Compare commits

..
28 Commits
Author SHA1 Message Date
jparkerweb 57134cf36e Merge pull request #7 from jparkerweb/sync/upstream-v2.4.5
v2.4.2: upstream web console updates (Cleanup, update banner, launcher banner, Notes threads)
2026-10-05 19:11:56 -07:00
jparkerweb c19a072a3a Sync upstream web console updates into v2.4.2: on-demand Cleanup of unused ~/.plan2code files, an UPDATE AVAILABLE banner from the release tags, a colored launcher with a Planny banner, Notes threads that keep the person's notes and read older reply shapes, the repo-local plan2code-model-update maintainer skill, and review fixes (413 for oversize bodies, notes kept on skip, scoped reconnect banner, live Overview tab, busy state on 409, menu keyboard focus, case-insensitive mentions) 2026-10-05 19:10:45 -07:00
jparkerweb cd8a1be568 Change image alt text for banner in README
Updated image alt text in README.md.
2026-10-05 18:56:03 -07:00
jparkerweb f968aeded8 Show a small spinner left of the web console's Questions and Ask tab names while Plan2Code is on it: Questions spins from Send until the answers are picked up, Ask while a picked-up question has no answer yet, and neither turns once the agent goes quiet or offline 2026-10-04 20:14:53 -07:00
jparkerweb fb71a02756 Hoist the Phase Checklist and Parallel Execution Groups to directly under the Summary in Document's overview.md template, so the phases are found without scrolling
AI Assisted
2026-10-04 18:44:23 -07:00
jparkerweb c8e65f12a6 Count a build on the web console's session meter while it runs, and score Implement + Review's built-in review once instead of twice
AI Assisted
2026-10-04 18:44:23 -07:00
jparkerweb 739e56c5cf Make the web console waiting screen's tasks pointer clickable: while a build runs, Task and the tasks tab's own name (Phase N tasks) show bold in the highlight color and open that phase's tasks tab 2026-10-04 15:52:20 -07:00
jparkerweb ade74cd636 Merge pull request #6 from jparkerweb/develop
Let Send through when an agent asks a question mid-build while still …
2026-10-04 14:24:23 -07:00
jparkerweb 48830c1afb Let Send through when an agent asks a question mid-build while still marked working: once an open question has an answer or note staged the web console unlocks Send at once instead of waiting out quietMinutes, and building.md now says a question stops the build (set waiting in the same post and wait) 2026-10-04 14:12:52 -07:00
jparkerweb b65481cb24 Merge pull request #5 from jparkerweb/develop
update web console, dashboard, implement-review, global launcher
2026-10-04 12:20:55 -07:00
jparkerweb 7d54941b75 Port upstream skill scripts into v2.4.1: deterministic workflow steps run as Node scripts beside each SKILL.md (specs, pathfinder, feedback payload, commit message, review scope, agent files, handoff path) with tests, the prompt limit rises to 20,000 chars, upstream review fixes are folded in (linked agent files never overwrite AGENTS.md, review stops with no base, paste-safe commit commands, Pathfinder number high-water mark and dry runs), and web console workspace folders are remembered per project and spec, with Finalize forgetting an archived spec's list
AI Assisted
2026-10-04 12:17:05 -07:00
jparkerweb 5cbc6edce2 Add the session meter rescale to the v2.4.1 CHANGELOG 2026-10-04 07:40:29 -07:00
jparkerweb 6b93bcb4c5 Scale the web console session meter with the work done: built phases score one point per 3 completed tasks (reported as run.tasks), every 2 answered questions add a point to their skill run, each Pathfinder research question adds 1 via the new pathfinder-research event, and thresholds move to yellow at 5, red at 10, ring full at 15 2026-10-03 22:31:14 -07:00
jparkerweb d49f2d5792 Keep Pathfinder from ending a session while its research subagents are still running: wait for every one, show a running count in the web console, mark finished research with a completion line, and re-run interrupted research on resume
AI Assisted
2026-10-03 10:23:38 -07:00
jparkerweb bfb47f3a8e Refresh agent docs, README and quick reference for v2.4.x: document plan2code-bot, launcher model menu, test/dev env vars, full npm test pipeline, GitHub-issues pathfinder backend, and dedupe command lists in favor of the README
AI Assisted
2026-10-03 09:38:55 -07:00
jparkerweb cb86034de9 Sync upstream v2.4.0 into v2.4.1: welcome dialog, bug/idea button with issue templates, version in Help, gear preferences and ocean default accent, plus web console screenshots, new landing page and web-console-first docs
AI Assisted
2026-10-02 20:51:34 -07:00
jparkerweb 0d7c2e140b Sync upstream v2.4.0 (web console, dashboard skill, implement-review, global launcher) with 24 review fixes and a look-around wake-up for Planny
AI Assisted
2026-09-28 19:37:41 -07:00
jparkerweb eefa7082ed Add pathfinder BRIEF mode, resolved-date tracking, and decide-menu-checkpoint cadence (v2.3.0)
AI Assisted
2026-09-03 07:18:34 -07:00
jparkerweb 8acd7b9d9f Modify image tag in README for better display
Updated image tag to include 'object-fit' style.
2026-08-26 22:24:02 -07:00
jparkerweb f026d3eebc Change banner image source to direct URL
Updated image source in README to use a direct URL.
2026-08-26 22:18:20 -07:00
jparkerweb 1b3360dc99 Add clipboard copy functionality for installation command in index.html 2026-08-22 10:10:06 -07:00
jparkerweb 16a736a2a9 Change alt text for banner image in README
Updated image alt text in README.
2026-08-22 09:58:20 -07:00
jparkerwebandDevin 8663eba113 Update the protected repository sync workflow
Keep the encrypted maintainer instructions current without mixing the opaque payload into the installer migration commits.

Generated with [Devin](https://devin.ai)

Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-08-20 13:42:56 -07:00
jparkerwebandDevin f40ddf5f8e Document the v2.2.0 skills-based distribution
Align release notes, version metadata, installation guidance, and maintainer documentation with the canonical skills workflow.

Generated with [Devin](https://devin.ai)

Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-08-20 13:42:43 -07:00
jparkerwebandDevin 57e32ab64f Delegate workflow installation to the skills CLI
Replace per-tool distribution generation with a canonical committed skills build so installation and updates share one format.

Generated with [Devin](https://devin.ai)

Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2026-08-20 13:42:31 -07:00
jparkerweb 483c3055b2 v2.1.1 - Failure Log convention in the init prompts 2026-08-12 13:54:43 -07:00
jparkerweb ca1e137a54 update init instructions 2026-08-12 13:15:50 -07:00
jparkerweb a7506952c8 Add a GitHub Issues backend to Pathfinder
Chart Step 1 now asks where the map should live: local files under
gitignored specs/ (the default) or a pathfinder:map issue whose decision
questions are its sub-issues, driven by gh. The pick is recorded as the
first Ground rules bullet and never re-asked; Auto-Discovery resolves
either backend, and an issue URL routes straight to github.

On github the local model maps onto the tracker's own primitives rather
than being simulated in issue bodies: sub-issues, native issue
dependencies for blocking, the assignee as the claim, one
pathfinder:<type>-<mode> label so type and mode cannot drift, and an
Answer comment plus a close reason that distinguishes a decision from a
question ruled out of scope. Both wiring calls key on the database id,
not the number. The map body therefore carries no checklist at all.

Sketches, secrets and the PLAN-DRAFT stay on local disk regardless --
plan2code-1-plan discovers its input with ls specs/ and has no notion of
a tracker. github is only offered after a five-check preflight, and the
offer names the repo's visibility, because a map on a public tracker
publishes the destination and the codebase recon. The Step 3 recon is
held until Step 6 so the no-fog off-ramp leaves no litter.

Depth lives in a seventh reference file; the orchestrator was
recompressed to absorb the new section within the 11,000 char limit by
removing duplication with its references. Adapted from the GitHub
tracker doc behind Matt Pocock's wayfinder skill (MIT).

AI Assisted
2026-08-08 19:00:26 -07:00
668 changed files with 406838 additions and 2416 deletions

No files matched your search

+178 -21
View File
@@ -5,9 +5,11 @@
```
plan2code/
├── src/ # Source workflow prompts (11 markdown files)
├── src/ # Source workflow prompts (13 markdown files)
│ ├── plan2code-0-pathfinder-references/ # Reference files for pathfinder skill
│ │ ├── brief.md # Dated plain-English decision briefs
│ │ ├── chart.md # MODE A: destination + frontier grills, templates
│ │ ├── github-issues.md # Issue-backed map (loaded only when Backend is github)
│ │ ├── grilling.md # Folded-in grilling + domain-modeling
│ │ ├── questions.md # On-disk question-file format + markers
│ │ ├── resolve.md # Per-type resolution + graduating the fog
@@ -25,6 +27,7 @@ plan2code/
├── plan2code-loop/ # Autonomous loop CLI tool (Node.js/TypeScript)
│ ├── src/ # TypeScript source
│ └── dist/ # Built output (tsup)
├── plan2code-bot/ # Autonomous end-to-end workflow test runner (TypeScript, tsup, Claude Agent SDK, vitest); user docs in .readme/test-bot.md
├── plan2code-metrics/ # Recursive self-improvement toolchain
│ ├── src/ # TypeScript source
│ │ └── prompts/ # Internal AI prompt templates (no char limit)
@@ -32,20 +35,42 @@ plan2code/
├── src/statusline-claude/ # Claude CLI status line (Node.js, zero deps, single file)
│ ├── statusline.js # Self-contained: config, git, formatters, render
│ └── statusline-config.json # Default config template
├── src/launcher/ # Global `plan2code` command (Node.js, zero deps)
│ ├── plan2code.js # Opens /plan2code in Claude Code or Devin in the cwd (or a picked folder)
│ ├── models.json # Curated model menu, copied to ~/.plan2code/bin/ on install
│ └── plan2code.ico # Desktop shortcut icon (Windows)
├── src/web-console/ # Local browser UI bundled into every skill (Node.js, zero deps, no build)
│ ├── console.mjs # CLI the agent drives: open, post, wait, chat, keep, status, stop, help
│ ├── server.mjs, lib.mjs, picker.mjs
│ ├── console.md, building.md # Agent-facing contract
│ └── public/ # Page assets served to the browser
├── scripts/ # Development scripts
│ ├── bench-web-console.mjs # Start-path timing
│ ├── capture-screenshots.mjs # Stages web console screens and saves the docs screenshots (needs puppeteer-core)
│ ├── check-skill-table.mjs # src/plan2code.md "What each skill reads and writes" table vs each skill's prompt and references
│ ├── parse-check.mjs # Imports every shipped browser module as real ESM
│ ├── screenshots/ # JSON payloads posted per shot (`ask` reuses `question.json`; `finish` follows `signoff.json`)
│ ├── test-web-console.mjs, test-launcher.mjs # node --test suites
│ ├── test-skill-scripts.mjs, test-script-<skill>.mjs, skill-script-helpers.mjs # Skill script suites and fixtures
│ └── validate-char-count.js # Pre-commit character count validator
├── dist/ # Generated distribution files (auto-generated)
│ ├── global-commands/ # For global installation (~/.claude/, etc.)
│ └── local-commands/ # For per-project installation (.claude/, etc.)
├── skills/ # Committed build artifact — one Agent Skill per src/ prompt
│ └── plan2code-<name>/ # SKILL.md plus references/ and scripts/ where present
│ # Generated by npm run build:skills; consumed by skills add
├── .husky/ # Git hooks (husky)
│ └── pre-commit # Runs character count validation
├── .claude/ # Repo-local Claude Code config (NOT installed by install.js)
│ └── skills/ # Maintainer-only dev skills, e.g. plan2code-publish/
├── docs/ # Documentation and assets
│ └── skills/ # Maintainer-only dev skills, e.g. plan2code-publish/, plan2code-model-update/
├── .github/ISSUE_TEMPLATE/ # bug.yml, idea.yml, config.yml (target of the console's feedback button)
├── .readme/ # User guides linked from README (web-console.md, web-console-design.md, walkthrough.md, autonomous-loop.md, metrics.md, status-line.md, test-bot.md)
├── docs/ # Landing page `index.html` served at plan2code.jparkerweb.com (GitHub Pages, `CNAME`) and assets
│ └── screenshots/ # Web console screenshots, <shot>-light.webp and <shot>-dark.webp (generated)
├── specs/ # Feature specs (if any in-progress)
├── install.js # Interactive installer (Node.js)
├── package.json # Root package (husky only, private: true)
├── version.json # Version metadata
├── CHANGELOG.md, QUICK-REFERENCE.md, LICENSE
├── .gitattributes, .editorconfig # Pin LF (see Code Style & Gotchas)
├── prototypes/, specs--completed/, dist/ # Local only, gitignored: design prototypes, archived specs, stale pre-2.2 build output. Ignore root dist/.
└── README.md # User documentation
```
@@ -53,25 +78,29 @@ plan2code/
| File | Purpose |
|------|---------|
| `install.js` | Main installer - generates and installs workflow files to AI tool directories |
| `install.js` | Interactive installer — builds `skills/` from `src/`, then delegates installation to the skills CLI |
| `src/plan2code-*.md` | Source workflow prompts (the "source of truth") |
| `scripts/validate-char-count.js` | Pre-commit validator ensuring all source prompts ≤ 11,000 chars |
| `skills/` | Generated Agent Skills committed for installation and drift verification |
| `scripts/validate-char-count.js` | Pre-commit validator ensuring all source prompts ≤ 20,000 chars |
| `version.json` | Version metadata (name, version, description) |
| `QUICK-REFERENCE.md` | User quick-reference card |
| `src/statusline-claude/` | Claude CLI status bar (included in `A` Install All + dev tools; also via Custom → S) |
| `src/statusline-claude/` | Claude CLI status bar (see Status Line below) |
| `src/launcher/plan2code.js` | Global `plan2code` command that opens the dashboard skill in Claude Code or Devin (installed by `I` and `A`; also via Custom → D) |
## Workflow Prompts (in `src/`)
| File | Step | Purpose |
|------|------|---------|
| `plan2code.md` | — | The dashboard: opens the web console's skill menu and launches the picked skill into its own session |
| `plan2code-init.md` | Init | Generate AGENTS.md as index + `.agents-docs/` section files (progressive discovery) |
| `plan2code-init-update.md` | Update | Update AGENTS.md with learnings; detects and routes edits to `.agents-docs/` files |
| `plan2code-0-pathfinder.md` | 0 | Chart a foggy idea as a local map of decision questions under `specs/<idea>/pathfinder/`, resolve one per session, hand a seeded PLAN-DRAFT to Step 1 |
| `plan2code-0-pathfinder.md` | 0 | Chart a foggy idea as a map (local files, or GitHub Issues) of decision questions under `specs/<idea>/pathfinder/`, resolve them one at a time, hand a seeded PLAN-DRAFT to Step 1 |
| `plan2code-quick-task.md` | quick | Lightweight planning for small tasks (standalone — not a pipeline step) |
| `plan2code-1-plan.md` | 1 | Requirements analysis & architecture |
| `plan2code-1b-revise-plan.md` | 1b | Mid-implementation revisions |
| `plan2code-2-document.md` | 2 | Create implementation specs |
| `plan2code-3-implement.md` | 3 | Execute implementation (phase by phase) |
| `plan2code-3-implement-review.md` | 3 | Execute one phase, run focused review and fixes, then request approval |
| `plan2code-review.md` | review | Post-implementation comprehensive review |
| `plan2code-4-finalize.md` | 4 | Validate, summarize, feedback, archive (steps 1–7, +optional 6.5) |
| `plan2code-handoff.md` | handoff | Compact the conversation into a self-contained handoff document for a fresh session |
@@ -79,6 +108,7 @@ plan2code/
## Naming Convention
Workflow files follow a strict naming pattern:
- **The hub:** `plan2code.md` (bare name — `bare: true` in `SOURCE_PROMPTS`, no suffix; it is the dashboard the rest hang off, not a step)
- **Utilities:** `plan2code-<name>.md` (single dash)
- **Numbered steps:** `plan2code-<N>-<name>.md` (single dash, number, single dash)
@@ -89,34 +119,130 @@ Examples:
## Reference Files
Some workflows use companion reference files for depth that exceeds the 11k char limit. The orchestrator (main workflow file) loads them via `Read` directives during execution.
Some workflows use companion reference files for depth that does not need to load on every invocation. The orchestrator (main workflow file) loads them via `Read` directives during execution.
**Pattern:** `src/<source-filename-without-extension>-references/` (e.g., `plan2code-review-references/`)
**How the installer handles them:**
- **Skill-directory platforms** (Claude Code, Agents, Crush, Devin): reference files are nested as `<skill-name>/references/`. Read paths use canonical `references/<file>.md`.
- **Flat-file platforms** (Windsurf, Cursor, Copilot, Continue): reference files are placed as a sibling directory. The installer rewrites Read paths to the sibling directory name (e.g., `plan2code-review-references/<file>.md`).
**How the installer handles them:** the directory is copied verbatim to `skills/<skill-name>/references/`, so `Read references/<file>.md` resolves consistently for every agent. There is one output format, with no flat-file sibling directory or path rewrite.
Reference files are NOT subject to the 11,000 character limit. The review workflow pioneered this pattern (`verification-protocol`, `dimensions`, `false-positives`, `session-end`); the init-update workflow also uses it (`ai-agent-file-sync` for its Step 7), `plan2code-4-finalize.md` uses it for STEP 6.5 (`community-feedback-submission`), and `plan2code-0-pathfinder.md` leans on it hardest (`chart`, `grilling`, `questions`, `resolve`, `handoff`, `trail` — the orchestrator is a dispatcher, the depth lives in the references). Other workflows can adopt it when a source file's detail exceeds the 11k limit.
Reference files are NOT subject to the 20,000 character limit. The review workflow pioneered this pattern (`verification-protocol`, `dimensions`, `false-positives`, `session-end`); the init-update workflow also uses it (`ai-agent-file-sync` for its Step 7), `plan2code-4-finalize.md` uses it for STEP 6.5 (`community-feedback-submission`), and `plan2code-0-pathfinder.md` leans on it hardest (`brief`, `chart`, `github-issues`, `grilling`, `questions`, `resolve`, `handoff`, `trail` — the orchestrator is a dispatcher, the depth lives in the references). Other workflows can adopt it when a source file carries detail that is only needed some of the time. The implement-review skill is a generated composition: `additionalReferences` in `SOURCE_PROMPTS` packages the canonical implement prompt, review prompt, and review reference directory into its own `references/` directory so those workflows are reused without source duplication.
**`Read` directives must sit at column 0.** `install.js` matches `/^(Read\s+)references\//gm` for the flat-file path rewrite — anchored, with no leading-whitespace tolerance. An indented or bulleted `Read` line is silently skipped, so flat-file platforms ship a `references/` path that does not exist there (they receive the reference dir as a *sibling*, named `plan2code-<name>-references/`).
## Skill Scripts
Anything a workflow step decides with one correct answer (parsing a spec's checkboxes, rebuilding Pathfinder's map markers, the commit message, which AI config files exist) lives in a script the agent runs, not in prose it follows. Prose keeps the judgment.
| Source | Ships to | How |
|---|---|---|
| `src/skill-scripts/<name>.mjs` | `skills/<skill>/scripts/<name>.mjs` | Shared: a skill lists the ones it runs in `scripts: [...]` on its `SOURCE_PROMPTS` entry. Skills install independently, so each gets its own copy |
| `src/skill-scripts/common.mjs` | every skill that has scripts | `SHARED_SCRIPTS_ALWAYS` in `install.js` |
| `src/<source>-scripts/<name>.mjs` | `skills/<skill>/scripts/<name>.mjs` | A skill's own scripts, mirroring `-references/`. A name that collides with a shared script is a build error |
| `version.json` (repo root) | `skills/<skill>/scripts/version.json` | Copied at build, so an installed skill can report its version |
| Script | Skills | Does |
|---|---|---|
| `specs.mjs` | plan, revise-plan, document, implement, implement-review, review, quick-task, finalize, handoff | `list` / `status` (Phase Checklist, groups, task markers, numbering, the implement selection verdict, consistency checks) / `metrics` / `mark` / `archive`. Imports `scanProject()` from the bundled `web-console/lib.mjs`, so the dashboard and the skills share one spec classifier |
| `commit-msg.mjs` | implement(-review), review, quick-task, finalize, init-update | The `<subject>` / ticket / `AI Assisted` command from the branch name |
| `review-scope.mjs` | review, implement(-review), quick-task | Review Mode's git fallback: files, lines, doc share, base branch |
| `agents-md.mjs` | init, init-update | AGENTS.md structure, budget, orphans, broken links, always-inline sections |
| `agent-files.mjs` | init, init-update | Detect and replace other agents' config files with AGENTS.md pointers |
| `pathfinder.mjs` (own) | 0-pathfinder | `reconcile`, `frontier`, `claim`, `resolve`, `rule-out`, `trail`, `gate`, `lint`, `brief-data` |
| `feedback-payload.mjs` (own) | 4-finalize | STEP 6.5's payload (real sha256 prompt hashes) and the gh / browser / print submission tiers |
| `handoff-path.mjs` (own) | handoff | Timestamped path, OS temp dir, git ignore check |
**Conventions** (`common.mjs`): Node 18+ built-ins only; one JSON object on stdout (`ok: false` plus `error`, `message`, `next` on a handled failure, the same line on stderr); exit 0 ok, 1 crash, 2 bad arguments, 3 not found, 4 invalid or check failed, 5 refused; `--help` documents commands and exit codes; writes take `--dry-run`; `PLAN2CODE_NOW` pins the clock for tests. A SKILL.md names a script as `node "<S>/<name>.mjs"`, where `<S>` is `scripts/` beside it.
**Tests:** `scripts/test-skill-scripts.mjs` (installer plumbing and the shared scripts), `scripts/test-script-<skill>.mjs` (a skill's own), with fixtures from `scripts/skill-script-helpers.mjs`. They run the built `skills/` tree, and one test fails if any SKILL.md or shipped reference names a `<S>/<name>.mjs` that skill does not ship. The per-skill scripts import `./common.mjs`, which only sits beside them in the built tree, so run them from `skills/`, not `src/`.
## Repo-Local Skills (`.claude/skills/`)
Maintainer-only Claude Code skills committed to the repo but **deliberately excluded** from `install.js` — they are dev tooling, not shipped product, so they never install to `~/.claude/skills/` and carry no version bump of their own (a product-version bump would wrongly imply a user-facing release); changelog mentions fold into the current version's entry.
Maintainer-only skills committed to the repo but deliberately excluded from the generated `skills/` product artifact:
- `plan2code-publish/` — cuts a GitHub Release from the top `CHANGELOG.md` entry once `CHANGELOG.md` / `version.json` / `package.json` agree and the version is ahead of the latest published release. Delegates tag creation to `gh release create --target main`.
- `plan2code-publish/` — publishes a GitHub Release after version files agree.
- `plan2code-changelog/` — validates release classification and keeps version files synchronized.
- `plan2code-model-update/` — refreshes the launcher's curated model menu (`src/launcher/models.json`) from `devin models list` and Claude Code's documented model aliases: shows a diff, applies only the changes the maintainer approves, validates the file and runs `npm test`. Terminal only, and it never commits.
- `sync-repo/` — decrypts the protected upstream-sync workflow in memory.
**Warning:** anything named `plan2code-*` placed under `~/.claude/skills/` is deleted by the installer's uninstall (`uninstallFiles()` in `install.js`) and by every re-install's pre-copy cleanup in `install()` (both match `/^plan2code-/` for the Claude Code skills target). Keep these skills repo-local only.
Do not run `skills add` against the repository root: recursive discovery also finds these maintainer skills. `install.js` targets `skills/` and passes explicit workflow names instead. Global install/uninstall cleanup removes `plan2code-*` workflow skills from user skill directories; it never operates on this repository's `.claude/skills/` directory, and `sync-repo` intentionally has no `plan2code-` prefix.
## Web Console
Optional local browser UI shipped with **every** skill. Source lives in `src/web-console/` and is fanned into each skill by an `additionalReferences: [{ source: 'web-console', target: 'web-console' }]` entry in `SOURCE_PROMPTS`, landing at `skills/<skill>/references/web-console/`. The bare `plan2code` skill is the hub: it opens the console as `--workflow dashboard`, the page draws the skill menu itself (from `SKILL_CATALOG` in `public/answers.js`, kept in sync with the install set by a test), and a card click reaches the agent as a `{ "i": "__launch", "skill": ..., "workflow": ... }` send — the dashboard then **resumes its own session** under the launched skill's workflow (`open --resume <sid> --workflow <w>` restates `workflow`/`title`/`specDir` on the live server) and runs that skill's file. `plan2code-review` now has a console of its own AND still runs inside builds on the page (a `review` verdict on Implementation's sign-off card, or the **Review it now** button on a finished quick task), so `plan2code-3-implement` and `plan2code-quick-task` also carry `review.md` and the review references, the same way `implement-review` does. The list of skills with a console is duplicated as `CONSOLE_WORKFLOWS` in `public/answers.js` (it decides whether a copied hand-off asks the next session for the console); a test fails if the two disagree.
| File | Role |
|---|---|
| `console.mjs` | the CLI the agent drives: `open`, `post`, `wait`, `chat` (the instant Quick question check), `keep` (copy a cited note attachment into `specs/<idea>/attachments/`), `status`, `stop`, `help` (a compact summary for an agent that found the binary but not `console.md`) |
| `server.mjs` | HTTP server, spawned **detached** by `open` (never as a harness background task, which get killed on compaction and at assorted timeouts) |
| `lib.mjs` | session paths, atomic writes, the patch merge, `validate()` (which also allows `templates` only as `"idea"` on a `text` item; anything else exits 3), `attachmentName()` (the kept-file name `<slug>-<id8>.<ext>`: the same upload always gets the same name), `repoRoots()` (returns `{ project, worktree, branch }`), `readOverview()`, and the three sweeps `open` runs fire-and-forget after it prints (see the cleanup design point below): `sweepUploads()`, `staleSessions()` / `removeSessions()` and `staleScratch()` / `removeScratch()`; plus `cleanupFindings()` / `removeCleanup()` for User Preferences → Cleanup |
| `public/` | `index.html`, `app.js`, `answers.js`, `labels.js`, `boot.js`, `chat.js`, `help.js`, `starters.js`, `workspace.js`, `meter.js`, `mentions.js`, `favicon.js`, `app.css`, `render.js`, `palette.js`, `favicon.svg` (Planny peeking up from the tab bar, served as `image/svg+xml`) with `favicon.png` (the same icon at 192 px, served as `image/png`) as the fallback, `vendor/marked.esm.js`, the chimes (`start-skill.mp3`, `bootup.mp3`, `start-stop.mp3`, `next.mp3`, `insert.mp3`, `sleeping.mp3`; which moment plays which is `SOUND_EVENTS` in `palette.js`, one switch per event saved in `looks.json`), and a `package.json` that only says `"type": "module"`, so Node (the server, `console.mjs`, `lib.mjs` and the tests import several `public/*.js` modules) parses them as ES modules without a reparse warning |
| `public/help.js` | the Help dialog's PURE rules: `HELP_TABS` (16 tabs: ids, order, labels, diagram flags; every tab uses the chosen highlight color), `helpTabFor()` (which tab opens, FR-8 order, first match wins: finished → Stopping and finishing, server gone or agent adrift → When something looks stuck, opened from the Workspace dialog → Your workspace, then Ask, a red-meter dashboard → Session meter, dashboard, a doc or the Overview, a build, else What this page is) and `helpKeyTarget()` (roving arrow keys, Home, End, wrapping). No module-scope DOM, so `node --test` covers it |
| `public/workspace.js` | PURE, no imports: the Workspace's name rules (`nameProblem()`, `NAME_MAX`, `defaultName()`, `freeName()` for the `-2` suggestion), the footer label and its width tiers (`footerLabel()`, `FOOTER_TIERS`) and `changeLine()`, the `Workspace: …` reply line an agent is handed per change. Imported by the server, `console.mjs`, `lib.mjs` and the page |
| `public/meter.js` | PURE, no imports: the session meter. `WEIGHTS` (the one place per-event points live; its keys are also `RUN_EVENTS`, the `run` events `post` accepts), `LAUNCH_COUNTED` (workflows whose launch itself scores), the built-phase sizing (`TASK_EVENTS`, the keys of `TASK_BONUS`: `implement-phase` +0 and `implement-review-phase` +1; `isTaskCount()` accepts a whole `tasks` from 1 to `MAX_TASKS` 999; `runPoints()` scores such a run `ceil(tasks / TASKS_PER_POINT)` (3) plus its bonus, and any other run its `WEIGHTS` value), `ANSWERS_PER_POINT` 2 (`answer` entries score per skill run, `floor(count / 2)`), `breakdownFrom()` (the click-the-ring list: one row per scoring entry plus one answers row per skill run), `pointsFrom()` (the count, its sum, folded from `ledger.ndjson` and deduplicated by id, never stored), `levelFor()` with `YELLOW_AT` 5 / `RED_AT` 10, `ringFraction()` (full at `RING_FULL` 15), `WORDS`, `tooltipFor()`, `RED_BANNER` and `meterView()` |
| `public/mentions.js` | PURE, no imports: the `@name` popover's logic — `mentionAt()` finds an `@partial` at the caret, `matchNames()` filters the workspace names, `applyMention()` splices a pick in and returns the new caret. Used by the Ask box, notes and text answers |
| `public/favicon.js` | The tab icon's status bar. `faviconState({ mood, spin, finished })` (PURE) maps the call `renderStatus()` makes for Planny's line to `busy` (the spinning ring), `ready` (`point`), `attention` (`adrift`, `gone`, or `work` without the ring — the stale line) or `plain` (finished, `happy`); `faviconSvg()` is `FAVICON_BODY` (kept equal to `favicon.svg` by a test) plus a bar along the bottom edge in the `FAVICON_COLORS` color (busy dashed on a white strip, so the states differ by pattern; the rest solid); `setFavicon()` swaps both `<link rel=icon>` hrefs only on a change — the SVG as a data URL, the PNG link drawn through a canvas. Called from `say()` in `renderStatus()`, which runs every second |
| `picker.mjs` | the per-OS native folder picker behind **Browse…** (`pickerCommand()` → `{ command, args, env }` or `null`; `pickFolder()` runs it with async `spawn`, so the server keeps serving while the dialog is open). Copied from `src/launcher/plan2code.js` (self-contained CommonJS, so it cannot be shared) with the prompt as a parameter; a drift test keeps the two in step |
| `update-check.mjs` | the dashboard's release check, server-only (never in `SERVABLE`): `compareVersions()` (numeric X.Y.Z, `null` otherwise), `latestTag()` (the highest `refs/tags/vX.Y.Z`, peeled `^{}` and pre-release tags skipped), `checkForUpdate({ installed, cacheFile, now, run })` (never throws) and `runLsRemote()` (`git ls-remote --tags` on the repo: no shell, `GIT_TERMINAL_PROMPT=0`, `GCM_INTERACTIVE=never`, SSH `BatchMode` when `GIT_SSH_COMMAND` is unset, `windowsHide`, killed at 10 s; git authenticates itself, the console never touches a credential). The answer, failures included, is cached for an hour in `~/.plan2code/console/update-check.json` (`UPDATE_CHECK_FILE`, in `PROTECTED_CONSOLE_FILES` so no sweep removes it). `server.mjs` starts it after the ready line and never awaits it; `stateFrame()` carries `update: { installed, latest }` only when a newer release exists (its `latest` is part of `haveKey()`, so a page that connected first still gets it), and `GET /version` returns `{ version, latest }`. Tests seed a fresh cached failure in every console home, so no test server reaches the network |
| `public/chat.js` | the Quick question chat's PURE rules, shared by the server and the page: `CHAT_LIMIT` / `CHAT_MAX_CHARS` / `CHAT_MAX_ATTACHMENTS` (the combined image + document cap; `CHAT_MAX_IMAGES` is kept as an alias of it), `chatOffline()` (the server's `409 offline` guard), and the page's view model (`chatView`, `sendState`, `nextAbout`, `contextOptions`, `unreadDot` / `shouldChime`). Ask context is the `{ history, about }` model: `nextAbout()` records the last 3 places visited in `history` and sets `about` only when the person picks one from **+ Add context** (`contextOptions()` lists that history plus *This spec*); dropping it or a successful send clears it, so the page never attaches context on its own. Imports only `answers.js`, no module-scope DOM, so `node --test` covers it |
| `public/starters.js` | PURE, no imports: the Role list (`ROLES`, `ROLE_NOT_SET`, `normalizeRole()`, `needsRoleNudge()`: the dashboard banner shows until `looks.json` carries any `role`), the 13 starter templates and their per-role orders (`TEMPLATES`, `TEMPLATE_ORDERS`, `templatesFor()`, `templateSwap()` for the replace prompt), and the Ask starter sets (`STARTER_SETS`, `STARTER_SET_FOR`, `startersFor()`). Imported by `app.js` and the tests; in `SERVABLE`, preloaded by `index.html` and parse-checked. The page owns the wording; a skill only flags its fresh-idea `text` card with `"templates": "idea"` |
| `public/answers.js` | the page's PURE answer-shaping logic (what an answer reads as, what a placeholder offers, whether a send carries an answer at all). Split out of `app.js` because `app.js` cannot be imported outside a browser, so anything left in it is untestable. `scripts/test-web-console.mjs` imports this directly. |
| `public/labels.js` | `taskLabel()` / `activityLabel()`: rewrites a build's `Task 4.2 of 6` activity line to `Task 2 of 6` (only for `implement` / `implement-review`; a plan's `Task 2.1` is a real section number). A separate module for the same reason as `answers.js`: pure, so `node --test` can import it |
| `public/boot.js` | `playWake()`: the dashboard's first-open wake-up. Self-contained (it owns its timers, the listeners that swallow input mid-run and the one measured glide transform; the page's `stir()` callback times the nudge between the click and the boot, which `app.js` fills with one snore; `skipEl` — the muted Skip button, and Escape — ends the run early with `onSkip` as the cue to silence queued sounds) and free of module-scope DOM access, so it can be imported by the tests and kept out of `app.js`'s render paths; `GLIDE_AT_MS` / `DONE_AT_MS` are its timings |
| `console.md` | the agent-facing contract, which is what every console prompt points at. Its **Workflow notes** section holds the per-workflow detail (Dashboard, Init and Init Update, Document, Review, Handoff, Pathfinder, Planning, Revision, Finalize), which keeps the prompts themselves to a one-line pointer |
| `building.md` | the build half of the contract, for `implement`, `implement-review` and `quick-task`: progress posts (`agent.activity`, `agent.quietMinutes`), the sign-off (with the review offer as a verdict), the finish, and running the review on the page mid-session |
Key design points:
- **First-run Welcome.** The first dashboard a person sees (after the wake-up finishes) opens `#welcome-modal`. Any close (button, Esc, backdrop) saves `looks.welcomeSeen: true` to `looks.json`, which survives "Back to default". Help's first tab has a **Show the welcome again** link (`data-action="welcome"`). Logic: `maybeWelcome()` / `openWelcome()` in `app.js`.
- **Top-bar icons.** `#btn-looks` (gear) opens `#looks-modal`, titled User Preferences: Role, Theme, Card width, Highlight color, Sounds and Models. `#btn-feedback` (megaphone) is a plain link to `https://github.com/jparkerweb/plan2code/issues/new/choose`, backed by `.github/ISSUE_TEMPLATE/bug.yml` and `idea.yml` (`config.yml` sets `blank_issues_enabled: false`). `DEFAULT_ACCENT` in `palette.js` is `ocean`.
- **Version in Help.** `GET /version` reads `version.json` beside `server.mjs`, falling back to `../../version.json` for a run from `src/` (`server.mjs` `/version` route). `renderHelpVersion()` in `app.js` fills `#help-version`. `computeExpectedSkills()` in `install.js` injects the root `version.json` into every `references/web-console/`, so a version bump changes every skill.
- **Models in User Preferences.** `GET` / `POST /models` in `server.mjs` read the shipped `~/.plan2code/bin/models.json` and read/write the user's `~/.plan2code/models.json` (`SHIPPED_MODELS_FILE` / `USER_MODELS_FILE` in `lib.mjs`). `PLAN2CODE_MODELS_DIR` relocates both for tests. The launcher reads the same two files for its model menu.
- **The wait is bounded.** `wait --seconds 240` exits `0` (answers), `10` (still waiting, `message: "Waiting on your answer in the web console"` to relay), `20` (server gone, resume it) or `30` (cancelled). The agent loops. Nothing blocks for the whole time a person takes.
- **The CLI hands the agent its terminal line.** `open`, `post` and `wait` (exits `0` and `10`) print `terminalLine: "→ Look at the web console: <url>"` (`terminalLine()` in `console.mjs`), and `console.md` asks the agent to end every terminal message with the latest one verbatim, so the tool builds the line rather than the model. In `wait` it is added to the printed result only; the collected `result.json` on disk never carries it. `open` also prints `swept: { sessions, scratch }` when it found anything to clean.
- **State fields written by `open`, not the agent.** Every `open` (fresh or resume) records `state.worktree` and `state.branch` from `repoRoots()` — the branch is read off `<gitdir>/HEAD` on the walk that already finds the gitdir (`""` on a detached HEAD or outside git; env-relocated repos ask git, as before) — and the page shows them in the footer as the workspace label (`footerLabel()`: `folder · branch`, or `folder + N folders · branch` once folders are added, stepping down through narrower tiers); a click opens the Workspace dialog. `state.wake = true` is set only by a fresh `open --workflow dashboard` and deleted by every resume (server restart, launch, Back to the dashboard); the page also keeps a per-sid `sessionStorage` guard so a reload never replays it.
- **`looks.width` is a person-level preference** alongside theme, accent and sounds in `looks.json`. `CARD_WIDTHS` in `public/palette.js` (`narrow` 720, `comfortable` 880 — default — `wide` 1080) is read by the page (`cardWidth()` validates a saved value) and by `server.mjs`, which paints `data-width` on first load; `app.css` maps `data-width` onto the single `--card-width` variable that every column rule uses.
- **`GET /overview` is the one route that reads a project file.** It serves `<spec>/overview.md` as `text/markdown` with `cache-control: no-store` (so the Overview tab is always fresh), or `404 { "error": "no overview" }`. A dashboard session honours `?spec=` only when it equals a `dir` in `state.scan.specs`; every other workflow uses `state.specDir` and ignores the query. The root is `state.worktree` (fallback `state.project`), and `readOverview()` resolves and `realpath`s the file and refuses anything outside the root's realpath, so `../`, absolute paths and outward symlinks all 404.
- **Month-old sessions are deleted whole.** `open` lists `staleSessions()` (directories matching the session-id pattern whose `state.json` mtime is older than 30 days, never the sid being opened — no exceptions for paused sessions or unread sends) and `staleScratch()` (top-level regular `.json` / `.mjs` files in the console folder older than 30 days, never `looks.json` or `console-dir` — `PROTECTED_CONSOLE_FILES` — never a directory or symlink) before it prints, and removes them after; `removeSessions()` / `removeScratch()` re-check every name against the session-id pattern / the scratch-name rule before deleting — not the age, the skipped sid or the file type, which the `stale*` listing already decided, so never hand them names that did not come from it. `console.md` tells agents to write temporary payload files into the session's own folder so they go with it.
- **User Preferences → Cleanup deletes on demand, from a list the server makes.** `POST /cleanup` (origin-checked) with `{ scan: true }` returns `cleanupFindings()`: `{ items: [{ id, where, name, kind, bytes }], bytes }`, ids `session:<sid>` (the `staleSessions()` rule, and every file in the folder that old too, never the serving sid), `console:<name>` (anything at the top of the console home but `PROTECTED_CONSOLE_FILES`, `sessions` and `runtime`, directories too) and `root:<name>` (anything at the top of `~/.plan2code` but `KEPT_PLAN2CODE_ENTRIES`: `console`, `bin`, `launcher.json`, `models.json`). Both stray kinds must be untouched for `STRAY_GRACE_MS` (an hour, by the newest `lstat` mtime anywhere inside, since rewriting a file never moves its folder's mtime; unreadable reads as fresh): the same folders hold `workspaces.json.lock` and `writeJsonAtomic()`'s `<file>.<pid>.tmp`, which live for milliseconds. `root:` is only scanned when the console home is `<PLAN2CODE_MODELS_DIR or ~/.plan2code>/console`, so a test or dev home never reaches the real folder. Paths never go to the page. `{ delete: [id] }` runs `removeCleanup()`, which re-scans and removes only ids still on the fresh list (symlinks unlinked, never followed), returning `{ removed, bytes, failed }`. The page always shows the list before **Delete these**.
- **The console's path is stated, not searched for.** Each Interface line names `references/web-console/` beside the SKILL.md and the `~/.plan2code/console/console-dir` fallback; the prompts with room (all but pathfinder, plan, document and finalize, which sit near the char limit) also name the canonical global install path `~/.agents/skills/<skill>/references/web-console/` (the skills CLI's real store; every agent dir links into it), and every `console.mjs open` writes `HERE` to `~/.plan2code/console/console-dir` (only `open`, so a `help` or dev run from `src/` never repoints it) so a later session can read the path from a fixed place. Without these the agent hunts the skill install dirs on every session.
- **`open` needs no payload, and runs first.** `--file` and `--title` are optional (the title defaults to the folder name). Every skill's Interface line puts `node "<D>/console.mjs" open --workflow <w>` BEFORE reading console.md or the project, and a test (`every skill opens the console before it reads anything`) keeps it that way; a blank session renders a starting screen ("Starting Pathfinder", Planny in flight) until the first post. Every skill replaces the flight with its own animated pose (init — init-update borrows it — pathfinder, plan — revise-plan borrows it — document, implement — implement-review borrows it — finalize, quick-task, review, handoff; only the dashboard keeps the flight), cloned by `startStage()` in `app.js` from `<template id="start-poses">` in `index.html` and animated under `.pose` in `app.css`; `START_POSES`/`startPose()` in `answers.js` is the list, and a test keeps the template in sync with it and every catalog workflow mapped to a pose. The same pose stands above the waiting pane (`.waiting-pane`, agent working with an `activity` and nothing to ask). `posePane()` keeps the pane on screen across renders when the pose is unchanged — `renderMain()` holds any `:scope > [data-pose]` child back from the clear — so the SVG never leaves the DOM, its animations never restart on a state frame, and Planny carries unbroken from launching → starting → waiting. The dashboard's launch hand-off resumes first and reads the launched skill's file after.
- **The start path is timed, and `open` is built around it.** Measure with `node scripts/bench-web-console.mjs` (`--browser-delay 900` for a real Chrome launch, `--roundtrip` for Send→agent and post→page, `PLAN2CODE_CONSOLE_TRACE=1` on `open` for a step timeline). Four things make it fast, and each is load-bearing:
- **No git process.** `repoRoots()` in `lib.mjs` finds the common dir and worktree top by walking to `.git` (following `gitdir:` and `commondir` files) — each `git rev-parse` cost 150ms+ on Windows. It asks git only when `GIT_DIR`/`GIT_WORK_TREE`/`GIT_COMMON_DIR` relocate the repo, and uses the native realpath so an 8.3 short name keys the same project git would.
- **The socket handover.** `open` spawns the server first, then binds the loopback port itself (the first `listen()` blocks ~50ms on Windows, overlapped with the server's boot), launches the browser on that URL immediately (Chrome via `start` takes ~900ms to its first request, the longest step of all), and hands the listening socket to the server over IPC (`bindLoopback()`/`handOff()` in `lib.mjs`, `--handoff` in `server.mjs`). Requests queue in the socket until the server takes it; one `open` already accepted is passed across as a socket and acknowledged (`conn-held`) before the channel is closed. No channel, or no socket within 3s, and the server binds its own port; `open` then opens the real URL.
- **The first paint needs no fetch.** `server.mjs` renders `index.html` with the session's title, badge and saved theme (`data-theme`, `data-looks-accent`, matched by accent rules generated from `palette.js` and appended to `app.css`) and the whole first state as an inline `<script type="application/json" id="boot">` (every `<` escaped; data blocks are not executed, so the CSP is unchanged). `modulepreload` hints fetch every module in parallel, the page draws in the same task as the script runs, and `/events?have=<key>` skips the opening frame when the page already holds that exact state. The chimes load on first use, not at start-up.
- **Reads are stat-gated.** `fileStamp()` (mtime:size:ino) decides whether `state.json`, `result.json` and `looks.json` changed before anything reads or parses them — per request, per heartbeat and in the 1s poll; the parsed state is cached (`readState()` returns it; `/submit` clones before editing). `wait` stats `result.json` every 50ms instead of parsing it every 400ms. Static assets are held in memory for the server's life and revalidate by ETag.
- **The Workspace and the session meter live beside `state.json`, one writer each, and outlive every hop.** Three session files, none of which `switchWorkflow()` ever touches, so both survive dashboard → skill → dashboard (a new console session starts with a count of 0, and with the folders remembered for its folder and spec, below):
| File | Writer | Holds |
|---|---|---|
| `workspace.json` | `server.mjs` | the folders (`id`, `path`, `name`, `description`, `original`, `nested`), `version`, the `changes` log, `missing` ids, and `remote` (`remoteWarning()`, now in `lib.mjs`, computed by the server at boot) |
| `ledger.ndjson` | `console.mjs` | append-only, one JSON line per entry. Meter events: `launch` (every fresh `open` and every resume into another workflow), `run` (a post's `run`, id scoped by launch, with `tasks` when a built phase reports one) and `answer` (one per item the person answered in a collected send, written by `ledgerAnswers()`, id `<launch>:answer:<send stamp>:<item>`). Also `folder-issue` reports |
| `workspace-cursor.json` | `console.mjs` | the last workspace `version` handed to the agent; advanced after the print, never backwards |
The page changes the workspace only through five routes: `GET /workspace` (`?check=1` re-stats every folder first), `POST /workspace/add` (path resolved against `state.worktree`, `~` expanded, case-insensitive duplicate check on win32/darwin; `not-a-folder`, `duplicate`, `bad-name`, `name-taken` + `suggest`), `POST /workspace/edit`, `POST /workspace/remove` (the original folder refuses) and `POST /workspace/browse` (one picker at a time, 409 `busy`; 404 `no-picker` when `remote` is set or no picker exists; `PLAN2CODE_PICKER_ECHO` stands in for the dialog in tests). `stateFrame()` carries `workspace` (folders, `missing`, `remote`, the current launch's folder `issues`) and `meter` (`meterView()` over the ledger), and both files are on the 1s poll; it also carries an optional `update` once the boot-time release check (`update-check.mjs`) finds a newer release. The agent hears of the workspace through `open` output (`workspace: { folders, missing? }`, which also seeds the cursor) and then each change once, as `workspace: [...]` plus `Workspace: …` reply lines on `wait` exit 0 or `chat`; a change alone never ends a `wait` slice (`pendingWorkspace` on exit 10). `post` and `open --file` lift `run` and `folderIssue` out of the patch into the ledger, never into `state.json`; an unknown event or a malformed issue is exit 3. **Remembered workspaces.** `lib.mjs` keeps `~/.plan2code/console/workspaces.json` (`WORKSPACES_FILE`, under `PLAN2CODE_CONSOLE_HOME` in tests): `projects[<case-folded worktree>] = { path, project, specs: { "specs/<name>": … } }`, each entry `{ original: { name, description }, folders: [{ path, name, description, nested? }], at }`. `rememberedWorkspace()` answers spec, else project, else nothing; `initialWorkspace(worktree, entry)` builds from it, both in the server's first `workspaceNow()` and in `open`'s read-only `workspaceOnOpen()`, so they agree. `workspace.json` carries `scope` (the spec key or `""`). Every user change (`bumpWorkspace`, not `missing`/`found`) calls `rememberWorkspace()`. `setScope()` swaps the list, logging `removed`/`added`/`renamed`/`described` changes so the cursor stays truthful; `POST /workspace/scope { spec }` (the dashboard picker, on selection and before a launch) seeds an unsaved spec from the project's list, and `followSpecDir()` (state poll, boot) moves the scope to a posted `specDir`, saving the current list as an unsaved spec's. Two writers, each a read-modify-write of one key: the server and `console.mjs forget --spec` (Finalize Step 6). The contract is in `console.md` (Workspace; Posting an update → Session meter) and `building.md` (per-phase `run`), so the skill prompts stay unchanged.
- **The filesystem is the transport.** Sessions live in `~/.plan2code/console/sessions/<sid>/`, deliberately not under `specs/`, so the console cannot trip the metrics scraper or the loop's marker parser, and survives `git clean`. A result written after the agent dies is picked up by the next session.
- **Attachments on notes and quick questions live in `<sid>/uploads/`**, never in the project: images as `<uuid>.jpg`, documents as `<uuid>.<ext>` with `ext` from `DOC_TYPES` in `answers.js` (text, code, PDF). The browser re-encodes images to JPEG and sends raw bytes to `POST /upload` as `image/jpeg`; documents go as-is as `application/octet-stream`, the extension taken by `docExt()` from the uncut `x-p2c-name`. `readRaw(req, limit, cb)` in `server.mjs` caps images at `MAX_BODY` (4 MB) and documents at `MAX_DOC_BYTES` (10 MB), and drains rather than destroys so the client gets its 413. Images need the `FF D8 FF` magic; documents pass `checkDocument()` in `lib.mjs` (`%PDF-` for PDF, strict UTF-8 with no NUL for text), else 415. The server names the file with `crypto.randomUUID()`. `GET /uploads/<id>.<ext>` serves images as `image/jpeg` and documents under a fixed type (`application/pdf`, else `text/plain`) with `nosniff` and, on text, CSP `sandbox` (a sandboxed PDF cannot open in the viewer), so a browser can show but never run one; `recordSubmitted()` writes every send's note onto its item as `sentNote { at, text, images, files }` (paths and names only), which `noteAwaiting()` in `answers.js` turns into the Sent box's attachment links and the Notes pane's pending note until the agent's next thread message; `DELETE /uploads/<id>.<ext>` takes any allowed extension; both stay contained under `uploads/`. `isOwnUpload()` returns the kind (`"image"` / `"file"` / `false`), and `attachmentsOk()` makes `/submit` comments and `/chat` messages carry images only in `images[]` and documents only in `files[]`, this session's uploads only, at most 5 together (`MAX_ATTACHMENTS`). A comment's `reply` gains one ` Image: <path> (<name>)` or ` File: <path> (<name>)` line per attachment, in attach order; a chat entry's lines come from `chatReplyLines()`, its images first, then its files. `sweepUploads()` bounds disk growth, and the session sweep removes the whole session a month after that. A note attachment a spec file cites is copied out first by `console.mjs keep` (`--session --upload --name --spec [--from]`): it checks both paths (the upload must sit under the session's `uploads/`, and the spec folder under `state.project`, real paths included, so a symlinked folder cannot carry the copy outside), copies with `COPYFILE_EXCL` and reuses an existing copy, prints `{ ok, path, link, image, reused }` with a forward-slash link relative to `--from`, and logs a `keep` event. It is the agent's hand, not the server's: it runs in the agent's shell only when the agent decides to cite a file, so the server still never writes into a project.
- **Answers carry their terminal token.** Each item ships `fallbackText` and `token`, so a click resolves to the literal word the prompt already expects (`approved`, `a`, `skip`). No workflow needs a second code path, and declining the console degrades to prose with nothing lost.
- **A build tells the page how long silence is normal.** The page calls a working agent stuck after two minutes without a post, which is right for a conversation and wrong for a twenty-minute build task. `agent.quietMinutes` (1 to 60, counted only while `working`) stretches the stuck, adrift and Send-grace thresholds; `quietLimitMs()` in `answers.js` is the one rule.
- **A finished quick task's review button is a request through the wait loop**, like the brief: `{ "i": "__review" }` or `{ "i": "__done" }`. The agent keeps waiting after posting a finish that carries `review`, and takes the review up by posting `"finish": null`, which puts the session back to work. Implementation's review needs no such machinery: it is a verdict on the sign-off card, asked before approval, so the session never left work to take it up.
- **Back to the dashboard is the same kind of post-finish request.** A finish with `finish.dashboard: true` shows the button; a press arrives as `{ "i": "__dashboard", "type": "dashboard" }`, and the agent runs `open --resume <sid> --no-open --workflow dashboard` then posts `"finish": null` with a fresh `menu`. The agent waits ~10 minutes after the finish for it (alongside `__review` / `__done` on a quick task); after that the page swaps the button for the `/plan2code` command. `finish.command` is optional — without one the page shows the "All done" line — and `validate()` rejects a `<placeholder>` command and a non-boolean `dashboard`. The session-level actions are therefore `__brief`, `__stop`, `__launch`, `__review`, `__done` and `__dashboard`; a new post-finish button should copy this precedent.
- **`__dashboard` also arrives mid-workflow.** A left-pointing triangle (`#btn-home`, filled with `--accent`) sits left of Stop session on every non-dashboard page. Its state comes from the pure `homeButtonState()` in `answers.js`: hidden on the dashboard, on a finish and during a stop, and otherwise gated by the same `turnGate()` Stop now uses (the person's turn, unless the agent has gone adrift). The confirm (`#home-modal`) reuses `stopWarning()`, and the submit carries the staged answers plus `__dashboard`, with `HOME_REPLY`. With no finish on the page, the agent treats it as a stop that lands on the menu: it saves its place as for `__stop`, then resumes as the dashboard with `"finish": null` + `menu`, with no finish and no stop in between (console.md → Stop requests → "Back to the dashboard, mid-workflow"; building.md for builds). `local.homeward` drives Planny's "Saving your place and opening the dashboard…" line until the through-the-dashboard reset clears it.
- **The Quick question chat is a second, append-only channel beside the card slot.** The page's **Ask** tab (always last, every workflow, still there after a finish) posts to `POST /chat` (a message, or `{ reset: true }` for New conversation) and `POST /chat/decision` (Approve / Decline on an edit the agent proposed). The server is the only writer of `<sid>/chat.ndjson` (one JSON line per `message` / `decision` / `reset`, each with a `seq`; `appendChat()` in `lib.mjs`); `console.mjs` is the only writer of `<sid>/chat-cursor.json` (the last seq handed over, advanced only after printing) and of the agent's replies in `state.json` → `chat.replies` (merged by `id`, checked by `validateChat()`: real `re`, proposal paths outside the running `specs/<idea>/`). The server never writes chat into `state.json` — `post` read-modify-writes it under `withStateLock()` — and `stateFrame()` merges inbox and replies into the frame's `chat` block (`conversation`, `messages`, `typed`, `limit`, `offline`) in memory. `/chat` never checks `pendingResult()`, so neither channel can block the other. `wait` keeps its exit codes 0/10/20/30: exit 0 gains `chat` (or is `type: "chat"` with no actions), and the instant `console.mjs chat` answers 0/10/20 without touching `result.json`, which builds run after every task. Guards: 10 typed messages per conversation (`409 limit`), 4,000 characters (`400 too-long`), own uploads only, of the right kind (`400 image`, `400 file`), `409 offline` when `chatOffline()` says no agent is there, `409 stale` for an old conversation. A hop through the dashboard (checked on every state change and at server start, so a restart still sees it) appends a `reset` and starts a new conversation. A lone reset never ends a `wait` slice (`deliverableChat()`); it rides along with the next message.
- **`validate()` in `lib.mjs` enforces four workflow rules mechanically**: max three open required questions, no internal vocabulary in a title, one recommended option, and no loop markers or metrics-scraper bait.
- **The palette is written twice for dark, and that duplication is the mechanism.** `@media (prefers-color-scheme: dark) :root:not([data-theme="light"])` follows the system; `:root[data-theme="dark"]` is the explicit choice from the colors panel. Drop the `:not()` and a system-dark browser overrides someone who asked for light. A chosen accent is applied as an inline custom property on `:root` (CSSOM, which the CSP allows), and the default is applied by REMOVING that override so the stylesheet's own per-theme values win. The server's `data-looks-accent` stand-in only lasts until `applyLooks()` runs, which removes it first thing — left on, its rule would beat "Back to default".
- **Help is in the page, built on first open.** The **?** button (`#btn-about`) opens `<dialog id="help-modal">`, which replaced the old "What this page is" About dialog. Its 16 panels and their icon sheet are hand-written HTML and SVG in `<template id="help-content">` in `index.html` (after `#start-poses`), cloned into `#help-panels` the first time **?** is pressed, so first load builds nothing and Help still reads fully once the server is gone. `help.js` supplies the tab list and rules, `app.js` the wiring (tab buttons, roving focus, `renderAbout()` for the live connection line on tab 1, `lightDismiss()`, and a Ctrl/⌘+Enter guard scoped to this dialog). SVG is styled by class and CSS variables only — a `style=` attribute is dropped by the CSP, and a test fails on one — and every number the prose quotes sits in a `data-fact` span a test checks against the constant it comes from.
- **Markdown is rendered to DOM nodes, never to an HTML string** (`public/render.js`), so agent-authored text that quotes a hostile source cannot execute. Only `marked.lexer()` is used.
Tests: `npm test` runs, in order, `validate-char-count.js`, `install.js --verify-skills`, `parse-check.mjs`, `check-skill-table.mjs`, then `node --test scripts/test-web-console.mjs scripts/test-launcher.mjs scripts/test-skill-scripts.mjs scripts/test-script-pathfinder.mjs scripts/test-script-finalize.mjs scripts/test-script-handoff.mjs`. The parse check exists because `node --check` parses a `.js` file as a CommonJS script and has already let a browser-fatal string literal through; it imports each shipped module as a real ES module instead. They live in `scripts/`, not `src/web-console/`, because everything under the latter ships into every skill.
**Adding a file under `public/` needs three edits, not one:** the file itself, the `SERVABLE` allow-list in `server.mjs`, and (for a module the page imports) a `<link rel="modulepreload">` in `index.html`. The set is hardcoded rather than a directory lookup so there is nothing to traverse; a new module missing from it 404s and the page dies at import with no useful error (a test fetches every import and fails naming the file). A module missing its preload still works, but loads serially after `app.js` and slows the first paint.
## Status Line
Optional Claude Code status bar living in `src/statusline-claude/`. Three-line bar (icon + content per line) showing model, project, branch, uncommitted diff stats, session duration + cost, context window usage, and plan/quota usage.
Optional Claude Code status bar living in `src/statusline-claude/`. Three-line bar beside a Planny icon: two content lines with a dashed rule between, showing model (+ effort), project/worktree, branch, uncommitted diff stats, session duration + cost, context window usage, and plan/quota usage.
**Design constraints:**
- **Zero runtime dependencies** — `statusline.js` is self-contained (config loader, git helpers, formatters, render). Copied verbatim to `~/.claude/plan2code-statusline.js` on install; no bundler step.
- **Stdin-driven** — all data comes from Claude Code's stdin JSON (`model`, `workspace`, `context_window`, `rate_limits`, `cost`). No API calls, no auth, no background processes.
- **Silent failure** — outer `try/catch` around `main()` plus `process.exit(0)` on missing stdin guarantees the script never crashes the CLI. All git ops are timeout-bounded (1.5s) and non-git workspaces short-circuit via `fs.existsSync('.git')`.
- **Silent failure** — outer `try/catch` around `main()` plus `process.exit(0)` on missing stdin guarantees the script never crashes the CLI. All git ops are timeout-bounded (1.5s) and non-git workspaces short-circuit via `isGitRepo()` (walks up to 10 parent dirs looking for `.git`).
- **Atomic settings writes** — installer writes `~/.claude/settings.json` via temp file + rename so a crash never leaves the file truncated.
- **Custom-config respect** — installer detects non-plan2code `statusLine` entries, prompts before replacing, and backs up to `statusline-previous.json`. Uninstall only removes `settings.statusLine` if it points to the plan2code bundle.
@@ -132,3 +258,34 @@ src/statusline-claude/
**Adaptive plan-usage display:** the formatter auto-selects between `5h/7d` rate-limit percentages (Pro/Max/Teams — when `rate_limits` present in stdin) and `Nk in · Nk out` session-token counts (Bedrock/Vertex/PAYG — when `rate_limits` absent). Segment is hidden when neither shape is available.
**Installer integration** lives in `install.js` under the `STATUS LINE INSTALLATION` section (`installStatusLine`, `uninstallStatusLine`). Included in `A` (Install All + dev tools); also available individually via Custom → `S`.
## Global `plan2code` Command
Typing `plan2code` in any directory opens the dashboard skill against the current project, **in that directory**, in one of two CLIs (the `CLIS` table in the launcher):
| CLI | Runs |
| --- | --- |
| Claude Code | `claude /plan2code --permission-mode bypassPermissions [args]` |
| Devin | `devin --permission-mode bypass [args] -- /plan2code` |
On a terminal it first prints a Planny banner (`renderBanner()`). With both on `PATH` it asks `Plan2Code opens inside an AI coding agent. Pick which one to use:` as a numbered menu, the last pick pre-selected and marked `(last used)` (Enter keeps it; a number, `claude` or `devin` also work); with one installed it uses that without asking, printing `Opening with <CLI>…` (as does a `--cli` pick); with neither it exits 127 naming both. `--cli claude` / `--cli devin` (or `--cli=<name>`) skips the menu and is never forwarded; an unknown name exits 2, an uninstalled one 127. The pick is saved as `lastCli` in `~/.plan2code/launcher.json`, merged beside `lastFolder`. Before spawning, unless the forwarded args carry `--model` / `--model=`, it asks `Which model should <CLI> use?` from the curated `src/launcher/models.json` (copied to `~/.plan2code/bin/models.json`, overwritten by every install) merged with the user's additions in `~/.plan2code/models.json` (edited in the console's User Preferences → Models). Enter keeps the model the CLI last used; a pick appends `--model <id>`. Ctrl+C at any launcher prompt exits 130. With no TTY on stdin there is no menu (last pick, else the first installed) and no model question: it prints that the CLI will use the model last used in it (switch it in that CLI on its own, or pass `--model <name>`). Headings, numbers, defaults, retry lines and the `plan2code:` error prefixes are colored with `paint()` (the escape values of `install.js`'s `COLORS`, copied), stdout only when stdout is a TTY and stderr only when stderr is, and never when `NO_COLOR` is set; piped runs print plain text and no banner. Other arguments are forwarded to the CLI (`plan2code --model opus`), and its exit code is passed back. Tests: `scripts/test-launcher.mjs` (run by `npm test`) spawns the launcher against stand-in `claude`/`devin` commands and a temp home.
**Design constraints:**
- **Stable copy, not the package dir** — the installer normally runs from an npx cache directory npm may delete, so `installLauncher()` copies `src/launcher/plan2code.js` to `~/.plan2code/bin/plan2code.js` and the shims point there. (The loop/metrics/bot wrappers point into `__dirname` and don't have this protection.)
- **npm-style shims** — written into npm's global bin (`npm config get prefix`; `<prefix>/bin` off Windows): an sh shim everywhere, plus `.cmd` and `.ps1` on Windows, so cmd.exe, PowerShell and Git Bash all resolve it. On macOS/Linux, when the npm bin isn't writable (system Node), it falls back to `~/.local/bin`.
- **Marker-based ownership** — every shim contains `plan2code-launcher`; `uninstallLauncher()` removes only files carrying it, leaving any other `plan2code` bin alone.
- **Resolving the CLI** — the launcher walks `PATH` (+ `PATHEXT` on Windows) itself. An `.exe` is spawned directly; a `.cmd`/`.bat` wrapper goes through `cmd.exe /d /s /c` with quoted args, since Node refuses to spawn batch files without a shell. The installer warns only when neither `claude` nor `devin` is on `PATH`.
- **No bin-name overlap** — the installer's own bin in `package.json` is `plan2code-install` (→ `install.js`), so `plan2code` belongs to the launcher alone. npx runs a package's only bin whatever it's named, so the README's `npx … plan2code.git` command is unaffected; keep it the only bin, or npx will need `-c plan2code-install`.
- **`--pick-folder` is shortcut-only** — the launcher strips it before building the CLI's args. The shortcut runs the CLI menu, then the folder picker, then the model menu. With it, a native folder picker chooses the cwd instead: on Windows, the Explorer-style `IFileOpenDialog` (`FOS_PICKFOLDERS`) via `powershell -STA -EncodedCommand`, its C# interop (`WINDOWS_PICKER_SOURCE`) passed in an env var and compiled with `Add-Type` (~0.5s), with a TopMost owner form so it doesn't open behind the console — Windows PowerShell's `FolderBrowserDialog` is the old tree-only dialog, and is kept only as the fallback when `Add-Type` is blocked (Constrained Language Mode, AppLocker); `osascript choose folder` on macOS, and `zenity` then `kdialog` on Linux (only when `DISPLAY`/`WAYLAND_DISPLAY` is set). If there's no picker, or the picker fails, it falls back to a typed `Project folder [last]:` prompt. A picker that exits 0/1 with nothing printed means the person cancelled, so the launcher exits 0. The picker starts on the last pick, stored in `~/.plan2code/launcher.json`. From the shortcut, a non-zero exit other than Ctrl+C (130 / `0xC000013A`) waits for Enter so the error doesn't vanish with the window. Typed in a terminal, `plan2code` behaves exactly as before.
**Desktop shortcut** — after the shims are written, `installLauncher()` (now async) calls `offerDesktopShortcut()`. It asks `Also add a Plan2Code shortcut to your desktop? (y/N)`, skips the question when stdin isn't a TTY, and refreshes an existing marked shortcut without asking. `installDesktopShortcut()` writes it to the real desktop (`[Environment]::GetFolderPath('Desktop')` on Windows, which covers OneDrive redirection; `xdg-user-dir DESKTOP` on Linux; `~/Desktop` otherwise):
| OS | File | Runs |
| --- | --- | --- |
| Windows | `Plan2Code.lnk` (written by `WScript.Shell`) | `pwsh.exe` (else Windows PowerShell) `-NoLogo -ExecutionPolicy Bypass -File "<bin>\plan2code.ps1" --pick-folder`, icon `src/launcher/plan2code.ico` (the web console's Peek-a-boo favicon, 16–256 px) copied to `~/.plan2code/bin/plan2code.ico` |
| macOS | `Plan2Code.command` (0755) | `exec "<bin>/plan2code" --pick-folder` in Terminal |
| Linux | `plan2code.desktop` (0755, `Terminal=true`, `gio … metadata::trusted`) | `"<bin>/plan2code" --pick-folder`, `Icon=` the web console's `favicon.png` copied to `~/.plan2code/bin/plan2code.png` |
The macOS and Linux shortcuts carry the `plan2code-launcher` marker as a comment line. A `.lnk` has nowhere to hide one (its Description is the hover tooltip), so it's recognized by what it runs: the UTF-16 strings `plan2code.ps1` and `--pick-folder` stored in the file (`plan2code.cmd` also matches, so shortcuts from before the PowerShell switch get refreshed). `uninstallLauncher()` removes only a recognized shortcut, plus the icons (`plan2code.ico`, `plan2code.png`), `bin/models.json` and `launcher.json` (never the user's `~/.plan2code/models.json`). A shortcut failure prints a warning and never fails the command install.
**Installer integration** lives in `install.js` under the `PLAN2CODE COMMAND` section (`installLauncher`, `uninstallLauncher`, plus the shortcut helpers). `install()` awaits it after a successful skills install (so both `I` and `A` get it), `U` removes it, and Custom → `D` installs it alone.
+25 -9
View File
@@ -4,19 +4,35 @@
## Code Style
- **install.js:** CommonJS, Node.js built-ins only (no external deps), ANSI colors via `COLORS` constant, readline-based prompts
- **plan2code-loop & plan2code-metrics:** TypeScript + ESM, built with tsup (target ES2022, moduleResolution: bundler)
- External deps: `@inquirer/prompts`, `chalk`, `execa`, `ora`
- **plan2code-loop, plan2code-metrics & plan2code-bot:** TypeScript + ESM, built with tsup (target ES2022, moduleResolution: bundler)
- External deps (loop, metrics): `@inquirer/prompts`, `chalk`, `execa`, `fs-extra`, `ora`
- External deps (bot): `@anthropic-ai/claude-agent-sdk`, `chalk`, `fs-extra`, `ora`
- Interactive CLI via `@inquirer/prompts` (select, input, confirm)
- **File operations:** Synchronous fs in all packages
- **File operations:** Synchronous fs by default; the web console's fire-and-forget sweeps and `readOverview()` use `fs.promises`
- **src/web-console:** ESM, Node built-ins only, no build step. It is bundled into every skill, so it must stay dependency-free and runnable as a plain `node console.mjs`. Anything added under `src/web-console/` ships to users — put tests in `scripts/` instead.
## Gotchas / Pitfalls
- **Version sync:** When adding a new version to `CHANGELOG.md`, also update `version.json` and `package.json` (root) to match. Check `README.md` for any version badges or references that need updating. The installer displays the version from `version.json` in its header. All three files (`CHANGELOG.md`, `version.json`, `package.json`) must always show the same version number.
- **Version sync:** When adding a new version to `CHANGELOG.md`, also update `version.json` and `package.json` (root) to match. `README.md` and `docs/index.html` (footer `Version X.Y.Z`) also carry the version. `version.json` is copied into every skill's `references/web-console/` (the console's Help shows it), so after a bump run `npm run build:skills` or `npm test` (`--verify-skills`) fails. The installer displays the version from `version.json` in its header. All three files (`CHANGELOG.md`, `version.json`, `package.json`) must always show the same version number.
- **CHANGELOG ordering:** Entries in `CHANGELOG.md` must be in reverse-chronological order — newest version at the top, oldest at the bottom. New entries are always inserted immediately after the file header.
- **CHANGELOG house format is not Keep a Changelog:** version headings are `## vX.Y.Z` — with a `v` prefix and **no date**. The release date is recorded only in `version.json`'s `releaseDate`. Category headings carry emoji: `### ✨ Added`, `### 🔧 Changed`, `### 🐛 Fixed`, `### 💥 Breaking`, `### 🗑️ Removed`, `### 📚 Documentation`. Older entries contain one-off variants (`🎁 Added`, `📦 Updated`, `📝 Documentation`, `🏎️ Improved`, `🧪 Testing`) — do not introduce new ones. The `.claude/skills/plan2code-changelog/` skill automates version selection and formatting; use it rather than hand-rolling an entry.
- **PowerShell mangles the CHANGELOG emoji:** `Get-Content` / `Select-String` render the `###` heading emoji as `?` under the default Windows console encoding, so a heading audit done that way reports garbage. Read `CHANGELOG.md` with a file-read or grep tool instead.
- **`.claude/skills/` is tracked:** a new repo-local skill only needs `git add` (see Architecture → Repo-Local Skills). A correction that is a workflow belongs there as a skill, linked from AGENTS.md, not as a Failure log line.
- **Loop `.gitignore` setup:** `ensureGitignore()` runs at startup in `Controller.run()` as a pre-flight step, not just inside `createTaskCommit()`. This is critical for phase mode where the Node controller doesn't handle commits — without it, `git add -A` would stage spec files.
- **Workflow file character limit:** All `src/plan2code-*.md` files must be ≤ 11,000 characters. A husky pre-commit hook enforces this. The 11,000 limit leaves buffer for platform-specific YAML headers (106-142 chars) to stay under Windsurf's 12,000 char limit.
- **Metrics internal prompts have no char limit:** Files in `plan2code-metrics/src/prompts/` are NOT subject to the 11,000 char limit — only `src/plan2code-*.md` consumer-facing prompts are.
- **Workflow file character limit:** All `src/plan2code-*.md` files (and the dashboard prompt `src/plan2code.md`) must be ≤ 20,000 characters (about 5k tokens, the usual Agent Skills guidance for a SKILL.md body). A husky pre-commit hook enforces this, and `plan2code-metrics` (`improver.ts`, `applier.ts`) applies the same number to proposed prompt edits, so change them together. History: 11,000 originally (headroom under Windsurf's old 12,000-char limit for platform-specific files, gone since 2.1.0's single-format build), 11,500 in 2.4.0, 20,000 in 2.4.1. The limit is now a prompt-size discipline only: the whole SKILL.md loads on every invocation while references load on demand, so situational detail still belongs in a reference file, and anything with one correct answer belongs in a script (see Skill scripts in the architecture doc).
- **Keep LF line endings:** `.gitattributes` pins `eol=lf` because `validate-char-count.js` counts the characters on disk, and a CRLF checkout adds about one per line to every prompt. Measure with `node scripts/validate-char-count.js` before trusting a near-limit prompt.
- **The pre-commit hook runs only the char-count check:** `.husky/pre-commit` is `node scripts/validate-char-count.js` alone. `skills/` drift, parse errors and test failures are caught only by `npm test`, so run it before committing. Sub-package tests are separate; see [Development Commands](./AGENTS-development-commands.md).
- **`skills/` is a committed build artifact:** edit `src/`, run `npm run build:skills`, and commit the regenerated skills in the same change. `npm test` runs `install.js --verify-skills` and fails on missing, unexpected, or stale files. Never edit `skills/` by hand.
- **Metrics internal prompts have no char limit:** Files in `plan2code-metrics/src/prompts/` are NOT subject to the 20,000 char limit — only `src/plan2code-*.md` consumer-facing prompts are.
- **User Feedback table format:** The `## User Feedback` markdown table in `overview.md` has a strict format the collector regex depends on. Field names must be exactly `Rating`, `Reason`, `Went Well`, `Went Poorly`. Pipe characters in values must be escaped as `\|`.
- **PLAN-DRAFT confidence numbers are scraped by regex:** when a `specs/<feature>/PLAN-DRAFT-*.md` contains no `<!-- METRICS_JSON ... -->` comment, `collector.ts` falls back to prose scraping (`collector.ts:186-241`). The overall-confidence pattern requires a literal `%`, but the four *breakdown* patterns (`collector.ts:201-204`) do **not** — `/[Rr]equirements?[:\s|]+(\d{1,2})/` and its siblings match a bare dimension word followed by whitespace, a colon, or a pipe and then digits. So a PLAN-DRAFT written by anything other than `/plan2code-1-plan` Phase 7 must keep both the `%` sign **and** bare `Requirements` / `Feasibility` / `Integration` / `Risk` followed by a number off the page — including innocent table rows like `| Requirements | 11 |`. Otherwise the metrics pipeline records a planning-step confidence that no planning step produced. `/plan2code-0-pathfinder` works around this by hyphenating the labels (`Requirements-clarity 22/25`), which breaks the character class.
- **Reference file sizing guideline:** Files in `src/plan2code-*-references/` directories target ~100-200 lines each (soft guideline; evaluate splitting above 300). They are NOT subject to the 11,000 character limit. The pre-commit hook (`validate-char-count.js`) only checks `src/plan2code-*.md` flat files — subdirectory contents are automatically excluded.
- **The splitting guideline has a hard ceiling — reference files cannot always be split:** two constraints bound it. (1) Each new reference costs the orchestrator a column-0 `Read references/<file>.md` line plus its fallback blockquote (~150-200 chars), and orchestrators near the 11,000 limit have no room to spend. (2) **The path rewrite is not recursive** — `syncPrompts()` rewrites `Read references/…` paths on the *orchestrator's* content only, so a `Read references/…` directive placed *inside* a reference file is never rewritten for flat-file targets; it ships as a dangling instruction pointing at a path that does not exist there. When a reference legitimately exceeds 300 lines (e.g. `plan2code-0-pathfinder-references/chart.md`), that is an accepted trade-off, not an oversight.
- **PLAN-DRAFT confidence numbers are scraped by regex:** `collector.ts` scrapes the latest `specs/<feature>/PLAN-DRAFT-*.md` prose. Overall confidence falls back to prose only when `<!-- METRICS_JSON ... -->` lacks `confidence` (`collector.ts:186-196`): `Confidence: 95%`, or a table row `| Confidence | 95` with no `%`. The four *breakdown* patterns (`collector.ts:200-203`) run on every draft, METRICS_JSON or not: `/[Rr]equirements?[:\s|]+(\d{1,2})/` and its siblings match a bare dimension word followed by whitespace, a colon, or a pipe and then digits. So a PLAN-DRAFT written by anything other than `/plan2code-1-plan` Phase 7 must keep `Confidence` followed by a number (with `%`, or as a table row) **and** bare `Requirements` / `Feasibility` / `Integration` / `Risk` followed by a number off the page — including innocent table rows like `| Requirements | 11 |`. Otherwise the metrics pipeline records a planning-step confidence that no planning step produced. `/plan2code-0-pathfinder` works around this by hyphenating the labels (`Requirements-clarity 22/25`), which breaks the character class.
- **Web console result-handoff invariant:** `result.json` holds answers a person pressed Send on. **An uncollected result (no `consumedAt`) is never overwritten and never deleted.** `/submit` and `/cancel` return 409 while one is pending, `open` clears the file only when it is already consumed, and `wait` stamps `consumedAt` **after** printing the result, never before, and exits by setting `process.exitCode` rather than calling `process.exit()`. That ordering is deliberate: an agent harness that kills a long shell command must not be able to land between "marked collected" and "handed over", and `process.exit()` can truncate an unflushed pipe. A repeat delivery is recoverable, a lost one is not. Two separate data-loss bugs came from breaking this in different places, and both were silent: the person sees "Sent", the agent never receives it. Any new code path that touches `result.json` must preserve the invariant, and `scripts/test-web-console.mjs` guards it. Related: the server writes each submitted answer onto its item as `submitted`, and `validate()` in `lib.mjs` must keep excluding that field from the completion-marker and metrics-scraper scans, which are meant for agent-authored text only. A person who typed `TASK_COMPLETE` into an answer would otherwise make every later patch unpostable.
- **Web console oversized bodies:** `readBody()` in `server.mjs` calls `req.destroy()` past `MAX_BODY`, so an oversized JSON body gets no response at all: the client sees a connection reset, never a status code. A route that must answer 413 has to stop buffering and drain the rest of the body instead, as `readRaw()` (the `POST /upload` reader) does; memory still stays capped.
- **Web console commands are tied to the directory `open` ran in:** `console.mjs` finds a running server through a handle file under a hash of the project directory (`handleFile(project, sid)` → `runtimeDirFor(project)` → `<RUNTIME_DIR>/<cwdHash(project)>/<sid>.json` in `lib.mjs`, with `RUNTIME_DIR` = `os.tmpdir()/plan2code-console`, or `$PLAN2CODE_CONSOLE_HOME/runtime`), and the project is the working directory of `open` / `open --resume`. Run `status`, `chat`, `wait` and `stop` from that same directory. From anywhere else `status` says `live: false`, `chat` / `wait` report `server-gone` (exit 20) and `stop` stops nothing, while the server is in fact still running. It is easy to mistake for a crash when a scratch session is opened from the repo and then driven from its session folder; check `server.log` and `events.ndjson` (a real exit writes an `exit` event or a stack) before chasing one.
- **Windows temp paths in web console tests:** `os.tmpdir()` can be an 8.3 short path (`C:\Users\JUSTIN~1.PAR\...`), while `server.mjs` canonicalises its session dir with `fs.realpathSync.native` to the long form. Any test comparing a server-returned path (e.g. an upload's `path`) against the test's `session` must pass the test side through `fs.realpathSync.native()` first, and compare case-insensitively on Windows.
- **Web console assets:** `computeExpectedSkills()` copies references as `Buffer`s, so binary files are fine. Adding a `public/` file needs the three edits in Architecture → Web Console. The vendored `marked.esm.js` sha256 is in `.readme/web-console.md`.
- **The default highlight color lives in two places:** changing it means editing `DEFAULT_ACCENT` in `public/palette.js` and the light, system-dark and explicit-dark `--accent*` values in `public/app.css`. No test ties them together.
- **Reference file sizing guideline:** Files in `src/plan2code-*-references/` directories target ~100-200 lines each (soft guideline; evaluate splitting above 300). They are NOT subject to the 20,000 character limit. The pre-commit hook (`validate-char-count.js`) only checks `src/plan2code-*.md` flat files — subdirectory contents are automatically excluded.
- **The splitting guideline has a hard ceiling — reference files cannot always be split:** each new reference costs the orchestrator a `Read references/<file>.md` line plus its fallback blockquote (~150-200 chars), which adds up across a prompt's 20,000-char budget. When a reference legitimately exceeds 300 lines (e.g. `plan2code-0-pathfinder-references/chart.md`), that is an accepted trade-off, not an oversight. Nested folders under `-references/` work: `references/` is the real path on every agent.
- **Windows global bins for loop, metrics and bot are wrappers, not `npm link`:** on win32 the installer writes `<name>.cmd` and `<name>.ps1` into `npm config get prefix`, each running `node <package>/dist/bin/<name>.js` (`install.js:1302-1323` for the loop; the metrics and bot installers do the same). After a source change, `npm run build` in the package is enough; no reinstall.
- **plan2code-metrics only runs from a repo checkout:** `analyzer.ts` and `improver.ts` load `src/prompts/analyze.md` / `improve.md` at runtime through `new URL('../src/prompts/...', import.meta.url)`, resolved relative to `dist/` (`analyzer.ts:11`, `improver.ts:13`). A packed or copied `dist/` without `src/prompts/` fails.
+90 -52
View File
@@ -4,12 +4,33 @@
## Common Commands
```bash
# Install dev dependencies (sets up husky pre-commit hooks)
# Install dev dependencies (sets up husky pre-commit hooks only; each sub-package
# below needs its own npm install, Node 18+)
npm install
# Run the interactive installer (always interactive — any CLI args are silently ignored)
# Run the interactive installer
node install.js
# Regenerate skills/ from src/ (non-interactive)
npm run build:skills
# Full check (about 70s): char limits, skills/ drift, browser-module parse check,
# dashboard skill-table check, web console + launcher node:test suites
npm test
# One web console or launcher test
node --test --test-name-pattern "<regex>" scripts/test-web-console.mjs
# Web console start-up benchmark (puppeteer-core optional)
node scripts/bench-web-console.mjs [--runs 7] [--workflow dashboard]
# Refresh the web console screenshots in docs/screenshots/ (needs a local Chrome or Edge).
# Rerun after any change a person would see on the console page, and open every image
# before committing: right screen, both themes, no username or home path in frame.
npm i --no-save puppeteer-core # once; never added to package.json
node scripts/capture-screenshots.mjs # every screen
node scripts/capture-screenshots.mjs --only dashboard # one screen
# Plan2Code Loop
cd plan2code-loop && npm install # First time setup
cd plan2code-loop && npm run build # Build the CLI
@@ -17,17 +38,35 @@ cd plan2code-loop && npm run build # Build the CLI
# Plan2Code Metrics
cd plan2code-metrics && npm install # First time setup
cd plan2code-metrics && npm run build # Build the CLI
cd plan2code-metrics && npm test # vitest (aggregator, community, improver)
# Plan2Code Bot (end-to-end workflow runner, Claude Agent SDK)
cd plan2code-bot && npm install && npm run build
cd plan2code-bot && npm test # vitest
```
The root `npm test` does not run the metrics or bot vitest suites; run them in their package. `npm run dev` in loop, metrics and bot is `tsup --watch`. The web console and launcher suites isolate themselves with `PLAN2CODE_CONSOLE_HOME`, `PLAN2CODE_MODELS_DIR` and `PLAN2CODE_NO_BROWSER=1` (`scripts/test-web-console.mjs:191`), so they never open a browser or touch your real console home. Almost all of the ~70s is the web console suite.
## Test/Dev Env Vars
| Variable | Effect |
|----------|--------|
| `PLAN2CODE_CONSOLE_HOME` | Replaces `~/.plan2code/console`: sessions, `looks.json` and the runtime handle dir all move under it (`src/web-console/lib.mjs`). Always set it for scratch runs so `stop --all` cannot reach real sessions. |
| `PLAN2CODE_NO_BROWSER` | `open` does not launch a browser (`NO_BROWSER` also works). |
| `PLAN2CODE_BROWSER` | Browser command line to run with the link appended, instead of the system default. |
| `PLAN2CODE_MODELS_DIR` | Moves the launcher model menus (`bin/models.json`, `models.json`) away from `~/.plan2code`, for tests. |
| `PLAN2CODE_PICKER_ECHO` | Tests only: `/workspace/browse` returns this as the picked folder instead of opening a picker (`-` means cancelled). |
| `PLAN2CODE_CONSOLE_TRACE` | Set to `1` on `open` to print a start-up step timeline to stderr. |
## Installer Menu Options
**Main menu:**
| Option | Action |
|--------|--------|
| `I` | Install Plan2Code workflow prompts for all platforms, plus `plan2code-loop` CLI |
| `A` | Everything in `I` plus `plan2code-bot`, `plan2code-metrics` (dev tools), and Claude Code status line |
| `U` | Uninstall Plan2Code files: prompts + `plan2code-loop` + `plan2code-metrics` + `plan2code-bot` + Claude Code status line (confirmation required) |
| `I` | Install the Plan2Code skills globally via the `skills` CLI, plus the global `plan2code` command (opens the dashboard in Claude Code or Devin) — no dev tools |
| `A` | Everything in `I` plus every dev tool: `plan2code-loop`, `plan2code-bot`, `plan2code-metrics`, and the Claude Code status line |
| `U` | Uninstall Plan2Code: skills + `plan2code` command + `plan2code-loop` + `plan2code-metrics` + `plan2code-bot` + Claude Code status line (confirmation required) |
| `C` | Open CUSTOM sub-menu |
| `Q` | Quit |
@@ -35,67 +74,66 @@ cd plan2code-metrics && npm run build # Build the CLI
| Option | Action |
|--------|--------|
| `L` | Show local (per-project) install instructions |
| `L` | Install the skills into the current project instead of globally |
| `O` | Install plan2code-loop CLI only |
| `M` | Install plan2code-metrics CLI only |
| `S` | Install Claude Code status line only |
| `B` | Install plan2code-bot CLI only |
| `D` | Install the global `plan2code` command only |
| `Q` | Return to main menu |
## Non-Interactive Flags
`install.js` takes no arguments for normal use, but exposes two build hooks. Any argument other than these two (when neither is present) exits 1 with usage; extra arguments beside either flag are ignored.
| Flag | Action |
|------|--------|
| `--build-skills` | Regenerate `skills/` from `src/`, pruning stale skills and reference files |
| `--verify-skills` | Compare committed `skills/` with `src/`; exits 1 on drift and is run by `npm test` |
## How the Installer Works
1. **Reads source prompts** from `src/plan2code-*.md`
2. **Generates platform-specific files** with appropriate headers (YAML frontmatter for some platforms)
3. **Writes to `dist/`** subdirectories organized by destination type
4. **Copies to target directories** (global: `~/.claude/commands/`, etc.)
1. **Builds `skills/` from `src/`** — one Agent Skill per source prompt, including uncommitted source edits.
2. **Checks the skills CLI is reachable** with `npx --yes skills --version`.
3. **Sweeps pre-2.2 install paths** and removes installed `plan2code-*` skills so renamed or retired prompts cannot survive as orphans.
4. **Delegates installation** to `npx --yes skills add "<repo>/skills" -g -s <skill names> -y`.
5. **Installs the global `plan2code` command** (only when step 4 succeeded). Custom → `L` runs the same flow without `-g` and without the launcher.
## Platform-Specific File Formats
The skills CLI owns distribution from step 4 onward. It stores canonical skills under `~/.agents/skills/` and links them into agents that maintain their own skill directory. Plan2Code no longer maintains platform-specific output formats.
| Platform | Extension / File | Local Dir | Global Dir | Header |
|----------|-----------------|-----------|------------|--------|
| Claude Code | `.md` | — | — | None |
| Cursor | `.md` | — | — | None |
| Copilot CLI | `.md` | — | — | YAML frontmatter |
| Continue | `.prompt.md` | — | — | YAML frontmatter |
| Windsurf | `.md` | — | — | YAML frontmatter |
| VS Code Copilot | `.prompt.md` | — | — | YAML frontmatter |
| Codeium | `.md` | — | — | YAML frontmatter |
| Claude Code (Skills) | `SKILL.md` in subdir | `.claude/skills/<skill-name>/` | `~/.claude/skills/<skill-name>/` | YAML frontmatter + `disable-model-invocation: true` |
| Agent Skills (Amp · Devin · OpenCode · Zed) | `SKILL.md` in subdir | `.agents/skills/<skill-name>/` | `~/.agents/skills/<skill-name>/` | YAML frontmatter (no disable flag) |
| Crush | `SKILL.md` in subdir | — (global only) | `~/.config/crush/skills/<skill-name>/` (Unix) / `%LOCALAPPDATA%\crush\skills\<skill-name>\` (Windows) | YAML frontmatter |
| Pi (pi.dev) | `.md` | `.pi/prompts/` | `~/.pi/agent/prompts/` | YAML frontmatter (`description`) |
**Invocation choices:**
| Choice | Reason |
|--------|--------|
| Explicit space-separated `-s <names>` | Avoids shell expansion and prevents unrelated directories under `skills/` from being installed. |
| No `-a` / `--agent '*'` | Uses the CLI's supported default agent set instead of requesting incompatible scope/agent combinations. |
| Captured output | Suppresses the CLI's duplicated banners while preserving real failures; unsupported-scope noise is filtered. |
## Skill Format
| Item | Value |
|------|-------|
| Path | `skills/<skill-name>/SKILL.md` |
| Skill name | `generateSkillName(prompt)`, such as `plan2code-1-plan` or `plan2code-init` |
| Frontmatter | `name`, `description`, `disable-model-invocation: true` |
| Reference files | `skills/<skill-name>/references/<path>`, copied from `src/<prompt>-references/` and `additionalReferences` (any file type; every skill bundles `web-console/` plus a copy of `version.json`) |
| Scripts | `skills/<skill-name>/scripts/<file>.mjs`, plus `common.mjs` and a generated `version.json` (see Skill Scripts in the architecture doc) |
`disable-model-invocation: true` is unconditional because these workflows are user-initiated. Agents that do not recognize the field ignore it.
## Editing Workflow Prompts
When modifying workflow prompts in `src/`:
See the `skills/` build-artifact gotcha in [Code Style & Gotchas](./AGENTS-code-style.md).
1. Edit the source file in `src/`
2. Run `node install.js` to regenerate distribution files
3. Test the workflow in your AI tool of choice
4. The `dist/` folder is regenerated automatically — don't edit files there directly
1. Edit the source file under `src/`.
2. Run `npm run build:skills`.
3. Test the workflow in an AI tool.
4. Commit regenerated `skills/` beside the source change; `npm test` fails on drift.
5. Never edit `skills/` by hand because the next build overwrites it.
## Adding a New Workflow Prompt / Skill
Adding a new prompt to `src/` is more than dropping in a file — the installer is
driven by an explicit registry and several docs enumerate the command set. When
you add a `src/plan2code-<name>.md`, do **all** of the following so nothing drifts:
1. **Create the source file** `src/plan2code-<name>.md` — body content **only**,
no YAML frontmatter (the installer generates frontmatter per platform). Keep
it **under 11,000 characters** (`npm test` enforces this).
2. **Register it in the installer.** Add an entry to the `SOURCE_PROMPTS` array in
`install.js` (`source`, `stepNumber`, `name`, `displayName`, `description`, and
`isUtility: true` for non-numbered utilities). If `stepNumber` is a non-numeric
label (e.g. `'handoff'`), add a matching case to `generateStepLabel()` so the
generated description reads correctly.
3. **Update every doc that lists the command set** — keep these in sync, they are
the canonical inventories:
- `README.md` — command table ("When to Use")
- `QUICK-REFERENCE.md` — Commands table
- `.agents-docs/AGENTS-architecture.md` — "Workflow Prompts (in `src/`)" table
- `CHANGELOG.md` — add an entry under the current version
- `docs/index.html` — **only** if the new prompt belongs to the core pipeline
shown there; utilities (like `init`, `quick-task`, `handoff`) are deliberately
omitted from that curated marketing list.
4. **Regenerate and validate:** run `node install.js` (regenerates `dist/`) and
`npm test` (character-count validator now covers the new file).
1. Create `src/plan2code-<name>.md` with body content only and keep it under 20,000 characters.
2. Register it in `SOURCE_PROMPTS` in `install.js` (with the web-console `additionalReferences` entry every skill carries); add a `generateStepLabel()` case when needed. List any shared scripts it runs in `scripts: [...]`, and put its own scripts in `src/plan2code-<name>-scripts/`. Add it to the skill/workflow table and the "What each skill reads and writes" table in `src/plan2code.md` (the latter is checked by `scripts/check-skill-table.mjs`), and to the workflow lists in `src/web-console/public/answers.js`, `public/meter.js` and the `console.mjs` usage text.
3. Update command inventories in `README.md`, `QUICK-REFERENCE.md`, `.agents-docs/AGENTS-architecture.md`, and `CHANGELOG.md`. Update `docs/index.html` only for core pipeline steps.
4. Run `npm run build:skills` and `npm test`.
+15 -8
View File
@@ -13,15 +13,15 @@ The loop uses an **LLM-driven discovery** approach:
## Loop Commands
```bash
# Build the loop CLI
cd plan2code-loop && npm run build
Run `plan2code-loop` in a project (fully interactive, no flags). Build commands: see [Development Commands](./AGENTS-development-commands.md).
# Run the loop (after linking) - fully interactive
plan2code-loop
```
The CLI auto-detects specs in `./specs/`, prompts for selection if multiple found, and handles session continuation interactively. Only `specs/<feature>/overview.md` is detected; archived specs in `specs--completed/` are ignored. Session state is stored per-spec in `specs/<feature>/.plan2code-loop/`.
The CLI auto-detects specs in `./specs/`, prompts for selection if multiple found, and handles session continuation interactively. Session state is stored per-spec in `specs/<feature>/.plan2code-loop/`.
Exit codes: 0 all complete, 1 max iterations, 2 interrupted (SIGINT/SIGTERM saves state), 3 error. The state dir holds `config.json`, `scratchpad.md` (LLM notes), `iteration.log` (NDJSON) and `spec.hash` (a mismatch prompts "Start fresh?"). Defaults: 100 max iterations, 3 min timeout, 5 retries.
## Supported Agents
Claude Code, GitHub Copilot CLI, Devin CLI, defined in `plan2code-loop/src/agents/` (one file each, registered in `agents/index.ts`).
## Loop Modes
@@ -36,6 +36,10 @@ The LLM must output one of these formats:
- `TASK_BLOCKED: 1.1 - Reason` - Cannot complete task
- `PHASE_COMPLETE` - Current phase finished (phase mode only)
- `LOOP_COMPLETE` - All phases finished
- `PREREQ_COMPLETE: P1.1 - Description` - Prerequisite verified
- `PREREQ_ASSUMED: P2.1 - Description` - Prerequisite that cannot be verified, assumed met
Parsing is in `utils/completion.ts` and is lenient: `TASK_COMPLETE: 1.1: x` and `TASK_COMPLETE[1.1]: x` also match.
## Key Source Files
@@ -44,4 +48,7 @@ The LLM must output one of these formats:
| `plan2code-loop/src/controller.ts` | Main loop orchestrator |
| `plan2code-loop/src/prompt/templates.ts` | Prompt templates for both loop modes |
| `plan2code-loop/src/utils/git.ts` | `createTaskCommit()` — handles task-mode commits with footer |
| `plan2code-loop/src/cli.ts` | Interactive CLI entry point |
| `plan2code-loop/src/cli.ts` | Interactive session setup (spec, agent, JIRA ID, loop mode, max iterations, resume/fresh) |
| `plan2code-loop/src/utils/completion.ts` | Marker parsing (`checkForCompletion` task mode, `checkForAllCompletions` phase mode) |
| `plan2code-loop/src/state/manager.ts` | Per-spec state dir, spec hash, iteration log |
| `plan2code-loop/src/agents/` | Agent adapters (claude-code, copilot-cli, devin-cli) |
+10 -8
View File
@@ -17,10 +17,7 @@ Collect → Aggregate → Analyze → Improve → Apply
## Metrics Commands
```bash
cd plan2code-metrics && npm run build # Build the CLI
plan2code-metrics # Run (fully interactive, no flags)
```
Run `plan2code-metrics` (fully interactive, no flags). Build and test commands: see [Development Commands](./AGENTS-development-commands.md).
## Metrics CLI Menu
@@ -33,6 +30,7 @@ plan2code-metrics # Run (fully interactive, no flags)
| Propose | AI improvement proposals with validation |
| Apply | Interactive diff review + file patching |
| Fetch community submissions | List/parse/import open community-feedback GitHub issues from jparkerweb/plan2code, close on success |
| Delete metrics data | Delete selected runs, all runs + aggregated data, or everything including proposals |
Community submissions arrive as GitHub issues labeled `community-feedback` on `jparkerweb/plan2code`, created by the finalize prompt's post-Step-6 submission flow; the "Fetch community submissions" option requires an authenticated `gh` CLI to list/close them.
@@ -49,6 +47,7 @@ Community submissions arrive as GitHub issues labeled `community-feedback` on `j
| `applier.ts` | Interactive diff review + file patching |
| `cli.ts` | Menu-driven interactive CLI (100% prompts, no flags) |
| `invoke-llm.ts` | Unified LLM interface (Claude Code, GitHub Copilot CLI, or Devin CLI) |
| `*.test.ts` | vitest suites for aggregator, community, improver (`npm test` in `plan2code-metrics/`) |
## User Feedback
@@ -68,8 +67,8 @@ Feedback is collected during finalize (Step 5) or retroactively via the CLI. Pip
## Supported AI Agents
- **Claude Code** (recommended): `claude` CLI with `--inputFile` for prompt delivery
- **GitHub Copilot CLI**: `copilot` CLI with stdin prompt delivery
- **Claude Code** (recommended): `claude --print --dangerously-skip-permissions`, prompt written to a temp file and piped to stdin (execa `inputFile`)
- **GitHub Copilot CLI**: `copilot --allow-all-tools -s`, prompt on stdin (default model `claude-sonnet-4`)
- **Devin CLI**: `devin` CLI with `--print --prompt-file <file> --permission-mode dangerous`
## Metric Targets
@@ -77,11 +76,14 @@ Feedback is collected during finalize (Step 5) or retroactively via the CLI. Pip
| Metric | Target | Direction |
|--------|--------|-----------|
| `avg_confidence` | ≥ 90 | higher is better |
| `avg_clarification_rounds` | ≤ 2.0 | lower is better |
| `avg_verification_gaps_found` | ≤ 2.0 | lower is better |
| `avg_parallel_groups` | ≥ 0.5 | higher is better |
| `avg_verification_items_added` | ≤ 1.5 | lower is better |
| `avg_task_completion_rate` | ≥ 0.95 | higher is better |
| `avg_blocker_count` | ≤ 1.5 | lower is better |
| `avg_completion_marker_success_rate` | ≥ 0.95 | higher is better |
| `avg_verification_failures_found` | ≤ 1.0 | lower is better |
| `archival_success_rate` | ≥ 0.99 | higher is better |
| `avg_user_rating` | ≥ 7.0 | higher is better |
Data stored in `.plan2code-metrics/` (runs/, aggregated.json, proposals/).
On start the CLI asks for the plan2code repo path (default: saved `plan2codeRepoPath` in `~/.plan2code-metrics.json`, else auto-detected up to 5 parent dirs) and the project path (default: cwd). Data is stored in `<project>/.plan2code-metrics/` (runs/, aggregated.json, proposals/).
+149
View File
@@ -0,0 +1,149 @@
---
name: plan2code-changelog
description: "Validate and fix the CHANGELOG.md version number before opening a PR. Reads main branch to determine the current latest version, classifies changes on the current branch, and proposes the correct next semver. Use this skill when the user mentions changelog, version number, preparing a PR, release version, semver check, or says 'check the changelog', 'what version should this be', 'prepare for PR', or 'fix the version'. Also use proactively when you notice a CHANGELOG entry that may have an incorrect version number."
---
# Plan2Code Changelog Validator
Ensure the CHANGELOG.md entry for the current branch has the correct semver version before a PR is opened. This skill exists because parallel branches independently pick version numbers that collide or leap-frog when merged — this validates against main's actual state right before the PR.
## Workflow
### Step 1 — Gather state
Run these commands to understand the current situation:
```bash
# 1. Current latest version on main
git show main:CHANGELOG.md | head -20
# 2. Current branch name (for ticket ID extraction)
git branch --show-current
# 3. What this branch changed (commit subjects)
git log main...HEAD --oneline
# 4. Files changed on this branch
git diff main...HEAD --name-only
```
Extract from main's CHANGELOG:
- The **latest version number** (first `## vX.Y.Z` line)
Note: on Windows, PowerShell's console encoding mangles the emoji in the `###` headings to `?`. Read `CHANGELOG.md` with the file-read or grep tool rather than `Get-Content` / `Select-String` when you need to see them.
Extract from the branch:
- The **list of changed files** to classify the change type
- The **commit messages** for changelog entry content
### Step 2 — Classify the change
Determine the change type by examining what was modified on this branch:
| Signal | Classification | Version Bump | Heading |
|--------|---------------|--------------|---------|
| An install target removed, or an existing workflow's contract broken | Breaking change | **Major** (X.0.0) | `### 💥 Breaking` |
| New workflow prompt (`.md` file under `src/`) | New workflow | **Minor** (x.Y.0) | `### ✨ Added` |
| New capability added to an existing prompt, or a new `.claude/skills/` skill | New capability | **Patch** (x.y.Z) | `### ✨ Added` |
| Behavioral changes to existing prompt(s), installer, or docs | Behavior change | **Patch** (x.y.Z) | `### 🔧 Changed` |
| Bug fix to existing prompt(s) or tooling | Bug fix | **Patch** (x.y.Z) | `### 🐛 Fixed` |
| A prompt, target, or file deleted | Removal | **Patch** (x.y.Z) | `### 🗑️ Removed` |
| README / `.readme/` / docs-site only | Documentation | **Patch** (x.y.Z) | `### 📚 Documentation` |
| Mix of the above | Use the **highest** bump (major > minor > patch) | Combine headings |
Use only the headings in this table — the CHANGELOG has historical one-off variants (`🎁 Added`, `📦 Updated`, `📝 Documentation`, `🏎️ Improved`, `🧪 Testing`) that should not be introduced in new entries.
### Step 3 — Compute the correct version
Starting from main's latest version:
- **Major bump:** increment the first number, reset the rest (e.g., `1.16.1` → `2.0.0`)
- **Minor bump:** increment the middle number, reset patch to 0 (e.g., `2.0.0` → `2.1.0`)
- **Patch bump:** increment the last number (e.g., `2.1.0` → `2.1.1`)
### Step 4 — Check the current branch's CHANGELOG
Read the current `CHANGELOG.md` on the branch. Look for:
0. **You are on `main` with no diff** — there is no branch to validate. Instead, compare the top CHANGELOG version against `git log` since the commit that released it: if commits have landed on `main` without a CHANGELOG entry, treat those commits as the change set and continue from Step 2. Say so explicitly rather than reporting "nothing to do."
1. **No entry exists yet for this branch's work** — the branch hasn't added a version entry above main's latest. Proceed to Step 5 to draft one.
2. **An entry exists but the version is wrong** — the branch has a version entry, but it doesn't match the computed correct version (common when branches were rebased or other PRs merged first). Report the discrepancy:
```
Version check for branch: {branch-name}
Main is at: {main-version}
Branch claims: {branch-version}
Correct version: {computed-version} ({classification})
The version needs to be updated: {branch-version} → {computed-version}
```
Ask: "Update the version to {computed-version}? (yes / no)"
3. **An entry exists and the version is correct** — report success:
```
Version check for branch: {branch-name}
Main is at: {main-version}
Branch version: {branch-version} ({classification})
Version is correct. CHANGELOG is ready for PR.
```
Stop here unless the user asks for content changes.
### Step 5 — Draft or fix the CHANGELOG entry
**If no entry exists**, draft a new one based on the commits and changed files. Match this repo's house format exactly — a bare `## vX.Y.Z` heading with **no date**, emoji `###` headings from the Step 2 table, and a blank line between bullets:
```markdown
## {computed-version-with-v-prefix}
### ✨ Added
- **{prompt-or-area}** — {what changed, and why it matters to someone installing it}
### 🔧 Changed
- **{prompt-or-area}** — {what changed}
```
Conventions to follow, drawn from existing entries:
- Bold lead-in naming the prompt, file, or area, then an em dash (`—`), then the description.
- Reference prompts by their command (`/plan2code-1-plan`) or path (`src/plan2code-init.md`), not by informal name.
- One paragraph per bullet is fine — this CHANGELOG favours substantive entries over terse one-liners, and a bullet may carry extra indented paragraphs for detail.
- A release with a big theme may open with a one-line summary paragraph directly under the `## vX.Y.Z` heading, before the first `###`.
Insert the new section directly below the `All notable changes...` line and above the previous version's heading.
Present the draft and ask for approval before writing.
**If the version is wrong**, update only the version number — preserve the existing content unless the user asks for content changes too.
After any changes, show the final CHANGELOG entry for confirmation.
### Step 6 — Sync `package.json` and `version.json` versions
After writing or updating the CHANGELOG entry, update the `version` field in `package.json` at the repo root to match the computed version:
1. Read `package.json` and `version.json` then check the current `version` value.
2. If it already matches the computed version, skip — no change needed.
3. If it differs, update the `"version"` field in both files to the computed version (e.g., `"version": "2.1.1"`).
4. Set `releaseDate` in `version.json` to today's date in `YYYY-MM-DD`. This is the only place a date is recorded — the CHANGELOG headings carry no date.
5. Include `package.json` and `version.json` in the same commit as the CHANGELOG changes.
This keeps `package.json`, `version.json`, and `CHANGELOG.md` in lockstep so `npm pkg get version` always reflects the latest release.
## Rules
- Never create a version entry without checking main first — the whole point is to derive the version from main's current state
- Always present changes before writing — the user should see and approve the CHANGELOG entry
- Match the existing file's format — `## vX.Y.Z` with no date, emoji `###` headings. Do not introduce Keep a Changelog's `## [x.y.z] - date` style
- Write for someone reading the release notes, not for someone reading the diff — say what the change lets them do
- If multiple change types exist (Added + Changed), use multiple headings under the same version
- `version.json`'s `releaseDate` is today's date, i.e. when the release is being prepared, not when the work started
- If the branch has no meaningful changes vs main (e.g., only non-shipping files changed), say so and ask if a CHANGELOG entry is actually needed
@@ -0,0 +1,116 @@
---
name: plan2code-model-update
description: "Refresh the plan2code launcher's curated model menu in src/launcher/models.json from `devin models list` and Claude Code's documented model aliases, with the maintainer approving every change before anything is written. Use this skill when the user says 'update the models', 'refresh models.json', 'new Claude model', 'devin models changed', or otherwise asks to bring the launcher's model list up to date."
---
# Plan2Code Model Update
Bring the launcher's curated model menu (`src/launcher/models.json`) up to date with the models each CLI offers today. Discover what Devin and Claude Code currently offer, show the maintainer a diff against the shipped file, apply only the changes they approve, validate the file, and run `npm test`. This skill never commits.
**Repo-local by design.** This skill lives in the repo's `.claude/skills/` and is intentionally NOT wired into `install.js` — it is a maintainer dev tool, not part of the shipped product, so it is never installed to `~/.claude/skills/`. Do **not** copy it there: the uninstaller (`uninstallSkills()` in `install.js`) and every re-install (`install()`) both run `cleanLegacyPaths()`, which deletes every entry matching `/^plan2code-/` under `~/.claude/skills/` (the `LEGACY_PATHS` row for Claude Code skills), so a copy placed there would be silently removed.
**Terminal only.** This skill runs in the terminal, with no web console session. Ask every question here.
## Workflow
### Step 1 — Preflight
Run from the repo root (the folder holding `install.js` and `src/launcher/`). Run `git status --porcelain -- src/launcher/models.json`. If it prints anything, warn — do not stop:
> `src/launcher/models.json` already has uncommitted changes. The diff below compares against the file as it stands on disk, including those changes.
### Step 2 — Read the current file
Read `src/launcher/models.json`. It has three keys: `_note` (a string), `claude` and `devin` (each an array of `{ "id", "label" }`). Keep the `_note` text exactly as read — Step 7 writes it back verbatim and the Validation rules check it.
### Step 3 — Discover Devin's models
If `devin` is on PATH (`command -v devin` in bash, `Get-Command devin` in PowerShell), run:
```bash
devin models list
```
Show its raw output to the maintainer before any interpretation — the format is not a stable contract. The output is long (hundreds of ids across about twenty families), so save it to a scratch file outside the repo, say where, and show the family headings in full. From it, take the explicit ids at Low, Medium and High effort only, plus `adaptive`. Skip every id the Validation rules forbid (`xhigh`, `max`, `fast`, `priority` endings), and also skip `none`-effort ids, ids with no effort suffix (such as `swe-1-7`, which is a Max model), and `fusion-*` combinations.
The 30-entry cap means the file can never hold every family at three efforts. Propose additions by family — newest and most-used first — and say how many entries that leaves.
If `devin` is not on PATH or the command fails, say so with the error, and switch to **guided manual edits** for Devin: ask the maintainer which ids to add, remove or relabel, one at a time, and carry those into Step 5 as the proposed list.
### Step 4 — Discover Claude Code's models
Look up Claude Code's current model documentation — the model configuration page in the Claude Code docs. Locate it on each run (search the Claude Code docs for model configuration) rather than using a remembered URL; docs move. List the documented model aliases (for example `default`, `sonnet`, `opus`, `haiku`, `opusplan`) and what each one means, and name the page you used. Keep `default` as the file's first entry even though the docs call it a special value rather than an alias. Bracketed aliases such as `opus[1m]` are valid ids; propose them only when the maintainer wants them in the menu.
If the lookup fails or the page does not list aliases, say so, and switch to **guided manual edits** for Claude, exactly as in Step 3.
### Step 5 — Show the diff
For each CLI, compare the discovered list with the current one by `id` and show one table:
| Change | id | Current label | Proposed label | Source |
| --- | --- | --- | --- | --- |
| added | `new-id` | — | New label | `devin models list` |
| gone | `old-id` | Old label | — | not in `devin models list` |
| relabelled | `same-id` | Old label | New label | Claude Code docs: <page> |
- **added:** offered by the CLI, not in the file.
- **gone:** in the file, no longer offered.
- **relabelled:** same `id`, a better or changed label. Propose a label in the file's existing style.
Summarise unchanged entries as a count under the table ("19 Devin entries unchanged") instead of listing them. If approving every **added** Devin row would take the list past 30 entries, say so above the table and by how many. Call out any likely rename before approval (see the rename warning under Validation rules). If both tables are empty, say the file is current and stop.
### Step 6 — Approve each change
Ask about each row, one at a time: **yes**, **no**, or **edit the label** (they type the label to use). Write nothing until every row has an answer. A "no" leaves that entry exactly as it is today.
### Step 7 — Edit `models.json`
Apply only the approved rows:
- Keep `_note` verbatim.
- Keep the existing order. Remove a **gone** entry where it stands, change a **relabelled** label in place, and append each **added** entry at the end of its list.
- Write two-space-indented JSON with a trailing newline, matching the file's current formatting.
### Step 8 — Validate
Check the edited file against every rule under Validation rules. On a failure, show the rule and the entry that broke it, and go back to Step 6 for the affected rows. Do not run tests on a file that fails validation.
### Step 9 — Test
Run `npm test` from the repo root and report the result honestly: pass, or the failing suite and its output. Never describe a failing run as passing.
### Step 10 — Remind about the release
A launcher model change reaches people only with the next release. Remind the maintainer to add a `CHANGELOG.md` line and keep `CHANGELOG.md`, `version.json` and `package.json` on the same version — `/plan2code-changelog` does both.
### Step 11 — Suggest a commit
Suggest a one-line commit message that says what changed, and do not run it:
```bash
git add src/launcher/models.json
git commit -m "Refresh the launcher's model menu"
```
Give `add` and `commit` as two separate commands (Windows PowerShell 5.1 has no `&&`).
## Validation rules
Check every rule before `npm test` (Step 8). They mirror the shipped-file test in `scripts/test-launcher.mjs` ("the shipped models.json is curated Devin at low/medium/high only") plus the menu's own needs:
- The file is valid JSON.
- `_note` is present and unchanged from what Step 2 read.
- `claude` and `devin` are both non-empty arrays.
- `devin` has at most 30 entries.
- Every entry has a non-empty string `id` and a non-empty string `label`.
- Ids are unique within each list, and labels are unique within each list.
- No Devin id ends in `xhigh`, `max`, `fast` or `priority`.
**Warn on renames.** When an entry is **gone** and a near-identical id is **added** in the same list (a changed version number or suffix, say `glm-5-3-high` → `glm-5-4-high`), call the pair out before approval as a likely rename. People's own lists in `~/.plan2code/models.json` merge with the shipped one by id (`mergeModels()` in `src/launcher/plan2code.js`), so a renamed id can bring back a user's old copy of the entry beside the new one. Approve the pair knowingly, or keep the old id.
## Rules
- **Never commit and never push** — Step 11 suggests the commit; the maintainer runs it.
- **Touch only `src/launcher/models.json`**, plus `CHANGELOG.md` only if the maintainer asks for the release line.
- **Nothing is written before every row is answered** (Step 6).
- **Never install this skill** — never add it to `install.js`, and never copy it to `~/.claude/skills/` (the uninstaller deletes `plan2code-*` entries there).
+56
View File
@@ -0,0 +1,56 @@
#!/usr/bin/env node
// set-synced.mjs: rewrite the "Last synced" line inside sync-repo.enc, in memory only.
// Decrypts, replaces exactly one anchored line, re-encrypts, then decrypts again to verify.
// Plaintext never touches disk and never passes through a shell pipe (PowerShell pipes
// re-encode UTF-8 through the console code page, which is how the body got mojibake).
// Usage: SKILL_PASSWORD=... node set-synced.mjs <file.enc> "<new value>"
// e.g. node set-synced.mjs sync-repo.enc "v2.4.0 (main @ 1a2b3c4)"
import { readFileSync, writeFileSync } from 'node:fs';
import { scryptSync, randomBytes, createCipheriv, createDecipheriv } from 'node:crypto';
const MAGIC = Buffer.from('SENC', 'ascii');
const VERSION = 0x01;
const SCRYPT = { N: 1 << 17, r: 8, p: 1, maxmem: 256 * 1024 * 1024 };
const KEYLEN = 32, SALTLEN = 16, IVLEN = 12, TAGLEN = 16;
// Anchored to a whole line that starts with "**Last synced upstream:". Command text elsewhere in
// the body can mention the phrase without ever matching, because it never starts a line this way.
const LINE = /^\*\*Last synced upstream: .*\*\*$/gm;
const password = process.env.SKILL_PASSWORD;
const [encPath, value] = process.argv.slice(2);
if (!password) { console.error('ERROR: set SKILL_PASSWORD env var.'); process.exit(2); }
if (!encPath || !value) { console.error('Usage: node set-synced.mjs <file.enc> "<new value>"'); process.exit(2); }
if (value.includes('*') || value.includes('\n')) { console.error('ERROR: value may not contain * or a newline.'); process.exit(2); }
function decrypt(b64) {
const blob = Buffer.from(b64.trim(), 'base64');
if (!blob.subarray(0, MAGIC.length).equals(MAGIC) || blob[MAGIC.length] !== VERSION) throw new Error('bad header');
let off = MAGIC.length + 1;
const salt = blob.subarray(off, off += SALTLEN);
const iv = blob.subarray(off, off += IVLEN);
const tag = blob.subarray(off, off += TAGLEN);
const d = createDecipheriv('aes-256-gcm', scryptSync(password, salt, KEYLEN, SCRYPT), iv);
d.setAuthTag(tag);
return Buffer.concat([d.update(blob.subarray(off)), d.final()]).toString('utf8');
}
function encrypt(text) {
const salt = randomBytes(SALTLEN), iv = randomBytes(IVLEN);
const c = createCipheriv('aes-256-gcm', scryptSync(password, salt, KEYLEN, SCRYPT), iv);
const ct = Buffer.concat([c.update(Buffer.from(text, 'utf8')), c.final()]);
return Buffer.concat([MAGIC, Buffer.from([VERSION]), salt, iv, c.getAuthTag(), ct]).toString('base64') + '\n';
}
let body;
try { body = decrypt(readFileSync(encPath, 'utf8')); }
catch { console.error('Decryption failed: wrong password or corrupted data.'); process.exit(1); }
const hits = body.match(LINE) || [];
if (hits.length !== 1) { console.error(`ERROR: expected exactly one "Last synced" line, found ${hits.length}. Nothing written.`); process.exit(1); }
if (body.includes('�')) { console.error('ERROR: body contains U+FFFD (encoding damage). Nothing written.'); process.exit(1); }
const next = body.replace(LINE, `**Last synced upstream: ${value}**`);
const out = encrypt(next);
if (decrypt(out) !== next) { console.error('ERROR: round-trip verification failed. Nothing written.'); process.exit(1); }
writeFileSync(encPath, out);
console.error(`Updated: ${hits[0]} -> **Last synced upstream: ${value}**`);
+1 -1
View File
@@ -1 +1 @@
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 truncated
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 truncated
+6
View File
@@ -0,0 +1,6 @@
root = true
[*]
end_of_line = lf
charset = utf-8
insert_final_newline = true
+3 -2
View File
@@ -4,8 +4,8 @@
# Git stores LF, and every text file is checked out as LF on all platforms.
# This is not cosmetic: scripts/validate-char-count.js measures the characters
# actually on disk, so a CRLF checkout adds ~1 character per line. The workflow
# prompts in src/plan2code-*.md run close to their 11,000 character budget
# (several sit above 10,800), and a CRLF working tree pushes them over — turning
# prompts in src/plan2code-*.md are counted against a 20,000 character budget,
# and a CRLF working tree inflates that count, turning
# `npm test` into a check that passes or fails depending on how the repo was
# cloned. Pinning eol=lf makes the count reproducible everywhere.
# ─────────────────────────────────────────────────────────────────────────────
@@ -16,6 +16,7 @@
# ─────────────────────────────────────────────────────────────────────────────
*.png binary
*.jpg binary
*.mp3 binary
*.jpeg binary
*.gif binary
*.ico binary
+78
View File
@@ -0,0 +1,78 @@
name: "🐛 Bug"
description: Something in Plan2Code or the web console is broken or behaves unexpectedly.
title: "🐛 Bug: "
labels: ["bug"]
body:
- type: textarea
id: what-happened
attributes:
label: What happened?
description: Describe the problem. Screenshots are welcome.
validations:
required: true
- type: textarea
id: expected
attributes:
label: What did you expect to happen?
validations:
required: true
- type: textarea
id: steps
attributes:
label: Steps to reproduce
placeholder: |
1. Run /plan2code
2. Pick ...
3. See ...
validations:
required: true
- type: dropdown
id: skill
attributes:
label: Which skill or tool?
options:
- Dashboard (/plan2code)
- Init / Init Update
- Pathfinder
- Plan / Revise Plan
- Quick Task
- Document
- Implement / Implement + Review
- Review
- Finalize
- Handoff
- Web console (page itself)
- Loop (plan2code-loop)
- Metrics
- Status line
- Installer
- Other
validations:
required: true
- type: dropdown
id: interface
attributes:
label: Where were you working?
options:
- Web console
- Terminal
- Not sure
validations:
required: true
- type: input
id: version
attributes:
label: Plan2Code version
description: Found in version.json, the installer banner, or at the bottom of the web console's Help dialog.
placeholder: "2.4.0"
- type: input
id: environment
attributes:
label: OS, Node version and agent CLI
placeholder: "Windows 11, Node 22, Claude Code"
- type: textarea
id: logs
attributes:
label: Console output or logs
description: Anything from the terminal or the browser console that helps.
render: shell
+1
View File
@@ -0,0 +1 @@
blank_issues_enabled: false
+29
View File
@@ -0,0 +1,29 @@
name: "💡 Idea"
description: Suggest a feature or an improvement to Plan2Code.
title: "💡 Idea: "
labels: ["enhancement"]
body:
- type: textarea
id: problem
attributes:
label: What problem or goal is this for?
description: What are you trying to do that is hard or impossible today?
validations:
required: true
- type: textarea
id: idea
attributes:
label: Your idea
description: What would you like Plan2Code to do?
validations:
required: true
- type: textarea
id: who
attributes:
label: Who would this help?
placeholder: "Product leads writing specs, engineers running the loop, ..."
- type: textarea
id: alternatives
attributes:
label: Alternatives you considered
description: Workarounds you use now, or other ways it could work.
+5 -1
View File
@@ -14,4 +14,8 @@ nul
handoffs/
node_modules/
package-lock.json
SYNC.md
SYNC.md
# Throwaway design prototypes (ai-assist-prototype). The decision they
# recorded lives in the code and the CHANGELOG; the files stay local.
prototypes/
+1 -1
View File
@@ -8,7 +8,7 @@ specs--completed/
.gitignore
CLAUDE.md
# Exclude generated files (installer generates dist/ dynamically)
# Exclude stale generated output from pre-2.2 checkouts
dist/
# Exclude dependencies (installer will run npm install if needed)
+1
View File
@@ -13,6 +13,7 @@ produces are unchanged.
| Approach | Best for |
|----------|----------|
| `/plan2code-3-implement` | Interactive control, reviewing each phase, logic that needs your judgment |
| `/plan2code-3-implement-review` | The same interactive control, with a focused review and its fixes applied before you approve the phase |
| `plan2code-loop` | Straightforward implementations, batch work, overnight runs |
The loop reads the same `overview.md` and phase files. You can start with the loop and finish by
+6 -6
View File
@@ -15,18 +15,18 @@ background processes.** Optional, and unrelated to the workflow itself.
On Pro / Max / Teams accounts, where rate limits are available:
```
◦ ◦ Opus 5 / high │ plan2code │ feature/pathfinder-github-iss │ +12 -3
╭●╮ ┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄
├■┤ 3h 5m ($4.62) │ ▰▰▰▰▰▱▱▱▱▱▱▱ 42% (84K) │ 5h: 28% · 7d: 61%
╭───╮ Opus 5 / high │ plan2code │ feature/pathfinder-github-iss │ +12 -3
│ ★ │ ┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄
╰┬─┬╯ 3h 5m ($4.62) │ ▰▰▰▰▰▱▱▱▱▱▱▱ 42% (84K) │ 5h: 28% · 7d: 61%
```
On Enterprise / Bedrock / Vertex / pay-as-you-go, where they aren't, the last segment becomes session
token counts instead:
```
◦ ◦ Sonnet 5 │ plan2code │ main │ +12 -3
╭●╮ ┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄
├■┤ 2m ($0.18) │ ▰▰▱▱▱▱▱▱▱▱▱▱ 18% │ 88k in · 3k out
╭───╮ Sonnet 5 │ plan2code │ main │ +12 -3
│ ★ │ ┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄┄
╰┬─┬╯ 2m ($0.18) │ ▰▰▱▱▱▱▱▱▱▱▱▱ 18% │ 88k in · 3k out
```
The icons down the left are Planny, the project mascot.
+1 -1
View File
@@ -19,7 +19,7 @@ Built for **maintainers**. If you're using Plan2Code to ship features, you don't
## Install
```bash
node install.js # I (everything) — or C → B (bot only)
node install.js # A (everything + dev tools) — or C → B (bot only)
```
## Run
+124 -64
View File
@@ -1,103 +1,163 @@
# Walkthrough: one feature, start to finish
A REST API for task management, from a sentence to archived specs. Every block below is a **separate
conversation** — that's the whole trick.
A REST API for task management goes from a sentence to archived specs, all on the web console.
Each session stays short, and the specs on disk carry everything from one session to the next.
← [Back to README](../README.md)
---
## Session 1 — Plan
## Session 1: Plan
Run `plan2code` in the project (or `/plan2code` inside your agent). The dashboard opens in your
browser with every step laid out as a card.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="../docs/screenshots/dashboard-dark.webp">
<img src="../docs/screenshots/dashboard-light.webp" alt="The Plan2Code dashboard: a Set up card, then the steps in order as cards, one of them marked Suggested.">
</picture>
Pick **Plan** and describe the idea in the box: *I want to build a REST API for a task management
application.* The same page turns into the Plan session.
The questions arrive in batches of up to three, each option with its trade-off written underneath.
Answer what is there and press **Send to Plan2Code**; the next batch follows.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="../docs/screenshots/question-dark.webp">
<img src="../docs/screenshots/question-light.webp" alt="A Plan question card, How tasks are paged, with three options, the first marked Suggested, each with its trade-off underneath.">
</picture>
The phase breakdown comes back as a list you can reorder and edit. Documents a skill writes on the
page open as tabs like this one, where settled parts read as decisions and anything still a guess
is marked as assumed and still open.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="../docs/screenshots/doc-dark.webp">
<img src="../docs/screenshots/doc-light.webp" alt="The What we've agreed tab for the task API: a settled summary, an assumed section on paging with a dashed edge, and a dimmed draft list of what is still to work out.">
</picture>
Plan proposes a tech stack, Fastify + Postgres + Zod, and does not pick it alone: a sign-off card
asks you to approve it. Plan then designs the architecture. At 92% confidence (the gate is 90%) it
asks whether to write the draft, and with 3 assumptions written down, it writes:
```
You: /plan2code-1-plan
I want to build a REST API for a task management application.
AI: 🤔 [REQUIREMENTS ANALYSIS]
Six questions. Four of them I need answered before I can size this.
...
AI: 🤔 [TECH STACK]
Proposing Fastify + Postgres + Zod. I don't pick this alone — sign off?
You: approved
AI: 🤔 [ARCHITECTURE DESIGN]
...
AI: 🤔 [TRANSITION DECISION]
Confidence: 92% (gate is 90%). Assumptions written to the draft: 3.
→ specs/task-api/PLAN-DRAFT-20260804.md
specs/task-api/PLAN-DRAFT-20260804.md
```
Two files land: the draft, and a `PLAN-CONVERSATION-*.md` log of how you got there.
Two files land: the draft, and a `PLAN-CONVERSATION-*.md` log of how you got there. The finish
card names the next step.
> **Started foggy instead?** Run `/plan2code-0-pathfinder` first. When its map clears it writes a
> `PLAN-DRAFT` that Step 1 picks up at Phase 4, with requirements and scope already answered.
> `PLAN-DRAFT` that Plan picks up at Phase 4, with requirements and scope already answered.
---
## Session 2 — Document
## Session 2: Document
On Plan's finish card, press **Back to the dashboard** (or start a fresh console session if the
session meter says so; see Session 3 below). Pick **Document**, with `specs/task-api` selected in
the picker at the top. It reads `specs/task-api/PLAN-DRAFT-20260804.md`.
The overview and each phase file appear as tabs as they are written, so you watch the spec take
shape:
```
You: /plan2code-2-document
specs/task-api/PLAN-DRAFT-20260804.md
AI: 📝 [DOCUMENTATION]
Creating specs/task-api/overview.md
Creating specs/task-api/Phase 1.md Project setup (6 tasks)
Creating specs/task-api/Phase 2.md Data model (8 tasks)
Creating specs/task-api/Phase 3.md API endpoints (11 tasks)
Creating specs/task-api/Phase 4.md Authentication (7 tasks)
Parallel execution groups: Phase 3 and Phase 4 don't share files.
specs/task-api/overview.md
specs/task-api/phase-1.md Project setup (6 tasks)
specs/task-api/phase-2.md Data model (8 tasks)
specs/task-api/phase-3.md API endpoints (11 tasks)
specs/task-api/phase-4.md Authentication (7 tasks)
```
Parallel execution groups: Phase 3 and Phase 4 don't share files.
---
## Sessions 3…N — Implement, one phase each
## Sessions 3 to N: Implement, one phase each
```
You: /plan2code-3-implement
specs/task-api/overview.md
Look at the session meter in the top bar first. Plan and Document add 2 points each, so after
those two the meter is at 4: **Session: getting long**, in yellow. That is the cue to close this
console session and start a fresh one for the build (run `plan2code` again). This is the first of
the four rules:
AI: ⚡ [PHASE 1: Project setup]
(Phase 1 is the next unchecked phase.)
...
✓ 6/6 tasks. Checkboxes updated in Phase 1.md and overview.md.
Next: Phase 2 — Data model (8 tasks). New conversation.
```
> **1 · Keep each session short.**
> Chaining a couple of small skills on the dashboard is fine. Start a fresh console session when the session meter turns red, and always before a big planning or build step. The specs on disk carry everything between sessions.
Repeat. New conversation each time. It always finds the next open phase itself.
On the fresh dashboard, pick **Implement** for `specs/task-api`. Phase 1 (Project setup) is the next
unchecked phase, so it starts there. The page shows the build as it goes: a progress bar counting
tasks, the phase's task list ticking as each task lands on disk, and a line beside Planny saying
what it is doing now. Use the **Ask** tab for a quick question without interrupting the build.
When you reach a parallel group, it offers you the choice — open a second agent, take the other
<picture>
<source media="(prefers-color-scheme: dark)" srcset="../docs/screenshots/build-dark.webp">
<img src="../docs/screenshots/build-light.webp" alt="A phase of a task API being built: 5 of 11 done on the progress bar, and the phase's task list with the first five ticked.">
</picture>
When all 6 of 6 tasks are ticked in `phase-1.md`, a **Completion report** tab opens with the
sign-off card: **Review the code first**, **Approve this phase**, or **I want changes**.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="../docs/screenshots/signoff-dark.webp">
<img src="../docs/screenshots/signoff-light.webp" alt="The Approve this phase card, with a button to read the completion report and three choices: Review the code first, Approve this phase, and I want changes.">
</picture>
After you approve, `overview.md` marks Phase 1 done, and the finish card gives the next command:
Phase 2, Data model (8 tasks), in a new session.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="../docs/screenshots/finish-dark.webp">
<img src="../docs/screenshots/finish-light.webp" alt="The finished screen after a phase sign-off: the phase is built, the next implement command with a Copy button, and a Back to the dashboard button.">
</picture>
Repeat. A fresh session each time. It always finds the next open phase itself.
When you reach a parallel group, it offers you the choice: open a second agent, take the other
phase, and the `[/]` marks keep them from colliding.
**Optional, any time:** `/plan2code-review` for an independent second opinion on what just landed.
Pick **Implement + review** instead when review should run automatically before phase approval.
**Review** remains available on the dashboard for an independent second opinion at any other time.
---
## Final session — Finalize
## Final session: Finalize
```
You: /plan2code-4-finalize
specs/task-api/overview.md
With every phase approved, the dashboard suggests **Finalize** for `specs/task-api`. Pick it in a
fresh session. Each check lands on the page as a report tab, and anything that needs your say is
a card:
AI: 🧹 [VALIDATION]
All 32 tasks verified against the phase specs. 2 gaps found and fixed.
- **Validation.** All 32 tasks verified against the phase specs. 2 gaps found and fixed.
- **Documentation review.** README needs the new `/tasks` endpoints; AGENTS.md is current. A card
lists the proposed doc updates, and you tick the ones to make.
- **Spec cleanup.** A confirm card, then `specs/task-api/` moves to `specs--completed/task-api/`.
AI: 🧹 [DOCUMENTATION REVIEW]
README needs the new /tasks endpoints. AGENTS.md is current.
AI: 🧹 [SPEC CLEANUP]
Moved specs/task-api/ → specs--completed/task-api/
Implementation complete.
```
The finish card says the implementation is complete, and gives the commit command when doc
updates landed.
---
## If requirements move mid-build
Don't patch the code and hope the specs catch up. Run `/plan2code-1b-revise-plan` — it edits the
specs (and only the specs), so the drawing and the build stay in agreement.
Don't patch the code and hope the specs catch up. Pick **Revise the plan** on the dashboard: it
edits the specs (and only the specs), so the drawing and the build stay in agreement.
---
## Prefer the terminal?
The same journey is four commands, each in a new conversation:
1. `/plan2code-1-plan` with the idea in a sentence. Writes `specs/task-api/PLAN-DRAFT-20260804.md`.
2. `/plan2code-2-document specs/task-api/PLAN-DRAFT-20260804.md`. Writes the overview and phase files.
3. `/plan2code-3-implement specs/task-api/overview.md`, once per phase. It finds the next open phase itself.
4. `/plan2code-4-finalize specs/task-api/overview.md`. Verifies, updates docs, and archives the spec.
Add `--web` to any of them (for example `/plan2code-1-plan --web`) to open that step straight on the
web console without being asked. Every command and argument is in the README's
[Prefer the terminal?](../README.md#prefer-the-terminal) section. You can switch between terminal and
console at any point, because the files under `specs/` are the only state.
## More
- [web-console.md](web-console.md): every screen, setting and recovery path
- [../QUICK-REFERENCE.md](../QUICK-REFERENCE.md): the one-page card
+380
View File
@@ -0,0 +1,380 @@
# Plan2Code Web Console: design proposal
**Status:** implemented in v2.4.0. Kept as the design record, because it explains
*why* the thing is shaped the way it is. For how to use it, see
[web-console.md](./web-console.md); for where the code lives, see
[AGENTS-architecture.md](../.agents-docs/AGENTS-architecture.md).
**Date:** 2026-09-20
**Scope:** an optional local browser UI that `plan2code-0-pathfinder` and `plan2code-1-plan` can offer at startup, built as a shared component other workflow steps can adopt later.
---
## 1. Why this is smaller than it looks
Pathfinder is already written against a *delivery-channel abstraction*. Its rules say:
> AskUserQuestion when available; prose Q blocks only as fallback.
and `grilling.md` specifies the question object in full: a plain two-or-three-word header, the probe with its why-it-matters, two to four genuine options each described by its trade-off with exactly one named as the recommendation, plus an automatic `Other` escape hatch.
That is a form. The console is not a new workflow, it is a **third delivery channel** alongside the structured tool and prose. The methodology, the file formats, the gates and the rules are untouched.
Plan is the opposite: pure prose Q and A with seven phases and four named approval gates. But every one of its gates already resolves to a literal typed token (`approved`, `needs changes`, `approve / refine / abort`). That gives us the second half of the idea.
### The keystone: answers map back to the literal token
Every console request carries two extra fields:
| Field | Purpose |
| --- | --- |
| `fallbackText` | the exact terminal phrasing the prompt would have printed |
| `token` | what each button resolves to: `approved`, `H`, `A`, `S 1,3,5`, `yes`, `skip` |
Clicking **Approve** in the browser is then byte-identical to typing `approved` in the terminal. No prompt logic ever forks. If the console is absent, refused, or times out, the agent prints `fallbackText` and the session continues in the terminal with nothing lost. Mid-session switching in either direction is free, because the markdown files under `specs/` remain the only source of truth.
This one decision is what keeps the integration from metastasising across twelve prompt files.
---
## 2. Hard constraints found during research
These are verified, not assumed. They shape everything below.
### 2.1 The character limit is the binding constraint
`scripts/validate-char-count.js` caps every flat `src/plan2code-*.md` at 20,000 characters since 2.4.1 (11,000 when the numbers below were measured), enforced by a husky pre-commit hook. Measured today:
```
10995 5 plan2code-0-pathfinder.md <- 5 characters of headroom
10676 324 plan2code-1-plan.md
10835 165 plan2code-2-document.md
```
Neither target prompt can absorb the integration inline. **Trimming pathfinder by roughly 250 characters is a prerequisite task, not a nice-to-have.** Reference files are exempt from the limit, so all of the console contract lives in one.
### 2.2 Packaging needs zero installer code changes
`computeExpectedSkills()` in `install.js` copies `additionalReferences` directories verbatim through `listFilesRecursive()`, which has no extension filter. The initial implementation fanned one shared source directory into two skills; the current installer fans it into every skill.
Current packaging constraints:
- Reference files are copied as `Buffer`s, so binary assets are supported; the shipped MP3 chimes exercise this path. Browser-served assets must also be added to the server's `SERVABLE` allow-list and covered by a test.
- `skills/` is generated build output (committed, and checked for drift by `npm test`) and `buildSkills()` prunes files it doesn't expect. Assets must originate in `src/` and never be hand-placed.
### 2.3 `specs/` is gitignored, which is an argument *for* the server
Three prompts repeat the warning that Glob silently fails on `specs/` and only a shell listing works. A Node server enumerating candidate spec directories server-side and handing the page a list is strictly better than either a terminal listing or a browser file input.
### 2.4 Formats that must round-trip byte for byte
- `<!-- METRICS_JSON {...} -->` comments
- the `## User Feedback` table (exact field names `Rating`, `Reason`, `Went Well`, `Went Poorly`, pipes escaped as `\|`)
- the five-state checkbox alphabet `[ ]` `[/]` `[x]` `[?]` `[!]`, with the invariant "never reset `[/]` to `[ ]`"
- the handoff string `**Status:** Phase 3 Complete - Resume at Phase 4`, plain ASCII hyphen, single spaces
The console carries all of these as opaque strings and never reformats them.
### 2.5 Two regex landmines
- `plan2code-metrics` scrapes confidence with patterns like `/[Rr]equirements?[:\s|]+(\d{1,2})/` that **do not require a percent sign**. Nothing the console writes under `specs/` may contain bare `Requirements` / `Feasibility` / `Integration` / `Risk` followed by a number.
- The loop parses `TASK_COMPLETE`, `PHASE_COMPLETE` and friends. Pathfinder already forbids these tokens under `specs/`.
Both are cheap to enforce mechanically in the server's write path.
---
## 3. Architecture
### 3.1 Placement
```
src/web-console/
console.mjs CLI: open | post | wait | status | stop
server.mjs the HTTP server (spawned detached by `open`)
lib.mjs paths, atomic writes, patch merge, validate()
console.md the agent-facing contract
public/index.html markup shell only
public/app.js the client (no inline script, see CSP below)
public/app.css
public/render.js markdown tokens to DOM nodes
public/vendor/marked.esm.js pinned 15.0.7, MIT, lexer only
```
Wired in with one data entry per consumer:
```js
{
source: 'plan2code-0-pathfinder.md',
// ...
additionalReferences: [{ source: 'web-console', target: 'web-console' }]
}
```
Lands at `~/.agents/skills/<skill>/references/web-console/` on all ~28 target agents. The server finds its own siblings via `path.dirname(fileURLToPath(import.meta.url))`.
The agent needs an absolute path to invoke it. `ai-assist-git-pr` already establishes the convention, proven on Windows, and we reuse its wording:
> `<SKILL_ROOT>` = the directory containing this `SKILL.md` (its install location). Use the absolute path you loaded it from.
### 3.2 The transport, which is the genuinely hard part
A turn-based agent must wait up to twenty minutes for a human. Claude Code's Bash tool caps at 600s and **harness-managed background tasks are not a safe place to wait**: there are open issues documenting them being killed at 5 to 20 minutes, killed on compaction, killed when a new prompt arrives, and leaked with no handle when spawned by subagents. Detached processes are explicitly immune.
So: **split into two processes with deliberately different lifetimes.**
```
# once, at handoff
node <SKILL_ROOT>/references/web-console/console.mjs open --file request.json
-> spawns server.mjs DETACHED, prints URL + session id, exits in under 300ms
# repeatedly, until done
node ... console.mjs wait --seconds 240
exit 0 -> stdout is the result JSON. Done.
exit 10 -> {"status":"waiting","elapsed":"6m12s","answered":3,"total":7,...}
Agent relays one line to the user and calls wait again.
exit 20 -> server is dead. Agent restarts it, re-reports the URL.
exit 30 -> human hit Cancel.
```
Why bounded rather than a single long block:
- The deadline lives in our code, not the harness's. `--seconds 240` covers a 20-minute session in five turns. Drop to `--seconds 90` for harnesses with a 120s default (Cursor auto-backgrounds and cannot be told not to).
- Every return is a free heartbeat: "still waiting, 3 of 7 answered, 6 minutes in" beats silence.
- Interruption is survivable. Ctrl-C the agent, compact the context, restart: the next `wait` re-attaches, because state is on disk.
The waiter uses `fs.watch` on the session **directory** (watching a file breaks on Windows when the file is replaced by rename) with a 500ms `stat` fallback as the authority, and pings `/health` every 5s for liveness. The result is delivered **through the filesystem**, not HTTP, which decouples the waiter from the server and is also the resumability story.
**Optional push accelerator, Claude Code only:** at `start`, also fire a backgrounded `until [ -f <dir>/result.json ]; do sleep 2; done`. If it survives, submit-to-agent latency drops to near zero. If it gets killed, nothing breaks, because the poll loop is the source of truth.
### 3.3 Port, lifecycle, security
| Concern | Decision |
| --- | --- |
| Port | `listen(0, '127.0.0.1')`, handle file in `os.tmpdir()/<tool>/<sha1(cwd).slice(0,8)>/`. This is Jupyter's `jpserver-<pid>.json` pattern and RFC 8252 §7.3. A fixed port is fatal here: you run git worktrees, so two agents in two checkouts of one repo is routine, and a fixed port makes that silently wrong. |
| Bind | the literal `127.0.0.1`, never `0.0.0.0`, never the string `localhost` (which may resolve to `::1` and produce the classic "server is up but the browser gets ECONNREFUSED" bug on Windows). |
| Auto-open | best-effort, never fatal. Skip when `DISPLAY` and `WAYLAND_DISPLAY` are both unset, or `SSH_CONNECTION` / `REMOTE_CONTAINERS` is set. Honour `BROWSER=` and `--no-open`. |
| Orphan prevention | a **server-side idle deadline** (30 min, bumped by any authenticated request) is the mechanism, not a nicety. Windows has no real SIGTERM, and a detached server with no console never sees `SIGINT`, so signal handlers cannot be relied on. Plus a 4h absolute cap, a `pagehide` beacon, an explicit `stop`, and a reap sweep on every `start`. |
| Liveness test | `process.kill(pid, 0)` **and** a `/health` check whose returned session id matches the handle. PID reuse is real; the sid match is what disambiguates. |
Security, six controls, about 60 lines:
1. Bind `127.0.0.1`.
2. **`Host` header allowlist.** This is the DNS-rebinding defence and the highest-value line in the server. Vite shipped GHSA-vg6x-rcgg-rjx6 for exactly this omission.
3. 128-bit token **in the URL path**, not the query string. Path placement also kills the Windows `cmd /c start` `&`-injection hazard, because the URL then contains no `&`.
4. First `GET /s/<token>/` sets `HttpOnly; SameSite=Strict` cookie and 302s to `/`. Token leaves the address bar, history and pasted screenshots; CSRF protection comes free.
5. `Origin` check on every mutating request.
6. Strict CSP with no `unsafe-inline`, which is why JS and CSS are separate files rather than one inlined HTML.
**On whether XSS matters here.** The tempting argument is "it is the user's own machine and the model would not write `<script>`". That misses the real risk. The page is same-origin with a server that reads and writes the user's files and holds the session token, so script in that page can post arbitrary answers back to the agent, which then acts on them with the agent's full tool permissions. And the markdown is only nominally agent-authored: in practice the agent is summarising a dependency README, a scraped page, a PR description, a log. Treat it as attacker-controlled.
The structural fix, rather than a filter: **vendor `marked` for `lexer()` only and render tokens to the DOM with `createElement` and `textContent`.** No HTML string is ever produced, so there is nothing for a sanitiser bypass to bypass, and the strict CSP holds. Roughly 120 lines of renderer. Hand-rolling the parser instead is a trap: hand-rolled markdown fails exactly where agent-authored technical documents live (a table inside a list item, a fenced block containing backticks) and its failure mode is usually an injection hole.
### 3.4 Browser transport
SSE primary (about fifteen lines on Node's built-in `http`), with automatic fallback to 3s interval polling after three consecutive errors. **Exactly one `EventSource` per page, ever**: browsers cap HTTP/1.1 at ~6 connections per origin, and a second stream per tab across three tabs wedges the origin with no error and no timeout. Multiplex everything through one stream with a `type` field.
The page shows the **agent's** liveness, not the server's. If the agent's turn crashed, the server is still happily up and a naive spinner lies forever. Stamp `agentLastSeen` on every waiter ping; when stale past two minutes the page says so plainly: "The agent has not checked in for 4 minutes. Your answers are saved. Check your terminal."
### 3.5 State on disk
The filesystem is the database and the result transport.
```
~/.plan2code/console/sessions/<sid>/
session.json sid, cwd, workflow, title, phase
request-<rev>.json the agent's payload, immutable per revision
draft.json in-progress answers, debounced ~500ms write
result.json written ATOMICALLY on submit (temp + fsync + rename)
events.ndjson append-only, doubles as SSE Last-Event-ID replay
```
Deliberately **not** under `specs/`: that keeps the console clear of the metrics scraper and the loop's marker parser, and survives a `git clean`.
Draft persistence is the highest-value feature in the tool and among the cheapest. Twenty minutes of a product manager's structured answers must never be lost to a closed tab. Server-side debounced write plus a `localStorage` mirror keyed by `sid + rev`.
Recovery matrix:
| Failure | Recovery |
| --- | --- |
| Tab closed or browser crashed | Reopen the URL, draft restored, agent never notices |
| Server died, agent alive | `wait` returns exit 20, agent runs `start --resume <sid>`, new port, same session dir |
| **Agent session died, human still working** | Human submits, `result.json` lands on disk. A fresh session runs `status`, finds the unconsumed result, picks up where it left off |
| Machine rebooted | Handle file is gone (correct), session dir survives, `--resume` works |
---
## 4. The interaction contract
### 4.1 Item envelope
```jsonc
{
"id": "i7", "topic": "t2", "kind": "choice",
"title": "Export format", // PLAIN 2-4 words, never internal jargon
"body": "markdown: the probe plus its why-it-matters",
"options": [ { "k": "A", "text": "CSV", "detail": "trade-off", "recommended": true } ],
"allowOther": true,
"required": true,
"status": "open",
"answer": null,
"thread": [],
"fallbackText": "the exact terminal phrasing",
"token": { "A": "a", "B": "b" } // maps a click back to the literal reply
}
```
Note `recommended` is **optional per option**. Handoff explicitly wants its inferred task offered as the recommendation; the `.gitignore` question in handoff explicitly does not ("present it, don't force it").
### 4.2 Primitives
Tier 1, needed for pathfinder and plan, build now:
| Kind | Covers |
| --- | --- |
| `choice` | Step 0 intent gate, all six destination probes, the 14 frontier axes, plan's coverage and scope questions |
| `multi` | plan's testing types (Unit / Integration / E2E / None) |
| `text` | sacrificial boundary, the embarrassment question, kebab-case feature name **with live pattern validation the terminal cannot do** |
| `confirm` | the Tier-2 sketch gate, the lock offer, abort confirmations |
| `recap` | the recap-and-confirm turn. Non-skippable by design |
| `review` | plan's four approval gates, `1b`'s spec-update gate, paper-sketch reactions |
| `checklist` | `legwork · HITL`: numbered steps plus Done / Blocked / Need help / Defer plus a values field |
| `menu` | the fork menu: up to three takeable questions plus Start fresh, one recommended, with the resume command as copy-to-clipboard |
| `list` | orderable and editable rows, for plan Phase 6's implementation phases |
Tier 2, design the schema to admit them but do not build yet: `findings-triage`, `table-review`, `form`, `path-picker`, `command-handoff`, `notice`, `progress`.
### 4.3 Rules the server enforces mechanically
Several prose rules become checks, which is the quiet advantage of routing through a server at all:
- **Reject a patch leaving more than three open required items.** Grilling says "the cap is three regardless of what the environment allows. The ceiling is the human's attention, not the tool's schema." A web UI tempts you to show ten. Make the rule mechanical.
- **Reject internal jargon in `title`**: `grill`, `frontier`, `fog`, `HITL`, `Locked:`, `Blocked by:`. There is an explicit do-not-say list in `grilling.md`; this is where to enforce it.
- **One authoritative gate per session.** `plan2code-3-implement-review` proves several prompts each carry an approval gate while only one may fire when composed. A `gate: { id, authoritative }` field plus a server guard.
- **Defer is a third verb**, distinct from reject.
- Reject forbidden tokens and bare confidence-word-plus-number patterns in anything written under `specs/`.
- Stamp every timestamp server-side. Models do not reliably know the time.
---
## 5. The UI
Three panes, plus view tabs, closer to pathfinder's own vocabulary than to a grilling UI.
```
┌──────────────────────────────────────────────────────────────────────┐
│ 🧭 audit-export · Pathfinder · Working 4 of 9 cleared ▓▓▓▓░░░░░ │
│ ● Waiting on you · 3 questions [Questions] [Map] [Draft] │
├───────────────┬──────────────────────────────┬───────────────────────┤
│ THE MAP │ What you end up with │ DISCUSSION │
│ │ │ │
│ ✓ Destination │ Why this matters: ... │ thread on this item │
│ ✓ Codebase │ │ │
│ ● Export fmt │ ○ A CSV, ... │ │
│ ○ Size cap │ ◉ B Parquet RECOMMENDED │ │
│ ⊘ Retention │ ○ C Both, ... │ │
│ │ ○ Something else ______ │ │
│ NOT YET CLEAR │ │ │
│ · fog item │ Not sure, skip for now │ │
├───────────────┴──────────────────────────────┴───────────────────────┤
│ 2 staged · Export format → B · +1 comment ⌘↩ [Send to Agent] │
└──────────────────────────────────────────────────────────────────────┘
```
### What to take from grill-with-ui
- **Batch staging with one explicit Send.** Everything is local and reversible until one deliberate press. For a non-technical user this removes the terror of "did I just commit to that?"
- **The three-tier commitment vocabulary**: staged (amber, removable) → sent and pending (dashed, spinner) → recorded (green check). Honoured consistently on every surface.
- **The whole stuck-agent apparatus**: live elapsed counter, disabled Send while working, a five-minute grace that re-enables Send with an explanation, server-gone detection with self-reconnect. About 40 lines, and more honest about agent latency than most commercial agent UIs.
- **Assumed regions.** The best idea in that repo: the generated artifact marks regions that depend on still-open questions with a dashed outline and "assumed, pending Q3". This transplants directly onto a plan taking shape.
- **Versioned artifacts with a staleness flag** and a one-line change note.
- `withInputs()` focus preservation, raw-string diffing before re-render, and the `prefers-reduced-motion` handling that keeps progress indicators legible rather than freezing them.
### What to change
- **Topics, not rounds.** Grill groups the nav by round, which is a temporal artifact: "questions the agent happened to ask on turn 4". Its own sample data has one round containing three unrelated subjects. Product people cannot use that. Group by topic, stable and nameable.
- **Show real progress.** Grill has no "N of M" counter at all. Pathfinder's trail footer already computes `4/9 cleared` and a plain-English confidence line. Put both in the header. In a 45-minute session with a PM, "how much longer is this" is the first question asked.
- **Recommendations become optional.** Grill always leads with one, which is right when the agent has an opinion and you are pushing on it. In discovery the agent frequently should not have an opinion, and pre-filling one anchors a PM into rubber-stamping. Neutral mode: no `rec`, or reframed as "what similar products do".
- **Real, accessible form controls.** Grill's options are `<li>` with `onclick`, so they are unreachable by keyboard and invisible to screen readers. Non-negotiable to fix if non-technical stakeholders are in scope.
- **Markdown actually renders.** Grill escapes everything and sets `white-space: pre-line`; there is no markdown at all. A plan document cannot be shown that way.
- **Rename everything.** "Grill", "durable", "Defer", "Reopen", "Why A." is a vocabulary for interrogating an engineer who already has a plan.
- **Do not dim rejected options to 45%.** In discovery the rejected options are often the most valuable thing in the room, and an engineer joining late needs to see why B lost.
### Two things the console unlocks that the terminal cannot
- **Tier-2 UI sketches.** `resolve.md` already sanctions building 3 to 5 structurally different UI variants switched by "a URL search param plus a small floating bar". The console can host them in a sandboxed iframe with a proper variant switcher. This is already in the methodology and currently awkward.
- **Anchored reactions.** `resolve.md` insists every sketch probe "names its row or element". With `anchors: { "row5": "b3" }`, clicking the probe highlights the row. That is the inline-commenting capability, scoped to something the methodology already demands.
### Room-screen polish (v2.4.0)
Built for a mixed product and engineering room reading one screen together.
- **One width variable.** `--card-width` on `:root` in `app.css` sets the column the card, the pager and the document tabs share (`max-width: min(var(--card-width), 100%)`, so a narrow window still wins); no fixed `720px` column rule is left outside a media query. Three presets, picked under *Card width* in the looks panel and saved as `looks.width`: `narrow` 720px, `comfortable` 880px (the default, and the variable's own value), `wide` 1080px, applied as `:root[data-width="…"]`. `CARD_WIDTHS` in `palette.js` is the same list for the page's validation and the server's first-paint `data-width`; **`palette.js` and `app.css` must agree**, which is the one place the number is written twice. The dashboard keeps its own 860px layout. Inside the card the `62ch` / `68ch` measure caps come off (`.card .md, .card .prompt { max-width: none; }`) so a wider card is not a narrow column in a wide frame; documents keep `.md`'s 68ch.
- **One box grows.** On the primary card, the answer box that is being typed in — a `text` question's box, a choice's *Something else* box once picked, or the note under a verdict / confirm / recap once shown — is `textarea.grow`, and the card takes the column's full height above the pager (`.card.is-primary.has-grow { flex: 1 1 auto; }`). `fitGrowBox()` in `app.js` writes the card's free height as the box's `min-height`, **never under 160px**; a `ResizeObserver` watches the main column rather than the box, so the box growing can never retrigger it, and it writes only on a change over 1px. `resize: vertical` with that moving floor makes the handle taller-only; past the fill the card scrolls. Every other box keeps its ordinary size, and the notes panel is untouched.
- **A switch per sound.** Every cue is an event in `SOUND_EVENTS` (`palette.js`), each naming its file and rank: `startSkill` (`start-skill.mp3`, played once the dashboard's **Start** is accepted) 5, `bootup` 4, `sessionStart` / `sessionEnd` (`start-stop.mp3`) 3, `question` / `skillReady` / `chatReply` (`next.mp3`) 2, `newTab` (`insert.mp3`) 1, `snore` (`sleeping.mp3`, looped by the wake-up outside the queue) 0. The queue, the held slot and the one-at-a-time rule in `app.js` work on events, one `Audio` per file shared by the events that use it. User Preferences shows a *Play sounds* master (`looks.sound`) over one box per event, saved flat as `sound<Event>` (`soundStartSkill`, …) because `looks.json` only holds flat values; `soundPrefs()` turns any missing or non-boolean switch on, so a file from before the switches keeps every sound. A cue plays only while the master and its own switch are both on; the master off greys the boxes but keeps their ticks, and ticking a box plays it.
- **Send breathes.** `#btn-send:not(:disabled)` runs `send-pulse`, a soft accent ring that spreads and fades every 2.4s; a disabled Send is still, and reduced motion switches it off.
- **Per-tab scroll memory.** An in-memory `Map` keyed by view id holds the scroller's position (the main column on a document or the Overview, the card on the questions view, with the item it belonged to), restored when the tab comes back. Page-local, never persisted, cleared when the session turns into or out of the dashboard.
- **The Overview tab.** Appears only while `GET /overview` returns a file for the session's spec (on the dashboard: the picker's spec; gone for *start from scratch*), fetched with `no-store` on load, on a spec change and on every tab click, so it always shows the file as it is on disk. Rendered through the same DOM-only markdown path as every doc, captioned read-only. It never becomes a doc: the page is not the record.
- **The footer slot.** The workspace label sits centred in the footer's middle track, 12px and muted: `footerLabel()` in `public/workspace.js` builds `folder · branch` (from the workspace's original folder and `state.branch`), or `folder + N folders · branch` once folders are added, and `renderWhere()` steps down through its narrower tiers (`FOOTER_TIERS`) until the text fits. A warning dot marks a missing or unreadable folder. A click opens the Workspace dialog (`#workspace-modal`), where each folder has its own **Copy path**. The notification pill is fixed over the same spot and covers it while it shows, on purpose: nothing there needs pressing while the pill is up. The dashboard keeps the footer for this label alone: `renderFooter()` puts `is-dash` on it, which hides the summary and Send (and on a narrow screen keeps the label, the one thing left in the bar); the footer only disappears there when there is no label to show.
- **The wake-up.** A fresh dashboard (`state.wake`, plus a per-sid `sessionStorage` guard) plays Planny powering on centre stage, as proven in `pathfinder/sketch-01`. The wake layer hides the dashboard behind a solid `var(--bg)` backdrop (`.wake::before`, so it follows the theme) whose opacity fades out over the same 0.6s as the glide, revealing the menu as he lands; the notification pill (`.banner`, z-index 46) stays above the layer (45). Before the first click or key he sleeps there on loop under a "Click to wake Planny" hint (`.wake.dormant`): Z's rising, slow breathing, slack arms swaying out of step, his feet pulsing like a standby light. While he sleeps the page tries to loop `sleeping.mp3` (browsers usually refuse until the first click, so this is normally silent). That click fades the hint and nudges him (`.wake.stirring`): he rocks gently about his feet (`wake-nudge`) while exactly one snore plays from the top, via the `stir()` callback `playWake()` awaits — timed by the snore's `ended`, capped at 6s, or a 2.6s timer when sounds are off or refused. Then the boot starts to `bootup.mp3` (in place of the opening chime; outranked only by the Start-skill cue, which cannot coincide with it because the wake layer swallows every click and key until the run ends), which fades out over 0.5s the moment the run ends or is aborted. The boot pops the Z's outward: 0.4s power-on and his feet blink · 1.35s feet plant · 1.5s / 1.9s arms lift off his sides · 2.5s the star opens lit, with a scan · 2.85s he glances left, right, then ahead · 3.2s smile · 3.6s hop and a cheer, star back to ink · 4.3s a 0.6s glide into the rail, measured once by `boot.js` and handed to CSS as `--wake-glide` · the menu groups rise in 60ms apart as he lands, and the note types in. Done by 5s (a fallback timer covers a hidden tab whose transition never fires). The beats are CSS keyframes on delays under `.wake.booting`, all inside `@media (prefers-reduced-motion: no-preference)`; there is no skip: from the waking click on, clicks and plain keys (no Ctrl/Meta/Alt) are swallowed so the nudge, boot and sound always play out; only a launch mid-run aborts it (snore stopped, boot-up faded), and under reduced motion `playWake()` resolves at once and the awake dashboard is simply there.
---
## 6. The startup question
Both prompts open with one ask. Pathfinder already uses `AskUserQuestion`, so this costs almost nothing:
> Run this session in the **web console** (a local page in your browser, better for walking through decisions with someone) or **here in the terminal**?
Recommended default is an open decision (see §8). Because the markdown files stay the source of truth, the user can switch either way mid-session at no cost, and the prompt should say so.
---
## 7. Work plan
| # | Task | Notes |
| --- | --- | --- |
| 0 | **Trim `plan2code-0-pathfinder.md` by ~250 characters** | Hard prerequisite. 5 characters of headroom today. Candidates: the `grilling.md` fallback blockquote, the Rules list. |
| 1 | `src/web-console/` skeleton: `console.mjs` start/wait/status/stop, detached spawn, handle file, reap sweep | The transport is the risky part, build and prove it first |
| 2 | `server.mjs`: the six security controls, SSE, atomic result write with Windows `EPERM` retry | |
| 3 | `index.html` + `app.js` + `app.css` + vendored `marked` lexer, token-to-DOM renderer | |
| 4 | Tier 1 primitives, staging, drafts, Send | |
| 5 | `console.md`: the agent contract | The only file the prompts point at |
| 6 | `SOURCE_PROMPTS` entries + `npm run build:skills` + commit `skills/` | Zero installer code changes |
| 7 | Pathfinder integration: startup ask, channel rule, `Read` line | ~250 chars, funded by task 0 |
| 8 | Plan integration | 324 chars headroom, fits, trim ~150 for margin |
| 9 | CHANGELOG + `version.json` + `package.json` in lockstep | `plan2code-publish` refuses to release on mismatch |
### Three things most likely to bite during implementation
1. **Forgetting `windowsHide: true` on the detached spawn.** It defaults to `false`, and `detached` on Windows forces a console window that "once enabled, cannot be disabled". Every handoff flashes a black box at the user.
2. **`fs.rename` `EPERM` from antivirus and search indexers on Windows.** Intermittent, unreproducible on demand, looks exactly like a logic bug. Retry three times with backoff from day one.
3. **Relying on `SIGTERM` for cleanup on Windows.** It does not arrive. Build the idle deadline first, not last, and comment why.
### Non-obvious portability notes
- grill-with-ui's heredoc idiom (`patch <<'EOF'`) is **not portable to PowerShell**, which is the default shell for Claude Code on Windows. Use a `--file <path>` form as the primary documented interface.
- `nohup ... &` is bash-only. Backgrounding happens inside `console.mjs start`, never in the shell command the agent writes.
- Always emit `node "<abs>/console.mjs"` quoted. Install paths contain spaces.
---
## 8. Open decisions
1. **Default channel.** Terminal (lower ceremony for an engineer working solo) or console (better for the cross-functional sessions that motivated this)? Recommendation: console marked as recommended in the ask, terminal as the low-ceremony alternative.
2. **Vendoring `marked`.** It is MIT and dependency-free, but it is the first vendored third-party code in the repo, against a house style of "Node built-ins only". The alternative, a hand-rolled parser, is materially worse on both correctness and security. Recommendation: vendor it, pin the version, record the SHA-256.
3. **Single-driver or multi-user.** Everything above assumes one browser, one driver, no identity or attribution. A facilitated workshop with one person at the keyboard works fine. Two people answering independently is a different product and reaches into the state model. Recommendation: single-driver for v1, say so explicitly.
4. **Third consumer.** After pathfinder and plan, `/plan2code-review` is the strongest case: two of its rules exist only because output goes to a terminal ("for 15+ findings, present top 10", "output tables as direct markdown, NOT inside code blocks") and a browser deletes both. Its `S 1,3,5` fix selection is literally a checkbox list compressed into a CLI token, and one findings UI serves both `/plan2code-review` and `/plan2code-3-implement-review`.
---
## 9. Honest assessment of size
This is not a weekend. Rough shape: `console.mjs` plus `server.mjs` around 700 to 900 lines, the client around 900 to 1,200, the contract file 300 to 400 lines of prose, plus the prompt trims and build wiring. The transport and lifecycle work (tasks 1 and 2) is where the risk concentrates, and it is worth proving end to end on one primitive before building the other eight.
A smaller first cut that still tests the whole idea: pathfinder only, `choice` and `text` and `recap` only, no artifact pane. That exercises the transport, the token mapping, the fallback path and the three-probe cap, which are the parts most likely to be wrong.
---
## Credits
The architecture owes a large debt to `grill-with-ui`, whose file-ownership model, staged-then-send interaction, commitment tiers and assumed-region idea are all borrowed here. The implementation is independent.
+799
View File
@@ -0,0 +1,799 @@
# The Web Console
## New here? A two-minute tour
**Launch it.** After installing, run `plan2code` in your project folder. It starts Claude Code or
Devin on the dashboard, and the page opens in your browser. In any other agent, run `/plan2code`
inside it, or add `--web` to a skill (`/plan2code-1-plan --web plan a lunch voting app`).
**What you see.**
- **The dashboard** has a card for every skill. The one it suggests next is flagged. Pick a card and
that skill starts on the same page.
- **Questions** come laid out with the trade-offs next to each option and a progress bar on top.
Nothing is sent until you press **Send**, and half-typed answers survive a closed tab.
- **Tabs** show the document taking shape, the build progress and, always last, **Ask**, where you can
put a quick question to the running agent.
- **Sign-off** asks for your approval at the end of a phase. You can run the code review first from
the same card.
**Where next.** [walkthrough.md](walkthrough.md) follows one feature from the first idea to the
last phase. The rest of this file is the reference: every card, button and rule. Prefer the terminal?
Run the same commands without `--web` and answer there; the files under `specs/` are identical.
---
The web console is the main way to use Plan2Code: a local page in your browser
where you pick a step, answer its questions, watch the documents and the build
take shape, and approve the work. Every skill runs on it, and `/plan2code` is
the front door: it opens a dashboard of all of them, and whichever you pick
starts right there on the same page.
It exists because the hardest part of adopting Plan2Code is not the engineering.
It is getting the person who actually knows what the product should do into the
room. Pathfinder and Planning are the two steps where that person matters most,
and they were the two steps that looked like a wall of terminal text.
## Quick start
1. Install Plan2Code (see [Install and start](../README.md#install-and-start) in the README).
2. Run `plan2code` in your project, or `/plan2code` inside your agent. The dashboard opens in your
browser.
3. Pick a card. The same page becomes that step's session.
4. Answer on the page and press **Send to Plan2Code**.
Starting a step by its own command instead? Add `--web` (for example `/plan2code-1-plan --web`) and
it opens on the page without asking.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="../docs/screenshots/dashboard-dark.webp">
<img src="../docs/screenshots/dashboard-light.webp" alt="The Plan2Code dashboard: a Set up card, then the steps in order as cards, one of them marked Suggested.">
</picture>
---
## What you get
- **Every question laid out**, with the trade-offs written next to each option
instead of scrolling past in your history.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="../docs/screenshots/question-dark.webp">
<img src="../docs/screenshots/question-light.webp" alt="A Plan question card, How tasks are paged, with three options, the first marked Suggested, each with its trade-off underneath.">
</picture>
- **A progress bar.** "4 of 9 settled" answers the first thing anybody asks.
- **Previous and Next under each card**, walking the questions still waiting, so
the list on the left is somewhere to jump from rather than the only way round.
- **An example you can accept.** Where a question comes with one, an *Autofill
suggestion* button puts it in the box as a real answer for you to edit. It
disappears as soon as you have typed anything of your own.
- **Answers you have settled read as answers**, not as a form you might still be
able to change: what you chose, stated in words, with everything else locked.
- **A question you have sent is held** until Plan2Code answers it, showing
what you sent. The rest of the page carries on working, so you can read ahead
or answer the next question while you wait.
- **The document taking shape**, in a tab, with the parts that are still guesses
clearly marked as guesses.
- **A brief, on demand** (Pathfinder). Press *Write a brief*, choose how far back,
and what comes back is a plain-English report of what has been decided, what is
still open and what happens next, in its own tab with Print and Save buttons.
It is the thing you take to a meeting.
- **Nothing is sent until you press Send.** Everything before that is local and
reversible, so nobody has to worry about committing to something by accident.
- **Nothing gets lost.** Close the tab, reload, restart your machine: your
half-typed answers are still there. Once you press Send, what you sent is
held until Plan2Code actually collects it, so it cannot be written over.
- **It tells you when the other half stops listening.** The terminal and this
page are two programs that fail separately, and a page that sat there saying
"one moment" forever was the worst way to find that out. Now it says so, and
says what to do: answering anyway still works, because what you send is held
on disk until something comes for it.
- **Starters picked for your role.** Tell *User Preferences* what you do
(engineer, product lead, architect, designer, QA, engineering manager) and
the starter templates on an idea card and the starter questions on the Ask
tab put the most useful ones first. It changes the order, never the
questions a skill asks (see *Your role* below).
- **Light or dark, and a highlight color you pick.** The gear button at the
top right opens *User Preferences*: your role first, then these. It follows
your system by default; set it to one or the other if
you would rather it stayed put. Remembered on your machine, so it carries
over to every session.
- **A card as wide as you want it.** The same panel has a *Card width*
setting — **Narrow** (720px), **Comfortable** (880px, the default) or
**Wide** (1080px) — that widens the question card, the Previous / Next pager
and the document tabs together. The dashboard keeps its own layout, and on a
small window the column simply fits the window. Remembered with the colors,
so it follows you into every session.
- **Text that uses the card.** A question's text runs the full width of the
card rather than stopping in a narrow column, and the answer box you are
typing in grows to fill the card's free height: the box of a text question,
the *Something else* box once you pick it, or the note under a verdict,
yes/no or recap once it shows. Other boxes stay small. Drag the handle to
make it taller; it never shrinks below the space it was given, and on a short
window the card scrolls instead of squeezing the box under 160px.
- **Send tells you it is ready.** While there is something to send, the Send
button pulses gently; it goes still when there is nothing to send.
- **Each tab remembers where you were.** Scroll halfway down a document,
jump to the questions, come back: you are where you left off. The same goes
for a long question card. It lasts as long as the tab is open.
- **The spec's overview, one click away.** When the session is working on a
spec that has an `overview.md`, an **Overview** tab shows it — read-only,
and read fresh from disk each time you open the tab, so an edit shows up
straight away. No file, no tab. On the dashboard it follows the spec picked
at the top, and disappears for *start from scratch*.
- **Where you are, at the bottom.** The footer names the folder the session
was launched from and its git branch (`plan2code · feature/…`), muted in
the middle. Click it to open the Workspace dialog (see *Your workspace*).
Outside a git repo it shows only the folder.
- **A chime when it's your turn.** The page plays a sound when new questions
arrive, when a new tab (a document, the overview) joins the strip, and when
the session opens or closes, so you can look away without missing your cue.
Starting a skill from the dashboard plays its own sound too. The same panel
has a *Play sounds* switch for all of them, and a box per moment (Planny
snoring, Planny waking up, starting a skill, session opens, skill is ready,
session ends, new questions, a reply on Ask, a new tab), so you can keep only
the ones you want. Ticking a box plays its sound; everything is remembered
like the colors.
- **Files in your notes.** Show instead of describe: attach a screenshot, a
photo or a document to any note or quick question (see *Attachments in notes
and questions* below).
- **A nudge when it might be waiting on you.** Your agent may stop in the
terminal to ask permission for something, and the page cannot see that. So
the dashboard and every starting screen carry a one-line tip about hands-off
modes (`auto` mode in Claude Code, `smart` mode in Devin) that can cut down
on those stops, though they may still need your attention, and when the agent
has been working and silent for about 30 seconds, Planny's line asks
*"Still working, or waiting on an approval in your terminal?"* — well before
the two-minute *"Still working on it"* line.
- **The tab icon says whose turn it is.** A bar along the bottom of Planny's
favicon follows the line beside him: dashed amber while Plan2Code is
working or picking up your answers, solid green when it is your move, solid
red when it has gone quiet or the connection is lost (busy is dashed so the
states differ by pattern, not only color). A finished session goes back to
the plain icon, so a console left in a background tab can be checked at a
glance.
- **Help.** The **?** at the top right opens a Help dialog with 16 topics,
one per tab down the left (across the top on a narrow window), opening on
the one for where you are. It is part of the page itself, so it still reads
fully when the page has lost its program.
- Works offline, on your own machine only. No account, no install, no network.
An Implement phase ends on a sign-off card: read the completion report, then
review the code first, approve the phase, or ask for changes. Implement +
review has already reviewed the code, so its card has no review button.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="../docs/screenshots/signoff-dark.webp">
<img src="../docs/screenshots/signoff-light.webp" alt="The Approve this phase card, with a button to read the completion report and three choices: Review the code first, Approve this phase, and I want changes.">
</picture>
The **Ask** tab is a chat beside the cards, for a quick question at any point.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="../docs/screenshots/ask-dark.webp">
<img src="../docs/screenshots/ask-light.webp" alt="The Ask tab: the question Where are tasks paged? and Plan2Code's answer naming the file.">
</picture>
## Using it
The easy way in is the dashboard:
```
/plan2code
```
A page opens with every skill laid out as cards — the setup pair, the pipeline
in order, and the utilities alongside — with a note about what it made of your
project (no AGENTS.md yet, a spec waiting on Phase 2) and a *Suggested* flag on
the card it would pick. Click one and Planny flies while it gets ready; a few
seconds later the same page is that skill's console, already running. No
command names to remember.
A freshly opened dashboard starts with Planny asleep in the middle of the page.
Click to wake him: he snores once while he gets a gentle shake side to side,
then about five seconds of him powering on to a boot-up sound (arms lifting
off his sides, his star opening, a hop and a cheer) before he glides into his spot at
the top and the menu rises in. (Browsers stay silent until your first click,
so the snoring usually starts with it; with *Sounds* off, the same wake-up
plays without audio.)
It plays only on a fresh `/plan2code`: not on a reload, not when you come
back with **Back to the dashboard**, and not after a launch. To skip it — from
the asleep screen or mid-animation — press the muted **Skip** button at the
bottom right or hit Escape; either goes straight to the dashboard, silent.
With reduced motion switched on in your system settings you get the awake
dashboard at once.
Or start any step directly by its command (the full list is in the README's
[Prefer the terminal?](../README.md#prefer-the-terminal) section).
The first thing it asks is how you would like to work:
> **How would you like to work through this?**
>
> - **In your browser** (suggested)
> - **Here in the terminal**
Pick the browser and a page opens at a `127.0.0.1` address, straight away:
opening the page is the first thing the skill does, before it reads your
project, so for the first few seconds you see a starting screen naming the step
("Starting Pathfinder") while it gets its bearings — for the core steps, with
Planny acting that step out. The questions replace it the
moment they are ready. Answer what is there, press **Send to Plan2Code**, and
it carries on. You can switch back to the
terminal at any point by saying so: the files under `specs/` are the same either
way, and neither side owns them.
While the page is open, the agent's messages in the terminal end with a
pointer back to it — `→ Look at the web console: http://127.0.0.1:…` — and
while it is waiting on you it says so in those words: *Waiting on your answer
in the web console*. Glance at the terminal, and it always tells you where to
look.
Know you want the page already? Pass `--web` with the command —
`/plan2code-1-plan --web plan a lunch voting app` — and it opens the console
without asking. The flag is dropped before the rest of the argument is read, so
it never becomes part of the request.
Sharing the link with a colleague on another machine will not work, and that is
deliberate. It is bound to your own machine only. To work through a session
together, share your screen.
For the same reason, the web console is only offered when the agent is running
on your own machine. An agent working in a cloud VM, a container or over SSH
stays in the terminal: a `127.0.0.1` link from somewhere else is a link you
cannot open.
### Your role
The first control in *User Preferences* (the gear button) is **Role**:
Engineer, Product lead / PM, Architect / tech lead, Designer (UX), QA / test
engineer, Engineering manager, or **Not set / none of these**, the default.
It only reorders things: the starter templates on an idea card and the starter
questions on the Ask tab. It never changes what a skill asks or does. It is
kept in `looks.json` with your colors, so it follows you into every session,
and **Back to default** sets it back to *Not set*.
Until you have chosen, the dashboard shows a one-line banner at the top —
*Pick your role to get starter templates ordered for you.* — with **Set up
preferences** and **Not now**. Choosing any role, *Not set* included, or
pressing *Not now* puts it away for good.
### Starter templates
The first "What's the idea?" box in Pathfinder, Plan and Quick Task has a row
of starter templates above it: **Blank** first, then outlines to fill in (an
engineering spec, product requirements, a design decision, a bug
investigation, and so on), in the order your role puts
them. With a role chosen, the first one carries a *For your role* pill. Pick
one and the box fills with editable text, `[blanks]` marked; if you have
already typed something, a strip asks **Replace** or **Keep mine** first.
Only what is in the box is sent, exactly as if you had typed it.
The templates live in the page (`public/starters.js`), not in the prompts. A
skill only marks its fresh-idea card with `"templates": "idea"`, and `post`
refuses that flag on anything but a `text` card. A resumed map or plan asks
no idea question, so it shows no templates, and the terminal never has them.
### Leave for later (Pathfinder)
Pathfinder's scope question is **Leave for later** — *"Which of these features
are we leaving out this time?"* It is a tick-any list of 3–5 features someone
might expect, each ending with what leaving it out means ("Scheduled exports
(Leaving this out = someone runs each export by hand)"), with a recommended
set to tick in one go. The features you tick become the first lines of the
map's *Out of scope*. Every option names a feature, never a rule to keep, so a
tick always means "not this time".
### Skipping a question
**Skip this one**, on an optional question, now moves you straight on: the card
reads *Skipped* for a beat (about a quarter of a second) and the next open
question comes up. On the last one, focus goes to **Send to Plan2Code**
instead. A click or a key press in that beat cancels the move. **Do not skip**
on a skipped card brings it back.
### Your workspace
The folder label in the middle of the footer — `shop · main`, or
`shop + 2 folders · main` once you have added some (shorter on a narrow
window) — opens the **Workspace** dialog: the folders the agent uses as
context for this console session.
- **The folder you started in** is always first. It can be renamed and given a
description, but not removed.
- **Add a folder** by pasting a path (`../shared-ui` counts from the folder
you started in; `~` is your home folder) or with **Browse…**, which opens
your system's own folder picker. Browse… is hidden when the page looks like
it is being reached from another machine (SSH, a dev container, a
Codespace). A path that is not a folder, one already in the list, and a name
already taken are each refused with a message (a taken name comes with a
**Use @name-2** button); a folder inside another one in the list is
accepted, with a note.
- **Every folder has an `@name`** (lowercase letters, numbers and hyphens) and
an optional one-line *What is it?*. Type `@` in the Ask box, a note or a
typed answer and a list of names pops up; the agent reads `@shared-ui` as
that folder.
- **What the agent does with them.** It is told the list when its session
opens and each change once after that. Added folders are read-only
reference: it reads them, reads each one's `AGENTS.md` (or `CLAUDE.md`)
before first using it, and changes one only if you ask. Your harness may
ask permission the first time it reads outside the project.
- **Marks.** *This folder no longer exists* when it is gone from disk; *The
agent can't read this yet* when the agent reported a failed read (in Claude
Code with a copyable `/add-dir <path>` to run); *Inside another folder in
the list* for a nested one.
The workspace carries across skills you start from the dashboard and back, and
it is remembered for next time:
- **Saved on every change**, in `~/.plan2code/console/workspaces.json` (beside
`looks.json`, never in the project, since the paths are your machine's),
keyed by the folder you started in.
- **One list per spec.** *Start from scratch* on the dashboard shows the
project's list. Picking a spec loads its own; the first time, it starts as a
copy of the project's, and after that the two are separate. A spec created
mid-run (Pathfinder, Plan) takes the list in force with it. The line under
the dialog's title says whose list it is.
- **Forgotten on archive.** Finalize's Spec Cleanup step says it will forget
the spec's list and runs `console.mjs forget --spec specs/<name>` after the
move.
### Session meter
A small ring pill in the top bar — **Session: fresh** (green), **Session:
getting long** (yellow), **Session: time for a fresh start** (red) — counts the
work this console session has done. Click it for the count, a list of each
step that added points and what it added (for example *Quick task started +1*),
and a **What is this?** link to Help.
Each skill run adds points (`WEIGHTS` in `public/meter.js`): Plan, Document,
Review, Init and Init update 2; Revise plan, Quick task and Finalize 1; a
Pathfinder map 2 and each question its research subagents settle 1; a built phase by its size,
one point per 3 tasks it completed (rounded up), plus 1 for Implement + Review
(a flat 2, or 3, if the agent did not report a task count); Handoff and the dashboard 0.
A build counts while it runs: each task the progress bar ticks off adds to the
build's row (*Implement + Review: 9 tasks built so far*), Implement + Review's
review stage adds its 1 as it starts, and the approved phase takes the row over
with its final score, so the meter never waits for the sign-off to catch up.
On top of that, every 2 questions you answer on the page add 1 point to the
skill run you answered them in (`ANSWERS_PER_POINT`), so a long Pathfinder or a
Plan with many clarifying questions weighs more than a short one. A review run on the
page after a quick task or a phase sign-off counts as a Review too. Green is
under 5, yellow from 5, red from 10, and the ring is full at 15. At red the dashboard
also shows a banner: starting a **new console session** keeps the agent sharp
(resuming this one carries the count along). It is a rule of thumb, not a
reading of the agent's real context, and it never blocks anything.
### While something is being built
Implementation, Implement + Review and a Quick Task spend most of their time
building rather than asking, so the page shows the build instead:
- **What it is doing right now**, beside Planny ("Task 5 of 9: the rate
limiter"), and the progress bar counting tasks done. Tasks are counted
within the phase, to match the bar — the phase is already named in the
headline — so you never see "Task 4.2 of 6"; the page rewrites that form if
an agent still sends it.
- **The phase's task list** in its own tab, ticking as each task lands on
disk, and **the phase file itself** (`phase-3.md`) rendered in the next —
the prerequisites and task specs, not just the checklist.
- **Any question the build needs answered** (which phase, a spec conflict,
failing tests) appears as an ordinary question, and the sign-off is a
**Completion report** tab with an *Approve this phase* card.
- A long task is not mistaken for a stall. The build says how long it may go
quiet, and the page only calls it stuck after that.
### Review before you approve
The *Approve this phase* card carries a **Review the code first** button beside
the approval. Press it and the same session runs a focused code review of
exactly the files the phase changed, before you are asked to approve anything.
The findings come back as a tab, and a *Which to fix* card lists every one with
its severity: tick the ones to fix (or use the *Critical and warnings* / *All
of them* quick picks), press Send, and it fixes and re-checks them before the
approval card comes back.
A quick task has no approval step, so its offer sits on the finished screen
instead: a large **Review it now** button above the next step, with **No
thanks, I am done** beside it. If the session has already closed, the button
turns into the `/plan2code-review` command to run in the terminal.
Implement + Review shows neither: it reviews every phase before sign-off
already.
### When a session ends
A finished session says so, and ends in one of three shapes:
- **A next step.** What happened, and the exact command to run next (with a
Copy button) — the next phase, Finalize, a commit.
- **All done.** Nothing is left to run, so there is no command to copy:
*"All done, nothing left to run."* Init, Init update and Handoff end this
way, and so do Review and Finalize when nothing landed that needs committing.
- **Paused.** You pressed *Stop session*. The card names what was saved and
gives the command that picks it up later. Nothing else is offered.
The first two also carry a **Back to the dashboard** button (beside the next
step, or as the main button under "All done"). Press it and the same page turns
back into the skill menu, in the same conversation, so the next skill starts
right there. The agent keeps listening for it for about ten minutes after the
session ends; after that, or once the session has closed, the button turns into
the `/plan2code` command instead. That command is on both of these screens
anyway, for starting the dashboard in a fresh conversation — the better choice
after a long skill, since a fresh conversation starts with a clean context.
You don't have to finish first. On every skill but the dashboard itself, a
small **triangle** pointing left, in your highlight color, sits beside *Stop
session*. Press it, confirm, and the agent saves your place exactly as a stop
would, with any answers you have ready going along. Then the page turns into
the dashboard instead of showing a finish card. Pick the same skill again to
carry on, or pick another one. Like Stop, it only works on your turn, and where
nothing is saved yet (early in a plan) the confirm says what would be lost.
### Attachments in notes and questions
Any note, any quick question on the Ask tab, and the fresh-idea box on
Pathfinder, Plan and Quick Task can carry files (on a card they travel with your
answer as a note attachment), three ways:
- the **+ Attach file** button beside the box (pick one or more files),
- **paste** a screenshot or a copied file straight into the box,
- **drag and drop** files onto the notes panel or the Ask composer.
Two kinds are allowed:
- **Images.** Each is shrunk and re-encoded in your browser (JPEG, 2000 px on
the long edge at most, phone-photo rotation kept) and shows as a thumbnail.
- **Documents**: text, code and PDF (`.md`, `.txt`, `.json`, `.yaml`, `.csv`,
`.log`, `.js`, `.ts`, `.py`, `.sql`, `.pdf` and the like), up to 10 MB each.
They upload exactly as they are and show as a chip with the file type, name
and size. The server checks the contents match the type (a PDF header, or
UTF-8 text with no NUL bytes), and a document is never opened from the page.
Up to 5 attachments per note and 5 per question, images and documents counted
together. Anything else (`.docx`, a file over 10 MB, a dotfile such as `.env`)
picked or dropped is refused with a message naming the file; a paste only
picks up the files it can attach, and leaves any text to paste as usual. Each
file uploads straight away while you carry on typing, and sits in a tray with a
remove button; removing one deletes its file. A failed upload offers retry or
remove and holds back only that note, so the rest of your answers still send.
The files are stored with the session, in
`~/.plan2code/console/sessions/<id>/uploads/`, never in your project, under a
random name the server picks. When you send, each note is followed by one line
per attachment:
```
Note on Header row: Please keep the header row.
Image: /home/sam/.plan2code/console/sessions/<id>/uploads/<uuid>.jpg (header.png)
File: /home/sam/.plan2code/console/sessions/<id>/uploads/<uuid>.pdf (api-spec.pdf)
```
and the agent opens those paths with its own file reader. Uploads from sessions
untouched for 30 days are swept away the next time a session opens.
**Keeping the ones a spec cites.** When the agent is about to write a file
under `specs/<idea>/` that links to one of your note attachments (a plan, the
overview, a phase file, a Pathfinder question or the map), it first copies the
attachment into `specs/<idea>/attachments/` with `console.mjs keep` and links
the copy, so the spec still shows it after the session is swept. The copy is
named `<slug of your file name>-<8 characters of the upload id>.<ext>`, so
keeping the same upload twice reuses it. Attachments nothing cites stay in the
session, and quick-question attachments are never kept. `keep` is run by the
agent, in its own shell; the server itself still never writes into your
project.
### Ask a quick question
The **Ask** tab, always the last tab, on the dashboard and in every skill,
opens a **Quick question** chat. Ask anything about the project's files, or
anything at all, without waiting for a card. The agent running the skill
answers at its next check-in (in a build, after the task it is on), and the
workflow carries on meanwhile. Asking never holds up a card send, and a card
send never holds up a question.
- **Context only when you add it.** A question starts with no context, so it
reads as a general question. **+ Add context**, beside *+ Attach file*,
lists the last 3 places you visited (question cards, document sections, a
spec picked on the dashboard) plus *This spec*; pick one and an *About: …*
chip sits above the box. × removes it, and it clears after the question is
sent.
- **Starter questions.** An empty conversation offers a few questions to start
from, chosen for the skill you are in and ordered by your role; the ones
that need context appear once you have added some. Clicking one fills the
box without sending it.
- **Pictures.** The same as notes: the button, paste, or drag and drop, up to 5
per question.
- **Ten questions per conversation.** The `N of 10` counter is always in view.
After the tenth is answered, **New conversation** starts a clean one. So does
switching skills through the dashboard.
- **Edits need your approval.** Ask for a change to a file and the agent
replies with a summary and the list of files it would touch, with
**Approve** and **Decline**. Nothing changes until you approve. Afterwards
it lists what changed, with line counts, and nothing is committed unless you
spell out a commit in your message. It never edits the running skill's own
files under `specs/`, and a chat message never answers a card or moves the workflow on: that is what the cards are for.
- **When you cannot send.** Two different lines explain it: "That's 10
questions in this conversation" means start a new one, and "No agent is
connected right now" means the session has finished or the agent has gone
quiet, so resume it to ask. Either way the conversation above stays
readable.
A reply that lands while you are on another tab puts a dot on **Ask** and
chimes once (if sounds are on).
### When an update is out
When a newer Plan2Code release is out, the dashboard shows an **UPDATE
AVAILABLE** banner at the top. Click it for the details: the version you
have, the new one (linked to its Releases page) and the exact command that
updates you, with a Copy button. **×** hides the banner for this session; the
next dashboard shows it again until you update.
The check uses your own git login to read the repo's release tags, once an
hour at most. If git cannot reach the repo (no git, no access, offline), you
simply see no banner, and nothing waits on it. The version in Help always
links to the Releases page.
### Keyboard
| Key | Does |
| --- | --- |
| `Ctrl` + `Enter` (`⌘` + `Enter` on a Mac) | Send |
| `Tab` / `Space` | Move between options and pick one |
| In the Ask box: `Enter` / `Ctrl` + `Enter` (`⌘` + `Enter`) | `Enter` is a new line; `Ctrl` + `Enter` sends the quick question, never the cards |
| `Esc` | Closes whichever dialog is open; on the dashboard, backs out of a pick |
| In Help: arrow keys, `Home`, `End` | Move between topics (the arrows wrap). `Ctrl` + `Enter` does nothing inside Help |
Every control is a real button, radio or checkbox, so screen readers and
keyboard-only navigation work.
## Models in User Preferences
The `plan2code` launcher's model menu comes from `models.json`, which every install overwrites with the authors' curated list. **User Preferences → Models** (Claude and Devin tabs) lets you add a model that came out before Plan2Code was updated; additions are kept in `~/.plan2code/models.json`, and **Reset to default** removes them. Restart `plan2code` to see a change. You can always run `plan2code --model <id>`, or switch model inside your agent before running a skill. Devin ids ending `-xhigh` or `-max` are refused. Maintainers refresh the curated list with the repo-local `/plan2code-model-update` skill, which shows a diff and changes only what they approve.
---
## Troubleshooting
| What happened | What to do |
| --- | --- |
| You closed the tab | Open the link again. Everything is where you left it. After about two minutes with no tab open the link stops working: ask the agent to resume the session for a new one. |
| The link stopped working | Tell the agent. It restarts the session and gives you a new link. |
| Your agent session died while you were mid-answer | Press Send anyway. Your answers are written to disk and are never discarded, including when the session is restarted. The next session picks them up. |
| The page says Plan2Code is not running | It has not checked in for a few minutes. Look at the terminal: it may have finished its turn, hit an error, or be waiting on you there. If it is idle, type `continue` in the terminal and it picks your answers up. Answering here still works, and what you send is held for it. |
| You are not sure anything is connected | The **?** at the top right opens Help. Its *What this page is* tab says what keeps the page alive and whether both halves are talking at that moment, and *When something looks stuck* covers what to check. |
| You would rather use the terminal after all | Say so. Nothing is lost. |
---
## How it works
Nothing is installed. The skills ship a small zero-dependency Node server
(`references/web-console/`), and the workflow step starts it when you say yes.
```
agent server (detached) your browser
| | |
|-- open (binds the port, starts ->| |
| the server AND the browser) ---|---------------------------->|
|<-- http://127.0.0.1:PORT/s/TOK/ |<---- the page, first state -|
| | already inside it |
|-- post the first questions ----->|---- pushed to the page ---->|
|-- wait --seconds 240 ----------->| |
| (returns every 4 minutes with | |
| progress, then waits again) |<---- you press Send --------|
|<-- your answers -----------------| |
```
The server runs detached rather than as a background task of the agent, because
agent background tasks get killed at various timeouts, and that would take your
half-finished answers with them.
**It is built to show you something fast.** A browser takes most of a second to
turn up after it is asked, so `open` binds the page's port itself and sends the
browser there before the server has even finished starting; the server takes
the port over a moment later, and the browser's request waits for it rather
than being refused. The page arrives with the session already inside it, so it
draws the real screen without asking for anything first, in your own theme and
colors from the first frame. On a Windows laptop the page is in front of you in
roughly the time the browser itself takes to open.
**It stays up while the page is open.** The page pings it every few seconds, so
the idle clock never runs down while you are there. Close the tab and it shuts
itself down about two minutes later; it stops after four hours regardless, and
the agent stops it outright when the session ends.
That also means **the server routinely outlives the agent**. It is the agent, not
the server, that reads your answers, and the two fail separately. The page tells
the two apart: the **?** button at the top right opens Help, whose *What this
page is* tab says which halves are talking right now (and *When something
looks stuck* covers what to check), and Planny stops looking busy once
nothing is listening rather than spinning at you indefinitely.
**State lives in three places**, and none of them is your repository (the one
exception being the note attachments the agent keeps for a spec, above):
| Where | What | Survives |
| --- | --- | --- |
| `~/.plan2code/console/sessions/<id>/` | the questions, your draft, your answers | 30 days after the session was last touched |
| `~/.plan2code/console/sessions/<id>/workspace.json`, `ledger.ndjson`, `workspace-cursor.json` | the workspace's folders (paths, names, descriptions — never their contents), the session meter's events, and how far the agent has been told of workspace changes. Kept beside the session, so they outlive every dashboard hop | with the session |
| `~/.plan2code/console/sessions/<id>/uploads/` | images and documents attached to notes and quick questions | 30 days after the session was last touched |
| `~/.plan2code/console/looks.json` | your role, theme, colors, sounds and card width | always |
| your system temp directory | the port, process id and session token | until reboot |
Each of the workspace and meter files has exactly one writer: the server owns
`workspace.json` (the page changes it only through the `/workspace` routes
under Security), and `console.mjs` owns `ledger.ndjson` and
`workspace-cursor.json`. The meter's count is never stored; it is folded from
the ledger every time (`public/meter.js`), so a repeated `post` never counts
twice. The logic lives in small pure modules the tests import directly —
`public/workspace.js` (names, the footer label, the change lines the agent
reads), `public/meter.js` and `public/mentions.js` (the `@name` popover) — and
**Browse…** runs the per-OS picker in `picker.mjs`, copied from the
`plan2code` launcher and kept in step with it by a test.
**Old sessions clean themselves up.** Each time a session opens, whole session
folders nobody has touched for 30 days are deleted — questions, drafts,
answers, uploads and all, paused or not — along with stray scratch files
(`.json` / `.mjs`) left at the top of `~/.plan2code/console/` for as long.
Your `looks.json` and the `console-dir` pointer are never touched, and neither
is the session being opened. Nothing of value goes with them: the page is
never the record, `specs/` in your project is.
**Or clean up on demand.** *User Preferences → Cleanup* → **Find files to
clean up** lists, with sizes, everything in `~/.plan2code/` that is not
needed: sessions untouched for 30 days (never the one you are in), anything at
the top of `~/.plan2code/console/` other than `sessions/`, `looks.json`,
`workspaces.json`, `console-dir` and `update-check.json`, and anything at the top of `~/.plan2code/` other than
`console/`, `bin/`, `launcher.json` and `models.json`. A stray is listed once
it has sat untouched for an hour, so a lock or temp file another session is
writing right now is never caught. **Delete these** removes
that list and nothing else. A console running from another home
(`PLAN2CODE_CONSOLE_HOME`) only ever looks inside that home.
---
## Security
It is a local dev server that reads files in your project (never writing to it)
and keeps its own state in your home folder, so it is built to be boring about
this:
- Bound to `127.0.0.1` only, never `0.0.0.0`. Not reachable from your network.
- A `Host` header allow-list, which is what stops a website you happen to be
visiting from talking to it through a DNS rebinding trick.
- A 128-bit token in the link, swapped for an `HttpOnly; SameSite=Strict` cookie
on first load so it leaves your address bar, your history and any screenshot.
- A strict Content Security Policy with no inline script allowed.
- Markdown written by the agent is turned into page elements directly and never
into an HTML string, so a document that quotes a hostile README cannot run
anything. It shows up as text, which is what it is.
- Uploads are named by the server (`<uuid>.jpg`, or `<uuid>.<ext>` with the
extension taken from the allow-list, via `crypto.randomUUID()`), never by the
file name you sent, and live only under the session's own `uploads/`. The
server checks the bytes: a JPEG header for images (4 MB cap), and for
documents a `%PDF-` header or strict UTF-8 with no NUL byte (10 MB cap).
Documents are served back only under a fixed type (`application/pdf`, or
`text/plain` for text and code) with `nosniff` (and `sandbox` on text), so
the Sent box can link them. A send may only point at
uploads inside that session's `uploads/`, images in `images` and documents
in `files`.
### Routes
The upload routes, the send, the overview read and the workspace. Like every page route, they are on `127.0.0.1`
behind the session cookie, and the ones that write also check the request's
origin.
| Route | Does |
| --- | --- |
| `POST /upload` | Raw bytes, original name in `x-p2c-name`. `Content-Type: image/jpeg`: writes `uploads/<uuid>.jpg` and answers `{ ok, id, kind: "image", path, url, name, size }`; not a JPEG → 415, over 4 MB → 413. `Content-Type: application/octet-stream`: a document, its extension from the name's allow-listed extension; writes `uploads/<uuid>.<ext>` and answers `{ ok, id, kind: "file", path, name, size }` with no `url`; not an allowed extension or contents that do not match → 415 `type`, over 10 MB → 413. Any other content type → 415. |
| `GET /uploads/<id>.<ext>` | Serves an upload: an image as `image/jpeg` (the thumbnail after a reload), a document as `application/pdf` or `text/plain; charset=utf-8` with `nosniff` and, on text, `Content-Security-Policy: sandbox`. Only `<uuid>.<allowed ext>` names under `uploads/`; anything else is 404. |
| `DELETE /uploads/<id>.<ext>` | Deletes an image or document removed before sending. 204, or 404 if it is not there. |
| `POST /submit` | The Send. A note may carry `images` and `files`, each `[{ path, name }]`: at most 5 together, images in `images` and documents in `files`, every path inside this session's `uploads/`, otherwise 400. |
| `POST /chat` | A Quick question `{ text, about?, images?, files?, conversation }`, or `{ reset: true, conversation }` for New conversation. Appended to the session's `chat.ndjson`. 400 `too-long` (over 4,000 characters), `image`, `file` or `bad`; 409 `offline`, `stale` or `limit` (10 per conversation). Never refused because a card send is still uncollected. |
| `POST /chat/decision` | Approve or Decline on an edit the agent proposed: `{ re, decision, text?, conversation }`. Once per proposal; does not count toward the 10. |
| `GET /overview?spec=specs/<name>` | The Overview tab's read of `<spec>/overview.md`, as `text/markdown` with `cache-control: no-store`. On the dashboard `spec` must be one of the scanned specs; elsewhere the session's own spec is used. The file's real path must stay inside the project, so `../`, absolute paths and symlinks pointing out all get 404. |
| `GET /workspace` | The Workspace dialog's list: folders, `missing`, `remote` and the agent's folder `issues`. `?check=1` re-checks every folder on disk first. |
| `POST /workspace/add` | `{ path, name?, description? }`. The path is resolved against the folder the session started in (`~` expanded) and must be a folder: 400 `not-a-folder`, `duplicate` (with its `name`; compared case-insensitively on Windows and macOS), `bad-name`, or `name-taken` with a `suggest`. Answers the new entry, `nested: true` when it sits inside another folder in the list. |
| `POST /workspace/edit` | `{ id, name?, description? }`. Renames or re-describes one folder: 400 `bad-name` / `name-taken`, 404 `unknown`. |
| `POST /workspace/remove` | `{ id }`. 400 `original` for the folder the session started in, 404 `unknown`. |
| `POST /workspace/browse` | Opens the system folder picker and answers `{ path }` or `{ cancelled: true }`; never adds anything itself. 409 `busy` while one is open, 404 `no-picker` when the session looks remote or no picker exists. |
## For maintainers
| | |
| --- | --- |
| Source | `src/web-console/` |
| Ships as | `skills/<every skill>/references/web-console/` |
| Wiring | one `additionalReferences` entry per skill in `install.js`; the dashboard's menu lives in `public/answers.js` (`SKILL_CATALOG`) |
| Agent contract | `src/web-console/console.md`, plus `building.md` for the build steps |
| Tests | `node --test scripts/test-web-console.mjs`, also run by `npm test` |
| Start-up benchmark | `node scripts/bench-web-console.mjs` (needs `puppeteer-core` and a local Chrome or Edge for the paint columns; `--browser-delay 900` models a real browser launch, `--roundtrip` times Send→agent and post→page) |
| Screenshots | `node scripts/capture-screenshots.mjs` (needs `puppeteer-core` and a local Chrome or Edge) |
| Dependencies | none at runtime. `marked` 15.0.7 (MIT) is vendored for its lexer only |
Edit `src/` only. `skills/` is a committed build artifact: regenerate it with
`npm run build:skills` and commit it with the source change, since `npm test`
(via `--verify-skills`) fails when it drifts from `src/`.
### Driving it by hand
You do not need an agent to poke at the console. Point `PLAN2CODE_CONSOLE_HOME`
somewhere disposable first, so you never touch a real session:
```bash
export PLAN2CODE_CONSOLE_HOME=/tmp/p2c-play # Windows: $env:PLAN2CODE_CONSOLE_HOME
cat > /tmp/payload.json <<'JSON'
{ "workflow": "pathfinder", "title": "Playing about",
"items": [ { "id": "q1", "kind": "choice", "title": "Export format",
"body": "Pick one.",
"options": [ { "k": "A", "text": "CSV", "recommended": true },
{ "k": "B", "text": "Excel" } ],
"token": { "A": "a", "B": "b" } } ],
"agent": { "status": "waiting" } }
JSON
node src/web-console/console.mjs open --file /tmp/payload.json # prints the URL, opens the browser
node src/web-console/console.mjs wait --session <sid> --seconds 60
node src/web-console/console.mjs stop --all
```
`node src/web-console/console.mjs help` prints the same command summary an
agent gets when it finds the binary before the docs.
`wait` exits `0` with the answers, `10` if nobody has finished yet, `20` if the
server died, `30` on Cancel. Pick `--seconds` to sit under whatever your harness
allows a single command: it is a poller, so killing it costs nothing, and a
collected result is printed before it is marked collected, which means an
interrupted handover can only repeat itself, never lose an answer. `status`
lists sessions and flags any whose answers were submitted but never collected.
When something misbehaves, the session directory is the whole story:
`state.json` (what the agent published), `draft.json` (what the person has typed
but not sent), `result.json` (what they sent), `events.ndjson` (an append-only
log you can `cat`), and `server.log`.
Set `--idle-ms 60000` on `open` if you want the server to shut itself down
quickly while experimenting.
Two environment variables help when the start is what you are looking at:
| Variable | Does |
| --- | --- |
| `PLAN2CODE_CONSOLE_TRACE=1` | `open` prints a timeline of its steps to stderr (roots found, server spawned, port bound, browser sent, state written, server listening). |
| `PLAN2CODE_BROWSER` | The command that opens the link instead of the system default, run through the shell with the link appended, e.g. `PLAN2CODE_BROWSER='firefox --new-tab'`. |
Every `console.mjs open` also writes its own directory to
`$PLAN2CODE_CONSOLE_HOME/console-dir` (or `~/.plan2code/console/console-dir`
when the override is unset): it is how an agent finds the console on a later
session without searching the skill install dirs.
The vendored file is `src/web-console/public/vendor/marked.esm.js`,
sha256 `7a7d9a521ac9384e0c3a075120a7c486cbd0c3c32cc5601bbb79a23e97403690`.
Only `marked.lexer()` is used; the rendering is ours, in `public/render.js`, and
builds DOM nodes rather than HTML so nothing needs sanitising.
### Refreshing the screenshots
The docs and the landing page show the console through the images in
`docs/screenshots/`, one light and one dark WebP per screen. Rerun the capture
after any change a person would see on the page: a new layout, a renamed
button, a different colour.
```bash
npm i --no-save puppeteer-core # once; never added to package.json
node scripts/capture-screenshots.mjs # every screen
node scripts/capture-screenshots.mjs --only dashboard # one screen
```
Each screen is staged with no agent, from the payloads in
`scripts/screenshots/` (its README names the file behind each screen), in a
throwaway sample project and console home that the script removes when it is
done. It prints each file's size and warns over 300 KB.
Open every image before you commit it: the right screen, both themes, nothing
cut off, and no username, home path or real project name anywhere on it.
+42 -9
View File
@@ -1,6 +1,6 @@
# AGENTS.md
This file provides guidance to AI coding agents like Claude Code (claude.ai/code), Cursor AI, Codex, GitHub Copilot, Devin, Zed, and other AI coding assistants when working with code in this repository.
This file provides guidance to AI coding agents working with code in this repository.
## Project Overview
@@ -12,7 +12,7 @@ Plan2Code is a structured 4-step workflow methodology for AI-assisted software d
## How to Use This File
This file is an index — each section below contains a brief summary and a link to a detail file in `.agents-docs/`. Read only the sections relevant to your current task. Full details (commands, tables, file lists) are in the linked files. The sections "Git Commit Messages" and "Project Overview" are fully inline here.
This file is an index — each section below contains a brief summary and a link to a detail file in `.agents-docs/`. Read only the sections relevant to your current task. Full details (commands, tables, file lists) are in the linked files. The sections "Project Overview", "How to Use This File", "Mascot", and "Keeping this file current" / "Failure log" are fully inline here and are never split into `.agents-docs/`.
## Architecture
@@ -34,13 +34,25 @@ Details: [Plan2Code Metrics](./.agents-docs/AGENTS-plan2code-metrics.md)
## Plan2Code Status Line (Claude CLI)
Optional CLI status bar (`src/statusline-claude/`) for Claude Code. Displays model, project, branch, context usage, usage stats (rate limits or token counts), git diff stats, and duration. Reads data directly from Claude Code's stdin JSON — no API calls, no auth, no background processes. Included in `Install All + dev tools` (`A`); also available via `install.js` Custom → S (opt-in).
Optional CLI status bar (`src/statusline-claude/`) for Claude Code. Displays model (+ effort), project/worktree, branch, git diff stats, duration + session cost, context usage, and usage stats (rate limits or token counts). Reads data directly from Claude Code's stdin JSON (plus timeout-bounded git calls): no API calls, no auth, no background processes.
Details: [Architecture](./.agents-docs/AGENTS-architecture.md) (see Status Line section)
## Web Console
Optional local browser UI (`src/web-console/`, zero dependencies, detached `node` server on `127.0.0.1`) bundled into every skill via `additionalReferences`. The bare `plan2code` skill opens its dashboard, a card menu that launches any skill in the same session. Pages cover questions, sign-off gates (with an in-page code review), build progress, the living document, an **Ask** chat, a read-only **Workspace** of extra folders (remembered per folder and spec in `~/.plan2code/console/workspaces.json`), a session meter, a built-in 16-topic Help, a first-run Welcome, User Preferences (gear) and a bug/idea link to GitHub issues. The agent-facing contract is `console.md` / `building.md`; the server never writes into a project.
Details: [Architecture](./.agents-docs/AGENTS-architecture.md) (see Web Console section) and [.readme/web-console.md](./.readme/web-console.md)
## Skill Scripts
Deterministic workflow steps run as Node scripts shipped in each skill's `scripts/` (shared ones in `src/skill-scripts/`, a skill's own in `src/<skill>-scripts/`, listed per skill in `SOURCE_PROMPTS` → `scripts`). Prose keeps the judgment. JSON on stdout, documented exit codes, tests run against the built `skills/` tree.
Details: [Architecture](./.agents-docs/AGENTS-architecture.md) (see Skill Scripts section)
## Development Commands
Build commands, installer menu options, how the installer generates platform-specific files, platform format table, and how to edit workflow prompts.
Build commands, installer menu options, non-interactive skill-build verification, skills CLI delegation, skill format, and how to edit workflow prompts.
Details: [Development Commands](./.agents-docs/AGENTS-development-commands.md)
@@ -52,11 +64,32 @@ Details: [Code Style & Gotchas](./.agents-docs/AGENTS-code-style.md)
## Mascot
The project has a mascot called "Planny" — an ASCII art robot that appears in installer output and workflow prompts. Mascot variants are defined in the `MASCOT` constant in `install.js` and appear in workflow markdown files.
The project has a mascot called "Planny": a box with a star for an eye, stick arms and two small feet. He appears in installer output, workflow prompts, the loop, the Claude Code status line (box, star and feet, no arms) and the web console. ASCII variants are defined in the `MASCOT` constant in `install.js` (and `plan2code-loop/src/utils/logger.ts`) and appear in workflow markdown files: arms down at rest, one arm waving when he is asking something, both arms up when he is celebrating. The eye is always the star.
The web console draws him as inline SVG in `src/web-console/public/index.html`, where he doubles as the page's state indicator: his star lights in the highlight colour and he waves when something is yours to do, the star turns and his feet walk while the agent works, it goes green when all is done, he shows no arms at all when the agent is not answering, and he fades with his star shut when the connection drops. The parts are classed (`shell`, `eye`, `smile`, `foot`, and one `arms a-<mood>` set per mood) and `app.js` switches mood with a single class, so the drawing stays in the markup and the styling in `app.css`. The nine skill start poses, the dashboard wake-up, `favicon.svg` / `favicon.js` and `src/launcher/plan2code.ico` all use the same design.
```
╭───╮
│ ● │
│ ◡ │
╰───╯
╭───╮
│ ★ │╱
╱│ ◡ │
╰┬─┬╯
```
## Keeping this file current
The `Failure log` section below is a recording of mistakes made by previous AI Agents while working with this code base.
When you make a mistake, get corrected, or discover something about this codebase that wasn't written down:
1. Add one line to the `Failure log` below, in the imperative, describing the correct behaviour.
2. Keep it specific to this repo. General advice belongs nowhere.
3. If this fix is a workflow rather than a rule, put it in `.claude/skills/` and link it from here.
4. Include the change in the same commit and mention it in your summary.
## Failure log
- Do not audit `CHANGELOG.md` headings through PowerShell — the emoji come back as `?`. See the gotcha for the correct approach.
- Sync upstream only through `/sync-repo`, and verify its baseline against the real tree before porting: an earlier sync silently skipped a whole upstream release (the `3-implement-review` skill) while the recorded version froze. Update the baseline with `.claude/skills/sync-repo/set-synced.mjs`, never with a shell pipe over the decrypted body.
- After a sync merge, run `npm test` before trusting a "clean" file: when upstream and plan2code fixed the same bug, `git merge-file` keeps both copies without a conflict (a sync once left `let pattern` declared twice in `app.js`, caught only by `parse-check`).
- When extracting upstream files in Git Bash, run `git show <ref>:<path>` with `MSYS_NO_PATHCONV=1` and `C:/` style paths (never `/c/`): path conversion rewrites `origin/main:.readme/x.md` into `origin\main;.readme\x.md`, the show fails, and the merge silently runs against an empty "theirs". Check every extracted file is non-empty before merging.
- After a sync, grep the result for the upstream owner slug and company name in any case (the `/sync-repo` body lists them): the rebrand map only rewrites URLs, so a bare owner slug once slipped through into `feedback-payload.mjs` (and its test) until it was fixed by hand to `jparkerweb/plan2code`. No upstream name may appear in plain text anywhere in this repo.
+195
View File
@@ -2,6 +2,201 @@
All notable changes to Plan2Code will be documented in this file.
## v2.4.2
### ✨ Added
- **web-console**: the Questions and Ask tabs show a small spinner left of their name while Plan2Code is on it. Questions (Dashboard on the dashboard) spins from Send until the answers are picked up, the same wait as *Sent. Waiting for Plan2Code.*; Ask spins while a question the agent picked up has no answer yet, exactly when the conversation's own ring does. Neither turns once the agent goes quiet or offline, so a dead session never reads as a slow one.
- **web-console**: **User Preferences → Cleanup**, above Models. **Find files to clean up** lists, with sizes, what `~/.plan2code` does not need: sessions untouched for 30 days (never the one serving the page), anything at the top of `~/.plan2code/console/` but `sessions/`, `looks.json`, `workspaces.json`, `console-dir` and `update-check.json`, and anything at the top of `~/.plan2code/` but `console/`, `bin/`, `launcher.json` and `models.json`. A stray is listed only after an hour untouched (`STRAY_GRACE_MS`), so a held lock or a temp file mid-rename never is. **Delete these** removes exactly that list and reports what went and the space freed. The server builds the list (`POST /cleanup`, `cleanupFindings()` in `lib.mjs`), sends no paths to the page, and on delete re-scans and removes only ids still found (`removeCleanup()`), unlinking symlinks without following them. A console running from `PLAN2CODE_CONSOLE_HOME` never looks outside that home. Help's User Preferences topic describes it.
- **web-console**: the dashboard shows an **UPDATE AVAILABLE · v<latest>** banner when a newer release is out. The console server finds the newest `vX.Y.Z` tag with `git ls-remote --tags` when it starts (no shell, prompts off, a 10 s limit, never waited on), caches the answer for an hour in `~/.plan2code/console/update-check.json`, and adds `update` to the state frame only when that release is newer. The banner opens a dialog with both versions, a Releases link and the update command with Copy; × hides it for that session. `GET /version` also returns `latest`, and the version in Help links to the Releases page.
- **launcher**: `plan2code` opens with a colored Planny banner and a plain-English intro, then the agent menu with cyan numbers, a `(last used)` marker on the remembered pick and a green default. With one CLI installed, or a `--cli` pick, it prints `Opening with <CLI>…` instead. The model menu, the folder prompt and every `plan2code:` error get the same styling. Color is used only on a terminal with `NO_COLOR` unset, so piped runs print plain text and no banner.
- **maintainer**: a repo-local `/plan2code-model-update` skill (`.claude/skills/`, never installed) refreshes `src/launcher/models.json` from `devin models list` and Claude Code's documented model aliases. It shows a diff, applies only the changes the maintainer approves, checks the file against the launcher's rules, runs `npm test` and never commits.
- **tests**: `scripts/test-web-console.mjs` covers Cleanup, the update check and the note thread; `scripts/test-launcher.mjs` covers the banner and color handling.
### 🔧 Changed
- **document**: `overview.md` puts the Phase Checklist and Parallel Execution Groups directly under the Summary, ahead of the Tech Stack, Architecture, Risks and Success Criteria, so the phases are found without scrolling. Every script finds these sections by heading, so specs written in the old order keep working.
- **web-console**: while a build runs, the waiting screen's pointer to the tasks list is now clickable. **Task** and the tab's own name (say **Phase 6 tasks**) show bold in the highlight color, and either one opens that phase's tasks tab.
- **web-console**: the session meter counts a build while it runs instead of only once the phase is approved. `console.mjs` writes a `progress` ledger entry whenever an Implement or Implement + Review post raises `headline.cleared`, and the build's row (*Implement + Review: 9 tasks built so far*) scores one point per 3 tasks done. The phase's final `run` takes that row over in place, so nothing counts twice and the meter never drops. A second phase in the same run starts its own count once its headline is back at `cleared: 0`.
- **web-console**: the **UPDATE AVAILABLE** banner and **Send to Plan2Code** (while enabled) carry a soft glow in the picked highlight color; with reduced motion Send's pulse stops and the glow stays.
- **web-console**: a request body over the size cap now gets `413 too large` from every JSON route instead of a connection reset or a `400`.
### 🐛 Fixed
- **web-console**: Implement + Review's built-in review no longer scores twice on the session meter. Its `review` run now scores the phase's 1-point review bonus as the review starts (not a standalone Review's 2), and the phase's final `run` leaves that point out. A 17-task phase now ends at 7 points, not 9. The breakdown also stops repeating the unit in "Code review run: review".
- **web-console**: Send no longer stays disabled when an agent posts a question mid-build while still marked `working` (seen on Implement + Review, where the question said "I'm carrying on while you decide"). Once an open question has an answer or a note staged, Send unlocks at once with "Still working · it reads this next", instead of waiting out the agent's `quietMinutes` (up to an hour); the server holds the send until the agent next looks. `building.md` now says a question stops the build: set `waiting` in the same post as the question and wait, never keep working with a question open.
- **web-console**: **Notes on this** showed the agent's replies as empty "Plan2Code" bubbles and dropped the person's own notes once they were picked up. The server now writes every note into the item's `thread` as it arrives, `post` lifts `from` / `role` and `md` / `body` / `content` into `{ who, text }` (and drops an echoed copy of the person's note), and an entry with no text is rejected with exit 3. Older `{ from, md }` entries still render. `console.md` gains a **Replying to a note** section.
- **web-console**: dropped the `comments` list on docs, which `post` appended and time-stamped but nothing ever displayed.
- **web-console**: a dashboard launch could strand the hand-off when the agent called a skill-invocation tool on the picked skill instead of reading its `SKILL.md` (every skill below the dashboard ships `disable-model-invocation: true`). `plan2code.md` and `console.md` now say to open the next skill's file with a file-read tool.
- **web-console**: the waiting pane pointed Document and other skills with no task list at a tasks tab that never exists there. The pointer now appears only when a build's `phase` or a quick task's `tasks` tab is posted. The nav tab spinner also gets its 7px gap back from the tab's name.
- **web-console**: skipping a question dropped a note already typed for it; the note now goes with the skip as `skip; <note>`. Ticking a checklist step after a skip now ends the skip instead of still sending `skip`.
- **web-console**: reconnecting cleared every banner, including a warning that arrived while offline; it now clears only the *Lost contact* banner. An open Overview tab now refreshes during a build.
- **web-console**: a `409` on Home, Stop or **Write the brief** now marks the page busy the way Send does, and the brief dialog says why it could not go.
- **web-console**: arrowing through a menu question kept dropping keyboard focus to the page body; menu radios now carry stable ids, a `value` and an `aria-label`. `@` mentions typed with a capital letter now match, and the attachment-limit banner uses `MAX_ATTACHMENTS` / `CHAT_MAX_ATTACHMENTS` instead of a hard-coded 5.
### 📚 Documentation
- **web console guide**: new *When an update is out* section, Cleanup under the session cleanup notes, and a pointer to `/plan2code-model-update` for the curated model list.
## v2.4.1
### ✨ Added
- **Screenshots of the web console**: seven screens (dashboard, question, living document, build, sign-off, finish, Ask tab) in light and dark, as WebP in `docs/screenshots/`. Readers see the theme that matches their system, on GitHub and on the site.
- **`scripts/capture-screenshots.mjs`**: stages each screen from the JSON payloads in `scripts/screenshots/`, in a throwaway sample project and console home, and captures it with puppeteer-core and a local Chrome or Edge. One command refreshes every image; `--only <shot>` refreshes one. `puppeteer-core` is installed with `--no-save` and never added to `package.json`.
- **New landing page**: `docs/index.html` rebuilt in the console's look, with the console above the fold.
- **Web console welcome**: the first dashboard a person sees opens a Welcome dialog with Planny waving, four tips and a **Start building** button. Closing it by any route saves `looks.welcomeSeen` in `looks.json`; Help's first tab has **Show the welcome again**.
- **Report a bug or idea from the console**: a megaphone button at the right of the top bar opens GitHub's new-issue chooser for `jparkerweb/plan2code`, with new **Bug** and **Idea** issue templates in `.github/ISSUE_TEMPLATE/` (blank issues off).
- **Version in Help**: the Help dialog's footer shows the running Plan2Code version, served by a new `GET /version` route. The skill build copies `version.json` into each skill's `references/web-console/` so the console can find it.
- **skills**: skill scripts. Every step a workflow decides with one correct answer now runs as a script the agent calls instead of prose it follows: Node 18+ built-ins only, one JSON object on stdout, documented exit codes, `--help`, `--dry-run` on writes. They install beside each `SKILL.md` in `scripts/`: shared ones from `src/skill-scripts/` (named per skill in `SOURCE_PROMPTS` → `scripts`), a skill's own from `src/<skill>-scripts/`, plus `common.mjs` and the release's `version.json`.
- **specs.mjs** (plan, revise-plan, document, implement, implement-review, review, quick-task, finalize, handoff): `list`, `status`, `metrics`, `mark`, `archive`. It reads gitignored `specs/` directly, parses the Phase Checklist and Parallel Execution Groups in the formats real specs use, counts tasks by marker (including `3.6a` and named `1.T` tasks), applies Implementation Mode's phase-selection table, and flags drift such as a phase marked `[x]` with open tasks. It shares the dashboard's spec classifier by importing the bundled `web-console/lib.mjs`.
- **pathfinder.mjs**: `reconcile`, `frontier`, `claim`, `resolve`, `rule-out`, `trail`, `gate`, `lint`, `brief-data` for maps on the `local` backend. The map's markers, the Trail Footer, the Clearing Gate's mechanical checks and hard caps, the standing-rule lint (loop tokens, checkboxes, reserved names, schema lines, `Blocked by:` cycles, the metrics collector's own scrape patterns) and a brief's date filtering are computed from the question files. Reconcile keeps each map row's own wording and only changes what the files contradict. Maps on the `github` backend keep doing these steps with `gh`, as `github-issues.md` describes.
- **feedback-payload.mjs** (finalize STEP 6.5): builds the payload from the archive, with real 12-character sha256 prompt hashes (the prose admitted the agent could not compute them), renders the issue, checks the 8 KB URL limit with `--truncate`, and runs the gh / browser / print tiers against `jparkerweb/plan2code`.
- **commit-msg.mjs**, **review-scope.mjs**, **agents-md.mjs**, **agent-files.mjs**, **handoff-path.mjs**: the commit command from the branch's ticket, Review's git scope and doc share, AGENTS.md's structure and orphans, the AI agent file sync, and the handoff's timestamped path and git ignore check.
- **tests**: `scripts/test-skill-scripts.mjs` and `scripts/test-script-<skill>.mjs`, run by `npm test` against the built `skills/` tree. One test fails if a `SKILL.md` names a script its skill does not ship.
- **web console**: the Workspace dialog's folders are remembered across console sessions, in `~/.plan2code/console/workspaces.json` beside `looks.json` (never in the project), keyed by the folder a session runs in and then by spec. Every change is saved. *Start from scratch* on the dashboard shows the project's list; picking a spec loads its own, which starts as a copy of the project's and is separate after that, and a spec created mid-run takes the list in force with it. The dialog says whose list it is. New `console.mjs forget --spec specs/<name>`, which Finalize's Spec Cleanup runs after archiving a spec.
### 🔧 Changed
- **skills**: the character limit for `src/plan2code-*.md` rises from 11,500 to 20,000 (about 5k tokens). It no longer guards Windsurf's old limit; it is a size discipline, applied by `validate-char-count.js` and the metrics improver alike.
- **finalize**: a run at exactly 75% complete now finalizes with notice (the table said >75% and <75%). Assumed `[?]` tasks count as completed and are listed; started `[/]` tasks count as incomplete. `verification_items_added` and Step 7's two judgment counts are passed to the script, which computes the rest.
- **finalize**: the community feedback issue title reads `[Feedback] v<version> - rating N/10`; ingestion matches on the `[Feedback]` prefix and is unaffected.
- **init**, **init-update**: both write the same pointer templates, from `agent-files.mjs`. Replacing `.cursor/rules/` now also removes Cursor's own `.mdc` rule files, so AGENTS.md really is the single source there, and files that already point at AGENTS.md are no longer offered again.
- **revise-plan**: re-opening a phase (`specs.mjs mark --to open`) also sets its phase file's Status back to In Progress.
- **review**: the branch scope compares `base...HEAD` (what this branch added since it left the base) rather than `main..HEAD`, and a trunk other than `main` / `master` is passed with `--base`.
- **pathfinder**: the Trail Footer's named legend numbers stops by position (`(blocked:3)` points at stop 3), and a brief's range is given as `this-week`, `since:<date>` or `<date>..<date>`.
- **init**, **init-update**: a CLAUDE.md (or other agent file) that is a symlink or hard link to AGENTS.md is reported as `linkedToAgents` and never replaced, since writing the pointer through it would overwrite AGENTS.md. A pointer file checked out with CRLF line endings still counts as already pointing.
- **review**: with no `main` or `master` and no `--base`, `review-scope.mjs` now stops with `no-base` instead of quietly counting only uncommitted changes.
- **commit-msg.mjs**: the add step is `git add -- <paths>`, paths starting with `@` or `-` are quoted, and subjects with `\`, `!` or curly quotes are refused. The output also carries `add` and `commit` separately for Windows PowerShell 5.1, which has no `&&`.
- **specs.mjs**: `mark` reads a Status line written as `Status:` or `**Status**:` too, and warns when a phase file has none; a path that is not a folder directly under `specs/` is refused (`not-a-spec`).
- **pathfinder**: the map header carries `**Next NN:**`, so a deleted question's number is never handed out again. `claim`, `resolve` and `rule-out` honour `--dry-run`. Notes indented under a map row survive a marker change. Research questions are fired unclaimed; the one-claim rule is for the questions a person works through.
- **finalize**: `feedback-payload.mjs` counts the same phase files `specs.mjs` does, `submit` refuses an oversize URL when `gh` is unavailable (`oversize`), and a bare spec name present in both `specs/` and `specs--completed/` is refused as ambiguous.
- Every script was checked against real specs and Pathfinder maps in other repositories and by four independent reviews; each issue found has a regression test.
- **install.js**: the closing message beside Planny now names the dashboard first: `plan2code` from any project, or `/plan2code` in your agent when the command could not be installed. The docs link moves to the line under it.
- **web-console**: User Preferences opens from a gear button instead of the three swatches, and the default highlight color is now ocean blue instead of rust.
- **web-console**: the dashboard rail marks the suggested next step with a small dot, since the center pane's Suggested pill can scroll out of view. Send pulses only while it is the primary button.
- **web-console**: when the agent stops checking in, the status line and Help now say to type **continue** in an idle terminal so it picks your answers up.
- **web-console**: `wait` exit 10 now carries a `next` instruction (relay one line, then wait again), and `console.md` opens with the one rule that keeps the page alive: every turn ends on a `wait` call, never on text.
- **web-console**: a finished quick task offers **Back to the dashboard** as well as **Review it now**, so the finished screen is never a dead end.
- **Windows**: the folder picker (the launcher's desktop shortcut and the Workspace **Browse…** button) now opens centred and in front of other windows instead of behind the terminal.
- **Launcher**: before asking which CLI to use, `plan2code` says what it is about to do and that running `/plan2code` inside your agent works too.
- **web-console**: the session meter now weighs the work a session actually did instead of a flat amount per step. A built phase scores one point per 3 tasks it completed (rounded up), plus 1 for Implement + Review, so a twelve-task phase weighs four times a three-task one; the agent reports the count as `run.tasks`. Every 2 questions you answer on the page add a point to the skill you answered them in, so a long back-and-forth in Plan or Pathfinder counts for more than a short one. Each Pathfinder research question adds 1 through a new `pathfinder-research` event, since research fills the context with nobody answering anything. The levels move to match: yellow from 5 points, red from 10, and the ring is full at 15.
### 🐛 Fixed
- **pathfinder**: charting no longer ends the session, or shows "Session ended" in the web console, while its research subagents are still running. They live in the terminal's process, so closing it on a "done" page killed them mid-write. The session now waits for every one (the page shows a running count), each subagent ends `## Evidence` with a `**Research complete:**` line, and Work Step 2 re-fires any research question missing it. `console.md` makes the same rule general: no `finish` while any subagent or background job is still out.
- **web-console**: reusing a settled question id for a different kind of question (say a multi-select turned confirm) now opens it as a new question instead of keeping the old answer.
### 📚 Documentation
- **Web console first**: README, QUICK-REFERENCE, the walkthrough and the console guide now open on the browser flow (install, run `plan2code`, pick a card) with screenshots, and keep every slash command in a "Prefer the terminal?" section. The walkthrough follows the task API through the web console, session by session, and the console guide opens with a Quick start and has Troubleshooting ahead of the internals.
- **Rule 1 is now "Keep each session short"**, worded the same in README, QUICK-REFERENCE, the walkthrough and the landing page: chain a couple of small skills on the dashboard, and start a fresh console session when the session meter turns red or before a big planning or build step.
- **Web console guide** opens with a two-minute tour; **QUICK-REFERENCE** opens with a Launch table (shell, inside an agent, straight into a skill with `--web`) and a "Terminal instead" note; the walkthrough mentions the Ask tab during a build and ends with links onward.
## v2.4.0
### ✨ Added
- **Web console**: an optional local browser UI that every skill offers at the start of a session ("web console or terminal?"). Questions, recaps, sign-off gates, build progress and the document taking shape are laid out in a page instead of scrolling past in the terminal, so the workflow is usable by product people and not only engineers. Zero dependencies, no build step: a detached `node server.mjs` bound to `127.0.0.1`, bundled into each skill as `references/web-console/`. Answers carry the literal terminal token they stand for (`approved`, `a`, `skip`), so declining the console or losing it mid-session falls straight back to the terminal with nothing lost. Full guide in `.readme/web-console.md`.
- **`plan2code` dashboard skill**: a new bare-named skill that opens the console as a card menu of every skill, with a spec picker, and launches the pick on the same page. **Back to the dashboard** works from finished screens and, mid-workflow, from a triangle beside **Stop session**.
- **Global `plan2code` command**: installed with the skills (`I` and `A`; alone via Custom → `D`; removed by `U`). Typed in any project it opens the dashboard in Claude Code or Devin, asking which when both are installed and remembering the last pick in `~/.plan2code/launcher.json`. `--cli claude|devin` skips the question and `--model <name>` skips the model notice; other arguments are forwarded. It starts the agent with permission prompts off (`--permission-mode bypassPermissions` in Claude Code, `--permission-mode bypass` in Devin); run `claude /plan2code` yourself if you want prompts. The launcher lives in `~/.plan2code/bin/` with npm-style shims (sh, plus `.cmd` and `.ps1` on Windows).
- **Desktop shortcut** (opt-in, asked after the command installs): `Plan2Code.lnk` on Windows, `Plan2Code.command` on macOS, `plan2code.desktop` on Linux. It opens a native folder picker first, starting on the last folder picked.
- **`plan2code-3-implement-review`**: a quality-gated Step 3 that implements one phase, runs a focused review and the selected fixes before sign-off, and marks the phase complete only after the reviewed result is approved. Step 2's hand-off now offers it first, with `/plan2code-3-implement` as the review-free alternative.
- **`--web` flag**: `/plan2code-1-plan --web plan a lunch voting app` opens the console without asking. Every next-step command the workflow prints (resume footers, NEXT STEP lines, the loop's completion banner) now carries `--web`.
- **Console pages for builds and reviews**: a build shows the current task, a task-count progress bar and the phase's task list ticking as tasks land. After an approved phase or a finished quick task, **Review it now** runs `/plan2code-review` on the page, with *Which to fix* presets for critical-and-warnings or everything.
- **Console conveniences**: nine question shapes with live validation and drag-to-reorder phases; drafts that survive a closed tab, reload, server restart or agent crash; an **Ask** tab for quick questions the agent answers at its next check-in (file edits need approval); image and document attachments; a read-only **Overview** tab; a **Workspace** of extra read-only folders addressed as `@name`; a green / yellow / red **session meter**; a **Role** preference that orders starter templates; nineteen highlight colours, light / dark themes and card widths; audio cues with per-event toggles; and a 16-topic Help dialog.
- **Planny in the console**: drawn in SVG in his own dock above the question list, where he carries the page's state: his star lights and he waves when it is your turn, the star turns and he walks on the spot while the agent works, it goes green when all is caught up, he shows no arms when the agent is not answering, and he fades with his star shut when disconnected. He flies through a starfield while the agent thinks, wakes up on a fresh dashboard, and acts out each skill on its starting screen, arms holding that step's prop. The tab icon adds a busy, ready or attention bar, and the desktop shortcut uses him too.
- **Tests**: `scripts/test-web-console.mjs` and `scripts/test-launcher.mjs` (run by `node --test`), `scripts/parse-check.mjs` (every shipped browser module parses as an ES module) and `scripts/check-skill-table.mjs` (the dashboard's "What each skill reads and writes" table matches the skills), all wired into `npm test`. `scripts/bench-web-console.mjs` measures time to first paint.
- **install.js**: an animated progress indicator during the two long install phases, so the installer no longer looks hung.
- **Curated launcher models**: the `plan2code` launcher's model menu reads `src/launcher/models.json` (copied to `~/.plan2code/bin/` and refreshed by every install) instead of asking the CLI, so Devin offers a short verified list at Low, Medium and High effort. **User Preferences → Models** (Claude and Devin tabs) adds models released after an install, kept in `~/.plan2code/models.json`, with **Reset to default**. `--model <id>` still skips the menu.
- **Quick-task build progress on the console**: an approved quick-task plan opens a **Build progress** tab that ticks each step, raises the progress bar and updates the activity line, as an Implement phase does.
- **Console extras**: sent notes and their attachments stay listed in the Sent box after a reload or in a second tab; the fresh-idea box on Pathfinder, Plan and Quick Task has **+ Attach file** (and takes a paste); clicking the session meter lists what each step added; git code blocks get a **Run it** button; the Ask tab may run one command the person spells out (a commit, a test run) and report the result; the wake-up animation has a **Skip** button.
### 🔧 Changed
- **Planny has a new look everywhere**: a box with a star for an eye, stick arms and two small feet, chosen from twenty studies. Every ASCII Planny in the workflow prompts, the installer, `plan2code-loop` and the Claude Code status line is redrawn to match: the eye is always the star, one arm waves when he is asking something, both go up when he is celebrating, and the status line shows his box, star and feet without arms.
- **Character limit raised from 11,000 to 11,500** for `src/plan2code-*.md` (`scripts/validate-char-count.js`, and the same number in `plan2code-metrics`'s `improver.ts` and `applier.ts`), so the console's Interface lines fit without cutting workflow rules.
- **package.json**: the installer's bin is renamed `plan2code` → `plan2code-install`, leaving the `plan2code` name to the launcher. The `npx` install command is unchanged.
- **install.js**: skills bundle the shared `src/web-console/` directory through `additionalReferences`, copied as bytes so the bundled chimes survive the build. Installed skills are listed by reading the skills store directly instead of `skills list` (seconds instead of over a minute on Windows), and removal is one variadic `skills remove`.
- **plan2code-0-pathfinder**: the duplicated Trail Footer forms fold into the fallback line, since `references/trail.md` defines them in full; the GitHub Issues backend is unchanged.
- **`.gitattributes`**: `*.mp3` is binary. `.editorconfig` pins LF for editors. `prototypes/` is gitignored.
- **Finished screens**: Review, Init, Init Update and Handoff end with **Back to the dashboard**, as Quick Task does; only a pause leaves it off.
- **Console sturdiness**: `wait` reports the server gone only after three missed health checks in a row, so a busy server is not taken for a dead one; a reused question id no longer inherits the old answer; polling stops redrawing an unchanged page and returns to the live stream (with backoff) once the server answers; `stop` gives a busy server a second, longer health check before leaving it alone; a fresh session in a linked worktree records the worktree as its folder; a text question whose `pattern` does not compile is refused at `post`.
- **scripts/validate-char-count.js**: the dashboard prompt `src/plan2code.md` is size-checked too.
## v2.3.0
### ✨ Added
- **`plan2code-0-pathfinder`: `Resolved:` date on question files** — a sixth schema line recording the day each decision landed. Reconcile backfills older maps from `Claimed:` (today if unclaimed), and resolved map rows now carry the date in italics beside the gist. On the GitHub Issues backend the issue's own close date fills this role, so there is nothing to write and nothing to backfill.
- **`plan2code-0-pathfinder`: BRIEF mode** — asking for a `brief`, `recap`, or `minutes` writes a dated, plain-English report to `pathfinder/briefs/brief-<YYYYMMDD>.md`: what was decided in the range, what was ruled out, what is ready to decide next, what is still blocked, and what is next. Zero pathfinder machinery in the file — no numbers, markers, or jargon — so it doubles as meeting minutes for a joint Product and Engineering session. New `references/brief.md`; BRIEF mode never claims or resolves a question. The brief file stays on local disk on both backends.
### 🔧 Changed
- **`plan2code-0-pathfinder`: MODE B cadence is now decide → menu → checkpoint** — questions still resolve strictly one at a time, but a session no longer ends after one. Recording a decision leads to a continue-or-stop fork-menu (new Trail Footer **Form C**, with the resume command always visible as the stop branch), and a fresh session is *recommended* after about three decisions — or after a heavy, contested, or locked one — to protect context quality rather than forced by the rules.
- **`plan2code-0-pathfinder`: every user-facing question uses the structured question UI when available** — intent gate, destination and frontier grills, recap confirmations, sketch reactions, `legwork · HITL` completion, and the new fork-menu. The old detail test that routed gnarly questions to prose Q blocks is gone: trade-offs go in option descriptions, composed answers come back through `Other`, and an artifact is shown in prose immediately before the tool rather than replacing it. Prose Q blocks are now fallback only, for when the tool does not exist.
- **`plan2code-0-pathfinder`: the breadth-first frontier grill gains seven product axes** — user value, success measurement, target user, rollout, analytics, support and docs, and legal/compliance, each with a sample probe. Inapplicable axes are named and skipped up front, and with Product in the room engineering and product batches alternate so neither team spectates. New group-session grilling rules: address the room, record who broke a tie, and never record a contested pick as decided — both positions go under `## Evidence` and the question stays open.
## v2.2.0
### ✨ Added
- **Committed Agent Skills build** — each workflow under `src/` now builds to `skills/<skill-name>/SKILL.md`, with companion references nested under `references/`. `npm run build:skills` regenerates the artifact, and `npm test` verifies that committed skills have not drifted from their source prompts.
- **Project-scoped skill installation** — Custom → `L` now installs Plan2Code directly into the current project through the skills CLI instead of printing manual copy instructions.
### 🔧 Changed
- **Installation now delegates to the skills CLI** — Plan2Code ships one canonical Agent Skill format instead of maintaining separate command, prompt, workflow, and skill outputs for individual tools. The installer builds `skills/`, checks `npx --yes skills`, removes stale Plan2Code skills, and runs `skills add` with an explicit workflow list. Installation now requires Node.js 18+ and network access; installed skills can be updated with `npx skills update -g`.
- **Legacy installation cleanup is automatic** — install and uninstall sweep files written by earlier per-tool installers so old commands cannot shadow the canonical skills. This includes the Gemini CLI files retained exclusively for uninstall compatibility.
- **Default installation is skills-only** — main-menu option `I` no longer installs `plan2code-loop`. The loop remains part of `A` and is still available separately through Custom → `O`.
- **One frontmatter contract serves every agent** — generated `SKILL.md` files always include `name`, `description`, and `disable-model-invocation: true`. Reference files remain nested under their owning skill, eliminating the old flat-file path rewrite and its column-position constraint.
## v2.1.1
### ✨ Added
- **Failure Log convention in the init prompts** — `/plan2code-init` now generates a `## Keeping this file current` / `## Failure log` pair at the end of `AGENTS.md`, so agents record repo-specific corrections as one imperative line each instead of relearning them. The section is always-inline (never split to `.agents-docs/`) and is listed in the generated `CLAUDE.md` pointer block. `/plan2code-init-update` gains a matching **Failure Log Audit** step that flags the section as missing on existing `AGENTS.md` files and offers to add it.
- **`plan2code-changelog` skill** (`.claude/skills/plan2code-changelog/`) — validates the CHANGELOG version before a PR is opened. Reads `main` for the current latest version, classifies the branch's changes to pick minor vs patch, and keeps `CHANGELOG.md`, `package.json`, and `version.json` in lockstep. Exists because parallel branches pick colliding version numbers independently.
### 🔧 Changed
- **`AGENTS.md` preamble shortened** in the init template — the generated header no longer enumerates specific agents (Claude Code, Cursor, Codex, Copilot, Devin, Zed) and reads "AI coding agents working with code in this repository."
- **`CLAUDE.md` added at the repo root** — a pointer file to `AGENTS.md`, matching what `/plan2code-init` now tells agents to generate.
- **`CLAUDE.md` template reconciled between the two init prompts** — `/plan2code-init` emitted the `CRITICAL — MANDATORY FIRST STEP` directive followed by a seven-item bullet list, while `/plan2code-init-update`'s Step 7 reference emitted the directive followed by a single prose line. Running one workflow after the other rewrote `CLAUDE.md` back and forth. `src/plan2code-init-update-references/ai-agent-file-sync.md` now carries the bullet-list form for both the `CLAUDE.md` and the generic reference template.
### 🐛 Fixed
- **Typos in the Failure Log template** — the template shipped in `src/plan2code-init.md` and `src/plan2code-init-update.md` said `AGETNS.md`, `mistage`, and "mistakes make by". Since agents copy this block verbatim into generated `AGENTS.md` files, the errors propagated into every project initialized with it.
## v2.1.0
### ✨ Added
- **GitHub Issues backend for Pathfinder** — `/plan2code-0-pathfinder` no longer assumes local files. Chart Step 1 now asks, HITL and never self-picked, where the map should live: **local** (the default — files under gitignored `specs/<idea>/pathfinder/`, private and solo) or **github** (a `pathfinder:map` issue whose decision questions are its sub-issues, driven by the `gh` CLI). The pick is recorded as the first `## Ground rules` bullet, never re-asked and never switched mid-map, and Auto-Discovery resolves either backend — an issue URL or number as the argument routes straight to `github`.
On `github`, the local model maps onto the tracker's own primitives rather than being simulated in issue bodies: a question is a **sub-issue** (`sub_issues` endpoint), blocking is GitHub's **native issue dependencies** (`dependencies/blocked_by`, so the frontier renders in GitHub's UI without opening the map), the claim is the **assignee**, `Type:` becomes a single `pathfinder:<type>-<mode>` label so type and mode cannot drift, `Locked: yes` becomes `pathfinder:locked`, and resolution is an `## Answer` comment followed by a close — `completed` for a decision, `not planned` for a question ruled out of scope. Both wiring calls key on the issue's **database id**, not its `#number`. The map body therefore carries no question checklist at all: the frontier is a live query, which removes the single largest source of drift in the local backend.
Three things stay on local disk whatever the backend: runnable sketches (`specs/<idea>/pathfinder/sketch-NN/`), anything secret, and the `PLAN-DRAFT-<date>.md` — `/plan2code-1-plan` discovers its input with `ls specs/` and has no notion of a tracker, so a draft that existed only as an issue would be invisible to the rest of the pipeline.
Guardrails carried over from the local backend's assumptions: `github` is only offered after a five-check preflight (`gh` present, authenticated, GitHub remote, issues enabled, push access), and the offer must name the repo's **visibility** in the same breath, because a map on a public tracker publishes the destination, the rejected alternatives, and the codebase recon. The Step 3 recon is held in-session and published at Step 6, so the Step 4 no-fog off-ramp leaves no litter on a shared tracker. A `gh` failure mid-session stops the session rather than falling back to local files, which would fork the map.
Depth lives in a seventh reference file, `src/plan2code-0-pathfinder-references/github-issues.md` — preflight, label set, the local↔GitHub equivalence table, create-then-wire charting, the frontier query, resolve and out-of-scope flows, reconcile, the trail footer, handoff, and a failure-mode table. Adapted from the GitHub tracker doc behind Matt Pocock's [`wayfinder`](https://github.com/mattpocock/skills/tree/main/skills/engineering/wayfinder) skill (MIT).
### 🔧 Changed
- **`src/plan2code-0-pathfinder.md` recompressed** to absorb the new `## Backend` section within the 11,000-character workflow-file limit — duplication between the orchestrator and its reference files was removed (the local layout and marker legend now live only in `questions.md`; Form A/B footer detail only in `trail.md`), and step text tightened. No behaviour was dropped.
## v2.0.0
Ports upstream v1.17.0 and v2.0.0 into Plan2Code.
+14
View File
@@ -0,0 +1,14 @@
# CLAUDE.md
**CRITICAL — MANDATORY FIRST STEP: You MUST read [AGENTS.md](./AGENTS.md) before responding to ANY user message, including simple questions. Do NOT skip this step regardless of how trivial the request appears. No exceptions.**
See AGENTS.md for complete project documentation including:
- Development commands and setup
- Architecture overview
- Workflow prompt reference
- Plan2Code Loop CLI details
- Code style and gotchas
- Keeping this file current / Failure log
- Section details in .agents-docs/
This file exists for Claude Code auto-loading. All AI coding agents should reference AGENTS.md.
+54 -47
View File
@@ -1,29 +1,50 @@
# Plan2Code Quick Reference
## Launch
The web console is the main way in: a local page where questions, documents, build progress and sign-offs live.
| How | Command | What happens |
| ------------------------- | ------------------------------------ | -------------------------------------------------------------------------------------- |
| From your project (shell) | `plan2code` | Opens Claude Code or Devin on the dashboard (`--cli claude` / `--cli devin` to choose) |
| Inside any agent | `/plan2code` | Opens the dashboard: a card for every skill, launched on the same page |
| Install / update | `npx --allow-git=all git+https://github.com/jparkerweb/plan2code.git` | Installs the skills and the `plan2code` command; re-run it to update |
| Straight into a skill | `/plan2code-1-plan --web` (any step) | Skips the "web console or terminal?" question and opens the page |
Pick a skill on the dashboard and the same page becomes that session. Details: [.readme/web-console.md](.readme/web-console.md) · worked example: [.readme/walkthrough.md](.readme/walkthrough.md).
## Commands
| Step | Command | Input | Output |
| ------- | ----------------------------- | --------------- | -------------------------------------------------- |
| Init | /plan2code-init | None | AGENTS.md file |
| Update | /plan2code-init-update | AGENTS.md | Updated AGENTS.md |
| 0 | /plan2code-0-pathfinder | A foggy idea | pathfinder/map.md + PLAN-DRAFT-<date>.md |
| quick | /plan2code-quick-task | Requirements | Conversational plan (standalone — not a pipeline step) |
| review | /plan2code-review | Scope guidance | Review findings + fixes |
| 1 | /plan2code-1-plan | Requirements | PLAN-CONVERSATION-<date>.md + PLAN-DRAFT-<date>.md |
| 1b | /plan2code-1b-revise-plan | Specs + changes | Updated specs |
| 2 | /plan2code-2-document | PLAN-DRAFT.md | overview.md + Phase files |
| 3 | /plan2code-3-implement | overview.md | Implemented code |
| 4 | /plan2code-4-finalize | overview.md | Archived specs |
| handoff | /plan2code-handoff | Conversation | Self-contained handoff doc in handoffs/ |
Every command below is what a dashboard card runs. Typed on its own, each one asks "web console or terminal?" first.
| Step | Command | Input | Output |
| -------- | ------------------------------- | --------------- | --------------------------------------------------------------------------------------- |
| — | /plan2code | None | Web console dashboard — a menu of every skill below, launched on the page |
| Init | /plan2code-init | None | AGENTS.md file |
| Update | /plan2code-init-update | AGENTS.md | Updated AGENTS.md |
| 0 | /plan2code-0-pathfinder | A foggy idea | pathfinder/map.md *or* GitHub Issues + PLAN-DRAFT-<date>.md |
| quick | /plan2code-quick-task | Requirements | A short plan, then the change built (standalone, not a pipeline step) |
| review | /plan2code-review | Scope guidance | Review findings + fixes |
| 1 | /plan2code-1-plan | Requirements | PLAN-CONVERSATION-<date>.md + PLAN-DRAFT-<date>.md |
| 1b | /plan2code-1b-revise-plan | Specs + changes | Updated specs |
| 2 | /plan2code-2-document | PLAN-DRAFT.md | overview.md + Phase files |
| 3 | /plan2code-3-implement | overview.md | Implemented code |
| 3+review | /plan2code-3-implement-review | overview.md | Reviewed implementation ready for approval (alternative to `/plan2code-3-implement`) |
| 4 | /plan2code-4-finalize | overview.md | Archived specs |
| handoff | /plan2code-handoff | Conversation | Self-contained handoff doc (OS temp dir by default, or a folder you pick such as `handoffs/`) |
**Terminal instead:** run any command without `--web` and answer the interface question with **terminal**. Nothing is lost: same questions, same gates, same files under `specs/`. You can switch between the two mid-feature.
## File Structure
```
specs/
└── <feature-name>/
├── pathfinder/ # From Step 0 (optional, if charted)
├── pathfinder/ # From Step 0 (optional, if charted locally)
│ ├── map.md # the map: destination, decisions, fog
│ └── questions/NN-<slug>.md # one decision question per file
│ ├── questions/NN-<slug>.md # one decision question per file
│ └── briefs/brief-<date>.md # on-demand plain-English decision briefs
│ # (GitHub Issues backend: map issue + sub-issues instead)
├── PLAN-DRAFT-<date>.md # From Step 1 (verified plan)
├── PLAN-CONVERSATION-<date>.md # From Step 1 (conversation log)
├── overview.md # From Step 2
@@ -37,9 +58,9 @@ Note: `<date>` uses YYYYMMDD format (e.g., `20250204`)
## Key Rules
- Start NEW conversation for each step (and each implementation phase)
- Keep sessions short: fresh console session when the meter turns red, and before a big plan or build step
- ONE phase per conversation (but parallel phases can run in separate instances)
- Reply "approved" to complete phases
- Approve phases on the sign-off card (or reply "approved" in the terminal)
- 90% confidence required before planning completes
- Never look in `specs--completed/` (it's archived specs)
@@ -50,27 +71,29 @@ Note: `<date>` uses YYYYMMDD format (e.g., `20250204`)
| `[ ]` | Pending | Not started |
| `[/]` | In Progress | Agent working (or paused) |
| `[x]` | Complete | Approved |
| `[!]` | Blocked | Couldn't be done; the line says why and what you need to do |
| `[?]` | Assumed | Couldn't be verified, assumed complete; check it |
## Parallel Execution
When phases have no file conflicts or dependencies, they can run simultaneously:
1. Documentation Mode auto-detects parallel-eligible phases
2. Implementation Mode shows selection UI with status for each phase
2. Implementation Mode asks which phase to take (a card on the web console), with the status of each
3. Run multiple `/plan2code-3-implement` instances on different phases
4. `[/]` status shows which phases are actively being worked on
## Quick Troubleshooting
| Issue | Solution |
| ---------------------- | ------------------------------------------------- |
| Lost context mid-phase | Attach spec files, say "resume from Task X.Y" |
| Wrong phase started | Say "abort", start correct phase |
| Issue | Solution |
| ---------------------- | ------------------------------------------------ |
| Lost context mid-phase | Attach spec files, say "resume from Task X.Y" |
| Wrong phase started | Say "abort", start correct phase |
| Need to change plan | Use `/plan2code-1b-revise-plan` |
| Multiple spec folders | Specify which: "Continue with specs/user-auth/" |
| Need AGENTS.md file | Use `/plan2code-init` to generate one |
| Update AGENTS.md | Use `/plan2code-init-update` after sessions |
| Run phases in parallel | Check Parallel Execution Groups in overview.md |
| Multiple spec folders | Specify which: "Continue with specs/user-auth/" |
| Need AGENTS.md file | Use `/plan2code-init` to generate one |
| Update AGENTS.md | Use `/plan2code-init-update` after sessions |
| Run phases in parallel | Check Parallel Execution Groups in overview.md |
## Workflow Decision
@@ -81,8 +104,9 @@ New to a project?
Learned something during a session?
└── /plan2code-init-update → Add learnings to AGENTS.md
Too unclaer to plan? (big idea, don't yet know what the questions are)
└── /plan2code-0-pathfinder → chart it, clear one decision per session
Too foggy to plan? (big idea, don't yet know what the questions are)
└── /plan2code-0-pathfinder → chart it, clear one decision at a time
├── continue-or-stop menu between decisions; fresh session recommended after ~3
└── then → /plan2code-1-plan (resumes at Phase 4)
Is it a quick, small task?
@@ -90,6 +114,7 @@ Is it a quick, small task?
└── No → /plan2code-1-plan (full workflow)
├── /plan2code-2-document
├── /plan2code-3-implement (repeat per phase)
│ ├── OR: /plan2code-3-implement-review (review gate before approval)
│ └── OR: plan2code-loop (autonomous alternative)
└── /plan2code-4-finalize
@@ -99,22 +124,4 @@ Need to revise mid-implementation?
## Autonomous Loop (Alternative)
The `plan2code-loop` CLI is an **alternative** to Step 3, not a replacement.
| Approach | Use When |
|----------|----------|
| `/plan2code-3-implement` | You want interactive control per phase |
| `plan2code-loop` | You want hands-off autonomous execution |
```bash
plan2code-loop # Fully interactive - auto-detects specs, prompts for options
```
### Loop Modes
| Mode | Description |
|------|-------------|
| **One task per loop** (default) | One task per agent invocation. Node handles git commits. |
| **One phase per loop** | All tasks in a phase per invocation. LLM handles git commits. Best for smart models with larger context. |
Session state stored per-spec in `specs/<feature>/.plan2code-loop/`
`plan2code-loop` is a hands-off alternative to Step 3. See [.readme/autonomous-loop.md](.readme/autonomous-loop.md).
+213 -96
View File
@@ -1,48 +1,93 @@
# Plan2Code
<img src="docs/banner.png" alt="Plan2Code — send the plan, the build follows" style="max-width:1024px;">
**A spec-driven workflow for AI coding agents.**
**A spec-driven workflow for AI coding agents. Send the plan — the build follows.**
<img src="docs/banner.jpg" alt="banner" data="object-fit:contain;" style="max-width:1024px;">
An AI agent is a fine builder and a terrible client. Plan2Code stops making it both: you approve a
plan, the plan becomes a set of phase documents in your repo, and the agent builds to those documents
one phase at a time. Progress lives in files instead of chat history — so the next session, the next
agent, and the next engineer all start from the same specs.
You work through Plan2Code on a page in your browser. The dashboard lists every step as a card, and
each step's questions, progress and sign-offs arrive on that same page while your agent does the
work. What you decide still lands in your repo as plain files: an approved plan becomes phase
documents, and the agent builds to them one phase at a time, so the next session, the next agent and
the next engineer all start from the same specs.
Six commands, each posted separately. Two of them are optional.
One command opens it, every step is a card, and the terminal is always there if you prefer it.
Version 2.0.0 · MIT · 📖 [plan2code.jparkerweb.com](https://plan2code.jparkerweb.com)
Version 2.4.2 · MIT · 📖 [plan2code.jparkerweb.com](https://plan2code.jparkerweb.com)
---
## Install
## See it
Requires [Node.js](https://nodejs.org/) 14 or later. Re-run any time to update.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="docs/screenshots/dashboard-dark.webp">
<img src="docs/screenshots/dashboard-light.webp" alt="The Plan2Code dashboard for a project mid-build: a Set up card, then the steps in order, with the Implement card marked Suggested and reading Phase 3 of 4 is next.">
</picture>
The dashboard, in a project halfway through a build. It suggests the next step for the spec you pick.
---
## Install and start
Requires [Node.js](https://nodejs.org/) 18 or later and network access — installation runs through
the [skills CLI](https://skills.sh). To update, re-run the install command.
```bash
npx --allow-git=all git+https://github.com/jparkerweb/plan2code.git
```
This fetches the installer to a temp directory, runs it, writes the slash commands for whichever
tools you pick, and cleans up after itself. The installed commands work independently from then on.
This fetches the installer to a temp directory, builds the workflow as Agent Skills, delegates
installation to `skills add`, and cleans up after itself. The installed skills work independently
from then on.
Either route lands you on the same menu:
The installer opens this menu (cloning and running `node install.js`, below, opens the same one):
```
╔═════════════════════════════════════════════════════════╗
║ INSTALL PLAN2CODE ║
╠═════════════════════════════════════════════════════════╣
║ I. INSTALL Install Plan2Code for all platforms ║
║ I. INSTALL Install Plan2Code skills everywhere ║
║ A. ALL Install Plan2Code + dev tools ║
║ U. UNINSTALL Remove Plan2Code files ║
║ U. UNINSTALL Remove Plan2Code skills and dev tools ║
║ C. CUSTOM Advanced options ║
║ Q. QUIT Exit ║
╚═════════════════════════════════════════════════════════╝
```
**Supported tools:** Claude Code · Cursor · Windsurf · Continue · Codeium (IntelliJ) ·
GitHub Copilot CLI · VS Code Copilot · Crush · Pi · Amp · OpenCode · Devin · Zed
### Start
1. In any project, run `plan2code`. It starts Claude Code or Devin on the dashboard, and the
dashboard opens in your browser.
2. In a project without an `AGENTS.md`, pick **Set up this project** first: Plan and Quick task
need it.
3. Pick a card.
Already inside your agent? Type `/plan2code` instead.
<details>
<summary>About the plan2code command</summary>
The installer adds `plan2code` as a global command (Custom `C` → `D` adds just the command). Run it from any project and the Plan2Code
dashboard opens right there, in **Claude Code** or **Devin**. With both installed it asks which and
remembers your pick; `plan2code --cli claude` or `--cli devin` skips the question, and `--model <name>`
chooses the model. It works in cmd, PowerShell, Git Bash, macOS and Linux shells, and needs the
Claude Code or Devin CLI on your `PATH`. It starts the agent with permission prompts off
(`--permission-mode bypassPermissions` in Claude Code, `--permission-mode bypass` in Devin); run
`claude /plan2code` yourself if you want prompts. The installer then offers a desktop shortcut (answer `y`),
which opens a folder picker first. Uninstalling (`U`) removes both.
</details>
**Supported tools:** every agent supported by the skills CLI, including Claude Code · Cursor ·
GitHub Copilot · Windsurf · Codex · Continue · Codeium · Zed · Amp · OpenCode · Devin · Crush · Pi ·
Gemini CLI · Cline · Roo · Kilo · Goose · Trae · Qwen Code.
The installer keeps one canonical copy of each skill under `~/.agents/skills/` and links it into
agents that maintain their own directory. To update, re-run the install command.
Use the installer rather than calling `skills add` against the repository root: recursive discovery
would also find maintainer-only skills under `.claude/skills/`. The installer targets `skills/`
explicitly.
<details>
<summary>Prefer to clone?</summary>
@@ -59,11 +104,54 @@ npm install && npx husky
---
## A session on the page
**Answer the questions.** They arrive a few at a time, each option with its trade-off written beside
it, and a progress bar says how far along you are.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="docs/screenshots/question-dark.webp">
<img src="docs/screenshots/question-light.webp" alt="A Plan question card, How tasks are paged, with three options, the first marked Suggested, each with its trade-off underneath.">
</picture>
**Watch the build.** Tasks tick off as they land, and the line beside Planny says what is being
worked on right now.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="docs/screenshots/build-dark.webp">
<img src="docs/screenshots/build-light.webp" alt="A phase of a task API being built: 5 of 11 done on the progress bar, and the phase's task list with the first five ticked.">
</picture>
**Sign off.** Read the completion report, then approve the phase or ask for changes. In Implement,
you can also have the code reviewed first; Implement + review has already done it.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="docs/screenshots/signoff-dark.webp">
<img src="docs/screenshots/signoff-light.webp" alt="The Approve this phase card, with a button to read the completion report and three choices: Review the code first, Approve this phase, and I want changes.">
</picture>
**Ask a quick question any time.** The Ask tab is a chat beside the cards, answered at the agent's
next check-in, and it never holds up your answers.
<picture>
<source media="(prefers-color-scheme: dark)" srcset="docs/screenshots/ask-dark.webp">
<img src="docs/screenshots/ask-light.webp" alt="The Ask tab: the question Where are tasks paged? and Plan2Code's answer naming the file.">
</picture>
There is more on the page when you need it. Notes on any question can carry screenshots and
documents. The folder label in the bottom bar opens your workspace, where you add other folders
as read-only context and point at them by `@name`. The session meter in the top bar turns from green
to red as a session grows long and suggests a fresh one. **Stop session** saves your place and gives
you the command to carry on later, and **Back to the dashboard** takes you home to pick the next
step in the same session.
---
## The workflow
```
┌╴╴╴╴╴╴╴╴╴╴╴╴┐
╎0 PATHFINDER╎ optional · new in 2.0 · for an idea too big or unclear to plan
╎0 PATHFINDER╎ optional · for an idea too big or unclear to plan
└╴╴╴╴╴╴┬╴╴╴╴╴┘
▼
┌────────────┐ ┌────────────┐ ┌────────────┐ ┌╴╴╴╴╴╴╴╴╴╴╴╴┐ ┌────────────┐
@@ -71,33 +159,37 @@ npm install && npx husky
│decide what │ │ draw it as │ │build to the│ ╎ optional ╎ │verify, sum,│
│ to build │ │phase specs │ │ drawing │ ╎ any time ╎ │ archive │
└────────────┘ └────────────┘ └────────────┘ └╴╴╴╴╴╴╴╴╴╴╴╴┘ └────────────┘
new chat new chat new chat/phase new chat new chat
short sessions: fresh when the meter turns red, and before a big step
├◀─────────────── one feature, start to archive ──────────────────▶┤
```
Every box is its own conversation. That is not a style preference — planning context leaking into
implementation is where most agent drift starts.
Each box is a card on the dashboard. Chaining a couple of small steps in one session is fine, but
start a fresh session before a big plan or build step: planning context leaking into implementation
is where most agent drift starts.
| Command | Use it when |
|---------|-------------|
| `/plan2code-0-pathfinder` | The idea is too big and unclear to plan. Charts it as decisions, clears one per session, hands a hot plan draft to Step 1 |
| `/plan2code-1-plan` | Starting a feature. Full requirements → architecture pass |
| `/plan2code-2-document` | Planning is done. Turn the plan into phase specs |
| `/plan2code-3-implement` | Build the next phase (one per conversation) |
| `/plan2code-review` | Independent second opinion on local changes, then optional fixes |
| `/plan2code-4-finalize` | All phases done. Validate, summarize, archive |
| `/plan2code-init` | Generate this repo's `AGENTS.md` so every agent starts informed |
| `/plan2code-init-update` | Fold what you learned this session back into `AGENTS.md` |
| `/plan2code-quick-task` | A small change that doesn't warrant the full sequence |
| `/plan2code-1b-revise-plan` | Requirements moved mid-build. Revise the specs, not the code |
| `/plan2code-handoff` | Compact this conversation into a doc the next one resumes from |
The dashboard's cards, and when each one lights up for the spec you pick:
| Card | What it is for | Lights up when |
|------|----------------|----------------|
| Set up this project | Reads your project and writes AGENTS.md, the guide every later step works from. | The project has no AGENTS.md yet |
| Refresh AGENTS.md | Brings AGENTS.md and its notes up to date with what the project has learned lately. | The project has an AGENTS.md |
| Pathfinder | A foggy idea becomes a clear list of decisions, settled one question at a time. | You start fresh, or the spec is still being explored |
| Plan | Requirements, technical design and your approval: the shape of what gets built. | You start fresh, or the spec has a finished Pathfinder map or a drafted plan |
| Revise the plan | Requirements changed mid-build? Adjust the plan without starting over. | The spec is written up into phases and not fully built |
| Document | The plan becomes a checklist of small tasks, grouped into phases, that a developer can just do. | The spec has a drafted plan |
| Implement | Builds the next phase, task by task, with a report and your sign-off at the end. | The spec has phases with tasks still to build |
| Implement + review | One phase built, then a careful review of the changes, then your sign-off. All in one run. | The spec has phases with tasks still to build |
| Finalize | Checks the work, writes the summary, files the specs away and suggests the commit message. | Every task in the spec is done |
| Quick task | Too small for the full process? A quick plan, then it just builds it. | Always |
| Review | A second opinion on the changes. Problems ranked, and you pick what gets fixed. | Always |
| Handoff | Pack this conversation into a document a fresh session can pick up cold. | Always |
---
## The four rules that do most of the work
**1 · A fresh conversation for each step, and each implementation phase.**
Step 3 gets a new chat per phase, not one chat for all of them.
**1 · Keep each session short.**
Chaining a couple of small skills on the dashboard is fine. Start a fresh console session when the session meter turns red, and always before a big planning or build step. The specs on disk carry everything between sessions.
**2 · No code until the plan hits 90% confidence.**
Step 1 will not finalize below the threshold. Under it, the agent keeps asking and keeps reading your
@@ -106,48 +198,20 @@ code — and writes every assumption down where you can argue with it.
**3 · Checkboxes are the state, not the chat.**
Progress lives in the spec files. Any agent, any session, resumes cold from them.
**4 · Reply `approved` to close a phase.**
Nothing advances on a guess about what you meant.
**4 · Approve a phase to close it.**
On the sign-off card, or by replying `approved` in the terminal. Nothing advances on a guess about
what you meant.
---
## What lands in your repo
```
your-project/
├── specs/
│ └── task-api/ ← in progress
│ ├── pathfinder/ ← only if you charted it in Step 0
│ │ ├── map.md the destination, the decisions, the fog
│ │ └── questions/NN-<slug>.md one decision per file
│ ├── PLAN-DRAFT-20260804.md ← Step 1: the verified plan
│ ├── PLAN-CONVERSATION-*.md ← Step 1: how you got there
│ ├── overview.md ← Step 2: phase list + parallel groups
│ └── Phase 1.md … Phase N.md ← Step 2: one-point tasks, self-contained
├── specs--completed/
│ └── auth-refresh/ ← Step 4 files finished work here
└── ...your code
```
Each feature gets a folder under `specs/` (the plan, the overview and one file per phase), and Step 4
moves it to `specs--completed/`. The layout and progress marks are on the
[quick reference](QUICK-REFERENCE.md#file-structure).
`specs/` is gitignored by default — it's your working drawing, not a deliverable. Share a folder
deliberately with `git add -f` when you want to.
### Progress marks
| Mark | Status | Meaning |
|------|--------|---------|
| `[ ]` | Open | Unclaimed. Any agent picks it up cold. |
| `[/]` | In progress | Claimed right now — which is how two agents run parallel phases without colliding. |
| `[x]` | Done | Built, self-reviewed against the spec, approved by you. |
```markdown
## Phases
- [x] Phase 1: Project setup
- [x] Phase 2: Data model
- [/] Phase 3: API endpoints ← an agent is on this now
- [ ] Phase 4: Authentication ← next available
```
Add `specs/` to your project's `.gitignore` (`plan2code-loop` adds it for you): it's your working
drawing, not a deliverable. Share a folder deliberately with `git add -f` when you want to.
Step 2 marks which phases don't share files. Open a second agent on one of those, and the `[/]` marks
keep the two out of each other's way.
@@ -156,21 +220,29 @@ keep the two out of each other's way.
## The six steps in detail
Each one travels on its own — a fresh conversation, opened and closed, with the specs on disk as the
only thing carried between them.
Each step is its own run, and the specs on disk are the only thing carried from one to the next.
Chain a small one onto the last, or start a fresh session for a big one (rule 1).
### 0 · Pathfinder 🧭 — optional, new in 2.0
### 0 · Pathfinder 🧭 (optional)
On the page: one question at a time with the options and their trade-offs, the map growing in its own
tab, and a **Write a brief** button for a printable summary.
Some ideas are too big and unclear to plan: you can feel the shape of the work but you can't write
it as requirements, so planning would just invent the answers. Pathfinder finds the *way* to the
destination; Step 1 then walks it.
1. **Name the destination** — one or two lines fixing what this effort is finding its way to. Settled
first, because it fixes scope.
first, because it fixes scope. It also asks where the map should live: **local files** under
gitignored `specs/` (private, solo — the default), or **GitHub Issues** (a map issue with one
sub-issue per decision, native blocking, so your team can see and work the frontier in the tracker).
2. **Chart the map** — a breadth-first grilling surfaces the open decisions. Anything you can phrase
*sharply* becomes a question file; anything you can only sense stays listed as fog.
3. **Clear one question per session** — resolving a question burns off the fog behind it, graduating
whatever just became sharp into new questions.
3. **Clear the questions one at a time** — resolving one burns off the fog behind it, graduating
whatever just became sharp into new questions. After each decision it offers a menu: take the next
question here, or start fresh — recommended after about three, to keep the agent sharp.
Ask for a **brief** at any point and you get a dated, plain-English summary of what has been
decided, what is still open, and what is next — meeting minutes, with no jargon in them.
4. **Hand off** — when nothing is left to decide, it writes a `PLAN-DRAFT` that
`/plan2code-1-plan` resumes from at Phase 4, with requirements, context, and scope already
answered.
@@ -182,13 +254,18 @@ react to) · `legwork` (manual work that has to happen before a decision is poss
It never answers its own questions, and it **plans, it never builds.** When the urge to just build it
arrives, the map is done. Skip Step 0 entirely when you already know what you're building.
**Out:** `specs/<feature>/pathfinder/map.md` + `questions/` → `PLAN-DRAFT-<date>.md`
**Out:** `specs/<feature>/pathfinder/map.md` + `questions/` (or a `pathfinder:map` issue and its
sub-issues) → `PLAN-DRAFT-<date>.md`. The draft is always a local file — that is what Step 1 reads.
### 1 · Plan 🤔
The agent works as a senior architect through six phases, stopping for you after each: requirements
analysis · system context (reading your actual codebase) · tech stack (needs your explicit sign-off) ·
architecture design · technical specification · transition decision.
On the page: questions in batches of up to three, the phase breakdown as a list you can reorder and
edit, and a sign-off at each stage.
The agent works as a senior architect through seven phases, stopping for you after each: requirements
analysis · system context (reading your actual codebase) · scope assessment (a large project saves and
resumes in a fresh session) · tech stack (needs your explicit sign-off) · architecture design ·
technical specification · transition decision.
It won't finalize below **90% confidence**, and every assumption it makes is written into the draft.
@@ -196,17 +273,23 @@ It won't finalize below **90% confidence**, and every assumption it makes is wri
### 2 · Document 📝
On the page: a question only where the plan leaves something open, each phase file appearing as its
own tab as it is written, and one sign-off at the end.
The plan becomes the drawing. One `overview.md` with the phase checklist, plus one file per phase of
one-story-point tasks. Each phase is **self-contained** — an agent opening `Phase 3.md` cold needs
one-story-point tasks. Each phase is **self-contained** — an agent opening `phase-3.md` cold needs
nothing else to build it. Unit and E2E tests are excluded unless you ask for them.
The overview also identifies the **parallel execution groups**: phases with no shared files or
dependencies, safe to run in separate agents at once.
**In:** the `PLAN-DRAFT`. **Out:** `overview.md` + `Phase 1…N.md`
**In:** the `PLAN-DRAFT`. **Out:** `overview.md` + `phase-1…N.md`
### 3 · Implement ⚡
On the page: a progress bar and the phase's task list ticking over as tasks land, then the completion
report and a sign-off card where you approve, ask for changes, or have the code reviewed first.
Point it at `overview.md` and it does the rest: finds the next unchecked phase, implements every task
exactly as specified, ticks tasks off as they land, then reviews its own work against the spec and
writes a completion summary.
@@ -215,8 +298,14 @@ One phase per conversation. It won't run tests unless the phase says to.
**In:** `specs/<feature>/overview.md`. **Out:** working code, and updated checkboxes.
Want a second pair of eyes before you sign off? `/plan2code-3-implement-review` builds the phase the
same way, runs a focused review and the fixes you pick, and only then asks for approval.
### Review 🔬 — optional, any time
On the page: a few questions on scope, the findings as a report you can read in full, and a list
where you tick what gets fixed.
An independent second opinion, not a rubber stamp. It figures out its own scope (conversation
context, your instruction, or the git diff as a fallback), analyses across 11 dimensions, and ranks
findings Critical / Warning / Suggestion. Every finding cites a file and a line, or it gets dropped —
@@ -227,6 +316,9 @@ implementation step, but there's no wrong time to run it.
### 4 · Finalize 🧹
On the page: the task audit and the summary as reports, a list of the doc updates to approve, and a
confirmation before anything is archived.
Validates every task against its phase spec, writes the summary and the list of files touched, flags
the docs that drifted (`README`, `CHANGELOG`, `AGENTS.md`), then archives the whole spec folder —
`pathfinder/` included — to `specs--completed/`. That folder is the record of *why* the code looks
@@ -236,23 +328,39 @@ like this.
---
## What to bring to each step
## Prefer the terminal?
| Step | Required input |
|------|----------------|
| 0 · Pathfinder | Nothing to start — just describe the idea. To continue: the feature name; it finds its own map |
| 1 · Plan | Nothing — describe the feature |
| 2 · Document | `specs/<feature>/PLAN-DRAFT-<date>.md`, or the planning conversation |
| 3 · Implement | `specs/<feature>/overview.md` — it detects the phase itself |
| Review | Scope guidance, e.g. "the last two phases", "just the auth module", "the whole PR". Auto-detects if you give none |
| 4 · Finalize | `specs/<feature>/overview.md` |
Every skill also runs right in the terminal. Each one asks at the start whether you want the web
console or the terminal; add `--web` to skip the question and go straight to the web console:
`/plan2code-1-plan --web plan a REST API for tasks`.
| Command | Use it when |
|---------|-------------|
| `/plan2code` | Open the dashboard in the web console: every skill below as a card, launched on the page |
| `/plan2code-0-pathfinder` | The idea is too big and unclear to plan. Charts it as decisions, clears them one at a time, hands a hot plan draft to Step 1 |
| `/plan2code-1-plan` | Starting a feature. Full requirements → architecture pass |
| `/plan2code-2-document` | Planning is done. Turn the plan into phase specs |
| `/plan2code-3-implement` | Build the next phase (one per conversation) |
| `/plan2code-3-implement-review` | Build the next phase and review it before you approve it (instead of `/plan2code-3-implement`) |
| `/plan2code-review` | Independent second opinion on local changes, then optional fixes |
| `/plan2code-4-finalize` | All phases done. Validate, summarize, archive |
| `/plan2code-init` | Generate this repo's `AGENTS.md` so every agent starts informed |
| `/plan2code-init-update` | Fold what you learned this session back into `AGENTS.md` |
| `/plan2code-quick-task` | A small change that doesn't warrant the full sequence |
| `/plan2code-1b-revise-plan` | Requirements moved mid-build. Revise the specs, not the code |
| `/plan2code-handoff` | Compact this conversation into a doc the next one resumes from |
---
## Troubleshooting
**Slash commands aren't recognised.** Re-run `node install.js` for that platform and restart your AI
tool. For a per-project install, check the directory isn't gitignored.
**The skills aren't recognised.** Re-run the installer and restart your AI tool. Confirm the global
install with `npx skills list -g`. For a project install, check the generated directories aren't
gitignored.
**Your tool doesn't read Agent Skills.** Since v2.2.0, Plan2Code ships only as skills. If your tool
has no skill support, paste the relevant `src/plan2code-*.md` manually or point it at the installed
copy under `~/.agents/skills/`.
**The agent starts coding during planning.** The prompts forbid it, but models drift. Say: "Stay in
planning mode. Do not write code yet."
@@ -269,13 +377,21 @@ the requirements.
**Too many or too few phases.** Fix it in Step 2 — a phase should be a logical grouping of work, not
a fixed size.
**The agent runs on another machine, or over SSH.** The page is served on `127.0.0.1` of the machine
the agent runs on, so your browser can only reach it there. A forwarded port often works; otherwise
work in the terminal, which loses nothing.
**Web console problems** (lost contact, a closed tab, a dead link): see
[.readme/web-console.md](.readme/web-console.md#troubleshooting).
---
## Customizing
The prompts are yours to edit. Common changes: add testing requirements in Step 2, move the 90%
confidence threshold in Step 1, restructure the `specs/` layout, or add review gates to Step 3.
Source files live in `src/`; re-run `node install.js` to push your edits out to every platform.
Source files live in `src/`; run `npm run build:skills`, then re-run `node install.js` to push your
edits out through the skills CLI.
---
@@ -283,6 +399,7 @@ Source files live in `src/`; re-run `node install.js` to push your edits out to
Core reference:
- **[.readme/web-console.md](.readme/web-console.md)**: the full guide to the web console, from the dashboard to the last sign-off
- **[QUICK-REFERENCE.md](QUICK-REFERENCE.md)** — the one-page card: commands, inputs, outputs, decision tree
- **[.readme/walkthrough.md](.readme/walkthrough.md)** — one feature from a sentence to archived specs, session by session
- **[AGENTS.md](AGENTS.md)** — architecture and contributor guide for this repo
Binary file not shown.

Before

Width:  |  Height:  |  Size: 6.1 KiB

After

Width:  |  Height:  |  Size: 3.1 KiB

BIN
View File
Binary file not shown.

After

Width:  |  Height:  |  Size: 71 KiB

BIN
View File
Binary file not shown.

Before

Width:  |  Height:  |  Size: 127 KiB

BIN
View File
Binary file not shown.

Before

Width:  |  Height:  |  Size: 1.9 KiB

After

Width:  |  Height:  |  Size: 2.6 KiB

BIN
View File
Binary file not shown.

Before

Width:  |  Height:  |  Size: 584 B

After

Width:  |  Height:  |  Size: 884 B

+6 -32
View File
@@ -1,33 +1,7 @@
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 64 64" width="64" height="64" role="img" aria-label="Plan2Code postage stamp">
<defs>
<!-- Perforated stamp silhouette: white keeps, black bites out.
Few, deep notches so the scalloped edge still reads at 16px. -->
<mask id="perf">
<rect width="64" height="64" fill="#000"/>
<rect x="2" y="2" width="60" height="60" fill="#fff"/>
<g fill="#000">
<circle cx="2" cy="2" r="5"/><circle cx="14" cy="2" r="5"/><circle cx="26" cy="2" r="5"/>
<circle cx="38" cy="2" r="5"/><circle cx="50" cy="2" r="5"/><circle cx="62" cy="2" r="5"/>
<circle cx="2" cy="62" r="5"/><circle cx="14" cy="62" r="5"/><circle cx="26" cy="62" r="5"/>
<circle cx="38" cy="62" r="5"/><circle cx="50" cy="62" r="5"/><circle cx="62" cy="62" r="5"/>
<circle cx="2" cy="14" r="5"/><circle cx="2" cy="26" r="5"/>
<circle cx="2" cy="38" r="5"/><circle cx="2" cy="50" r="5"/>
<circle cx="62" cy="14" r="5"/><circle cx="62" cy="26" r="5"/>
<circle cx="62" cy="38" r="5"/><circle cx="62" cy="50" r="5"/>
</g>
</mask>
</defs>
<g mask="url(#perf)">
<!-- printed stamp -->
<rect x="2" y="2" width="60" height="60" fill="#D33A38"/>
<!-- the 2, set as a franking-machine numeral -->
<g fill="#FBFAF6">
<rect x="16" y="13" width="32" height="8"/>
<rect x="40" y="13" width="8" height="22"/>
<rect x="16" y="27" width="32" height="8"/>
<rect x="16" y="27" width="8" height="24"/>
<rect x="16" y="43" width="32" height="8"/>
</g>
</g>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 32 32" role="img" aria-label="Plan2Code">
<rect x="6" y="2.6" width="20" height="25.4" rx="4.4" fill="#eef3fa" stroke="#3b332c" stroke-width="2.2" />
<rect x="9.6" y="27.6" width="4.8" height="2.6" rx="1.3" fill="#3b332c" />
<rect x="17.6" y="27.6" width="4.8" height="2.6" rx="1.3" fill="#3b332c" />
<path d="M16 8.8 L17.41 12.26 L21.14 12.53 L18.28 14.94 L19.17 18.57 L16 16.6 L12.83 18.57 L13.72 14.94 L10.86 12.53 L14.59 12.26 Z" fill="#2455a4" />
<path d="M12.2 20.6 Q16 24.2 19.8 20.6" fill="none" stroke="#3b332c" stroke-width="1.8" stroke-linecap="round" />
</svg>

Before

Width:  |  Height:  |  Size: 1.6 KiB

After

Width:  |  Height:  |  Size: 636 B

+381 -349
View File
@@ -3,387 +3,419 @@
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Plan2Code — Spec-Driven Workflow for AI Coding Agents</title>
<meta name="description" content="Send the plan — the build follows. You approve a plan, it becomes phase documents in your repo, and the agent builds to those documents one phase at a time.">
<title>Plan2Code: plan and build with your AI agent in your browser</title>
<meta name="description" content="Plan and build with an AI coding agent on a page in your browser. You answer the questions and approve each phase, and the specs land in your repo as plain files.">
<meta name="color-scheme" content="light dark">
<link rel="icon" href="favicon.ico" sizes="32x32">
<link rel="icon" type="image/png" sizes="32x32" href="favicon.png">
<link rel="icon" href="favicon.svg" type="image/svg+xml">
<link rel="apple-touch-icon" href="apple-touch-icon.png">
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=Big+Shoulders+Display:wght@500;700;800&family=Karla:wght@300;400;500;600;700&family=Caveat:wght@500&display=swap" rel="stylesheet">
<style>
:root{
--air:#E4ECF6; --air-2:#D3DEEE; --paper:#FBFAF6;
--ink:#16233D; --ink-soft:#5E6E8A;
--red:#D33A38; --blue:#2455A4; --kraft:#E5D8BE;
/* The web console's palette and type (src/web-console/public/app.css), so the
site and the product look like one thing, with the Plan2Code blue as the
accent: #2455a4 in light, and the console's Ocean highlight (palette.js),
which the screenshots use, in dark. System fonts only, like the console. */
:root {
--bg: #f6f3ee;
--panel: #ffffff;
--panel-2: #fbf9f6;
--ink: #1d1a17;
--ink-2: #5d564d;
--ink-3: #8f8779;
--rule: #e4ddd1;
--rule-2: #cfc6b6;
--accent: #2455a4;
--accent-soft: #ecf2f9;
--accent-ink: #ffffff;
--accent-line: #9ab4d3;
--ok: #2f6b45;
--focus: #1f6feb;
--code-bg: #1d1a17;
--code-ink: #f0ece5;
--shadow: 0 1px 2px rgba(29, 26, 23, 0.05), 0 8px 24px rgba(29, 26, 23, 0.06);
--shadow-lg: 0 2px 4px rgba(29, 26, 23, 0.06), 0 24px 60px rgba(29, 26, 23, 0.12);
--sans: ui-sans-serif, system-ui, -apple-system, "Segoe UI", Roboto, Helvetica, Arial, sans-serif;
--serif: "Iowan Old Style", Charter, Georgia, Cambria, serif;
--mono: ui-monospace, SFMono-Regular, "Cascadia Mono", Menlo, Consolas, monospace;
--radius: 10px;
--width: 1120px;
--gutter: 24px;
}
*{margin:0;padding:0;box-sizing:border-box}
body{background:var(--air);color:var(--ink);font:400 16.5px/1.65 'Karla',system-ui,sans-serif;-webkit-font-smoothing:antialiased}
.bs{font-family:'Big Shoulders Display',sans-serif;font-weight:800;letter-spacing:.005em}
.hand{font-family:'Caveat',cursive}
.wrap{max-width:1140px;margin:0 auto;padding:0 34px}
a{color:inherit}
:focus-visible{outline:2px solid var(--red);outline-offset:3px}
/* ── SIGNATURE: airmail chevron border framing the page on all four sides ── */
:root{--fr:11px}
.frame-t,.frame-b,.frame-l,.frame-r{
position:fixed;z-index:60;pointer-events:none;
background:repeating-linear-gradient(-45deg,
var(--red) 0 14px, var(--paper) 14px 28px,
var(--blue) 28px 42px, var(--paper) 42px 56px);
}
.frame-t,.frame-b{left:0;right:0;height:var(--fr)}
.frame-t{top:0}
.frame-b{bottom:0}
.frame-l,.frame-r{top:0;bottom:0;width:var(--fr)}
.frame-l{left:0}
.frame-r{right:0}
body{padding:var(--fr)}
/* ── header (sticky, sits just under the top frame) ── */
.top{
background:var(--paper);border-bottom:2px solid var(--ink);
position:sticky;top:var(--fr);z-index:40;
}
.top .wrap{display:flex;align-items:center;gap:16px;padding-top:13px;padding-bottom:12px}
.brand{font-family:'Big Shoulders Display',sans-serif;font-weight:800;font-size:26px;letter-spacing:.01em;text-transform:uppercase}
.brand i{font-style:normal;color:var(--red)}
.avion{
margin-left:auto;border:1.5px solid var(--blue);color:var(--blue);
font-size:10px;letter-spacing:.22em;text-transform:uppercase;font-weight:700;padding:5px 11px;
@media (prefers-color-scheme: dark) {
:root {
--bg: #171513;
--panel: #201d1a;
--panel-2: #262220;
--ink: #f0ece5;
--ink-2: #b6afa3;
--ink-3: #8b8376;
--rule: #332e29;
--rule-2: #46403a;
--accent: #619fe5;
--accent-soft: #1a2533;
--accent-ink: #0c1117;
--accent-line: #3d5978;
--ok: #74c28f;
--focus: #6aa3ff;
--code-bg: #0f0d0c;
--code-ink: #f0ece5;
--shadow: 0 1px 2px rgba(0, 0, 0, 0.4), 0 8px 24px rgba(0, 0, 0, 0.35);
--shadow-lg: 0 2px 4px rgba(0, 0, 0, 0.45), 0 24px 60px rgba(0, 0, 0, 0.5);
}
}
/* ── hero: the envelope ── */
.hero{padding:56px 0 70px}
.env{
background:var(--paper);border:2px solid var(--ink);position:relative;
box-shadow:10px 12px 0 rgba(22,35,61,.14);padding:40px 42px 44px;
* { box-sizing: border-box; }
html { scroll-behavior: smooth; }
body {
margin: 0;
background: var(--bg);
color: var(--ink);
font: 17px/1.6 var(--sans);
-webkit-font-smoothing: antialiased;
overflow-wrap: break-word;
}
/* cancellation waves */
.cancel{position:absolute;right:34px;top:34px;width:210px;height:110px;pointer-events:none}
.cancel .ring{fill:none;stroke:var(--red);stroke-width:2.2}
.cancel .wv{fill:none;stroke:var(--red);stroke-width:2.2;opacity:.85}
.cancel text{font-family:'Karla',sans-serif;font-weight:700;fill:var(--red);letter-spacing:.06em}
/* stamp */
.pstamp{
position:absolute;right:250px;top:34px;width:86px;height:104px;background:var(--air-2);
border:2px solid var(--ink);padding:8px;text-align:center;
img { max-width: 100%; height: auto; }
a { color: var(--accent); text-underline-offset: 3px; }
a:hover { text-decoration-thickness: 2px; }
code { font-family: var(--mono); font-size: 0.88em; }
:focus-visible { outline: 2px solid var(--focus); outline-offset: 3px; border-radius: 4px; }
h1, h2, h3 { font-family: var(--serif); font-weight: 600; line-height: 1.15; margin: 0; }
h2 { font-size: clamp(1.8rem, 4vw, 2.5rem); margin-bottom: 12px; }
h3 { font-size: 1.45rem; margin: 8px 0; }
.muted { color: var(--ink-2); }
.wrap { max-width: var(--width); margin: 0 auto; padding: 0 var(--gutter); }
.narrow { max-width: 780px; margin: 0 auto; padding: 0 var(--gutter); text-align: center; }
.skip { position: absolute; left: -9999px; top: 0; background: var(--panel); color: var(--ink); padding: 10px 16px; z-index: 10; }
.skip:focus { left: 0; }
/* Planny, drawn as the console draws him: shell, star eye, lip, arms up. */
.planny { display: block; margin: 0 auto; width: 44px; height: 60px; overflow: visible; }
.planny .shell { fill: var(--panel); stroke: var(--ink-2); stroke-width: 1.3; }
.planny .smile, .planny .arm path { fill: none; stroke: var(--ink-2); stroke-linecap: round; }
.planny .smile { stroke-width: 0.9; }
.planny .arm path { stroke-width: 1.2; }
.planny .arm circle, .planny .foot { fill: var(--ink-2); }
.planny .eye { fill: var(--accent); }
/* Screenshots: framed like a window, one per theme via <picture>. */
.shot { display: block; border: 1px solid var(--rule-2); border-radius: var(--radius); overflow: hidden; background: var(--panel); box-shadow: var(--shadow); }
.shot img { display: block; width: 100%; }
.shot.big { border-radius: 14px; box-shadow: var(--shadow-lg); }
/* The install command and its copy button. */
.cmd { position: relative; display: inline-flex; align-items: center; gap: 12px; max-width: 100%; background: var(--code-bg); color: var(--code-ink); border-radius: var(--radius); padding: 10px 10px 10px 18px; text-align: left; }
.cmd pre { margin: 0; min-width: 0; overflow-x: auto; font: 14.5px/1.5 var(--mono); white-space: pre; }
.cmd pre span { color: var(--accent-line); user-select: none; }
.copy { flex: none; display: inline-flex; align-items: center; gap: 6px; border: 0; border-radius: 7px; background: var(--accent); color: var(--accent-ink); font: 700 13px/1 var(--sans); padding: 10px 12px; cursor: pointer; min-height: 40px; }
.copy:hover { filter: brightness(1.08); }
.copy svg { width: 16px; height: 16px; fill: none; stroke: currentColor; stroke-width: 2; stroke-linecap: round; stroke-linejoin: round; }
.copy .tick { display: none; }
.copy.-done .clip { display: none; }
.copy.-done .tick { display: block; }
.then { margin: 12px 0 0; color: var(--ink-2); font-size: 15px; }
/* Hero: centered headline, then the dashboard at full width. */
.hero { padding: 40px 0 32px; background: linear-gradient(to bottom, var(--panel), var(--bg)); }
.hero .narrow { max-width: 960px; }
.hero h1 { font-size: clamp(2.2rem, 5.6vw, 3.7rem); line-height: 1.06; margin: 12px 0 14px; }
.hero .lede { font-size: 1.2rem; color: var(--ink-2); margin: 0 auto 22px; max-width: 60ch; }
.stage { max-width: 1200px; margin: 0 auto; padding: 8px var(--gutter) 72px; }
/* Sections. */
section.s { padding: 72px 0; }
section.alt { background: var(--panel); border-block: 1px solid var(--rule); }
.intro { max-width: 62ch; margin: 0; }
/* How a session goes: alternating screenshot and text rows. */
.row { display: grid; grid-template-columns: minmax(0, 7fr) minmax(0, 5fr); gap: 48px; align-items: center; margin-top: 56px; }
.row.flip { grid-template-columns: minmax(0, 5fr) minmax(0, 7fr); }
.row.flip .shot { order: 2; }
.row .k { font-family: var(--serif); font-size: 3rem; line-height: 1; color: var(--accent); }
.row p { color: var(--ink-2); margin: 0; }
/* The pipeline: a vertical line of steps, then the utilities. */
.line { list-style: none; margin: 36px 0 0; padding: 0; position: relative; }
.line::before { content: ""; position: absolute; left: 11px; top: 10px; bottom: 26px; width: 2px; background: var(--rule-2); }
.line li { position: relative; padding: 0 0 26px 48px; }
.line li::before { content: ""; position: absolute; left: 4px; top: 7px; width: 16px; height: 16px; border-radius: 50%; background: var(--panel); border: 2px solid var(--accent); }
.line li.opt::before { border-style: dashed; }
.line h3 { display: inline; font-family: var(--sans); font-size: 1.15rem; font-weight: 650; }
.chip { display: inline-block; margin-left: 8px; vertical-align: 2px; font-size: 11px; font-weight: 700; letter-spacing: 0.08em; text-transform: uppercase; color: var(--ink-2); border: 1px solid var(--rule-2); border-radius: 999px; padding: 2px 8px; }
.line p { margin: 4px 0 0; color: var(--ink-2); max-width: 64ch; }
.line code { color: var(--ink-2); font-size: 13px; }
.utils { display: grid; grid-template-columns: repeat(3, minmax(0, 1fr)); gap: 12px; margin-top: 12px; }
.util { border: 1px solid var(--rule-2); border-radius: var(--radius); background: var(--bg); padding: 16px; }
.util h3 { font-family: var(--sans); font-size: 1rem; font-weight: 650; margin: 8px 0 4px; }
.util p { margin: 0; font-size: 14.5px; color: var(--ink-2); }
.util .chip { margin-left: 0; }
/* What the page gives you: a two-column list. */
.feat { columns: 2; column-gap: 56px; margin-top: 36px; }
.feat div { break-inside: avoid; padding: 18px 0; border-top: 1px solid var(--rule); }
.feat h3 { font-size: 1.2rem; margin: 0 0 4px; }
.feat p { margin: 0; color: var(--ink-2); }
.feat-shots { display: grid; grid-template-columns: repeat(2, minmax(0, 1fr)); gap: 24px; margin-top: 40px; }
.feat-shots figure { margin: 0; }
.feat-shots figcaption { margin-top: 10px; font-size: 14.5px; color: var(--ink-2); }
/* The rules: numbered. */
ol.rules { list-style: none; padding: 0; margin: 32px 0 0; counter-reset: r; }
ol.rules li { counter-increment: r; display: grid; grid-template-columns: 3.5rem minmax(0, 1fr); gap: 16px; padding: 22px 0; border-top: 1px solid var(--rule); }
ol.rules li::before { content: counter(r); font-family: var(--serif); font-size: 2.6rem; line-height: 1; color: var(--accent); }
ol.rules h3 { font-family: var(--sans); font-size: 1.15rem; font-weight: 650; margin: 0 0 4px; }
ol.rules p { margin: 0; color: var(--ink-2); }
/* Prefer the terminal? */
.term-band { display: grid; grid-template-columns: minmax(0, 1fr) minmax(0, 1fr); gap: 48px; align-items: center; }
pre.term { background: var(--code-bg); color: var(--code-ink); border-radius: var(--radius); padding: 22px; margin: 0; font: 14.5px/1.8 var(--mono); overflow-x: auto; }
/* Closing install and footer. */
.close { text-align: center; }
.close .muted { margin: 0 0 22px; }
footer { border-top: 1px solid var(--rule); padding: 26px 0 32px; color: var(--ink-2); font-size: 14.5px; }
footer .wrap { display: flex; flex-wrap: wrap; justify-content: center; gap: 8px 22px; }
footer a { color: inherit; }
@media (max-width: 820px) {
.row, .row.flip, .term-band, .feat-shots { grid-template-columns: minmax(0, 1fr); gap: 20px; }
.row.flip .shot { order: 0; }
.row { margin-top: 44px; }
.feat { columns: 1; }
.utils { grid-template-columns: minmax(0, 1fr); }
section.s { padding: 56px 0; }
}
.pstamp .in{border:1px solid rgba(22,35,61,.35);height:100%;display:flex;flex-direction:column;justify-content:center;gap:2px}
.pstamp b{font-family:'Big Shoulders Display',sans-serif;font-weight:800;font-size:26px;line-height:1;color:var(--blue)}
.pstamp span{font-size:8px;letter-spacing:.12em;text-transform:uppercase;color:var(--ink-soft);font-weight:700}
@media(max-width:900px){.cancel,.pstamp{display:none}}
.kick{font-size:10.5px;letter-spacing:.24em;text-transform:uppercase;color:var(--blue);font-weight:700}
h1{font-family:'Big Shoulders Display',sans-serif;font-weight:800;font-size:clamp(3rem,9vw,7rem);line-height:.86;text-transform:uppercase;margin-top:14px;max-width:13ch}
h1 em{font-style:normal;color:var(--red)}
.addr{
margin-top:32px;padding-top:26px;border-top:1.5px dashed rgba(22,35,61,.3);
display:grid;grid-template-columns:1fr 1fr;gap:36px;align-items:end;
@media (max-width: 600px) {
:root { --gutter: 16px; }
body { font-size: 16px; }
.hero { padding-top: 40px; }
.cmd { display: flex; }
.cmd pre { font-size: 13px; }
ol.rules li { grid-template-columns: 2.5rem minmax(0, 1fr); }
}
.addr .to{font-size:10px;letter-spacing:.2em;text-transform:uppercase;color:var(--ink-soft);font-weight:700}
.addr p{margin-top:10px;color:var(--ink-soft);max-width:44ch}
.acts{display:flex;flex-wrap:wrap;gap:12px}
.btn{
display:inline-block;text-decoration:none;background:var(--ink);color:var(--paper);
font-family:'Big Shoulders Display',sans-serif;font-weight:700;font-size:19px;letter-spacing:.1em;
text-transform:uppercase;padding:12px 28px;transition:background .18s,transform .18s;
@media (prefers-reduced-motion: reduce) {
html { scroll-behavior: auto; }
* { transition: none !important; animation: none !important; }
}
.btn:hover{background:var(--red);transform:translateY(-2px)}
.btn.-o{background:transparent;border:2px solid var(--ink);color:var(--ink);padding:10px 26px}
.btn.-o:hover{background:var(--ink);color:var(--paper)}
/* ── postage meter strip ── */
.meter{background:var(--ink);color:var(--paper)}
.meter .wrap{display:grid;grid-template-columns:repeat(auto-fit,minmax(140px,1fr));gap:0}
.meter div{padding:24px 18px;border-left:1px dashed rgba(251,250,246,.28)}
.meter div:first-child{border-left:none}
.meter b{display:block;font-family:'Big Shoulders Display',sans-serif;font-weight:800;font-size:2.4rem;line-height:1;color:#8FB6E8}
.meter span{display:block;margin-top:3px;font-size:9.5px;letter-spacing:.14em;text-transform:uppercase;opacity:.66;font-weight:700}
/* ── dispatch (steps as airmail envelopes) ── */
.dispatch{padding:80px 0 40px}
.sect{display:flex;align-items:flex-end;gap:16px;border-bottom:2px solid var(--ink);padding-bottom:10px}
.sect h2{font-family:'Big Shoulders Display',sans-serif;font-weight:800;font-size:clamp(2rem,5vw,3.4rem);text-transform:uppercase;line-height:.95}
.sect span{margin-left:auto;font-size:10.5px;letter-spacing:.16em;text-transform:uppercase;color:var(--ink-soft);font-weight:700;padding-bottom:5px}
.letters{display:grid;grid-template-columns:repeat(auto-fit,minmax(400px,1fr));gap:22px;margin-top:30px}
.ltr{
background:var(--paper);border:2px solid var(--ink);position:relative;
padding:26px 26px 28px 40px;display:flex;flex-direction:column;
box-shadow:6px 7px 0 rgba(22,35,61,.12);transition:transform .2s,box-shadow .2s;
}
.ltr:hover{transform:translate(-3px,-3px);box-shadow:9px 10px 0 rgba(22,35,61,.16)}
/* chevron spine */
.ltr::before{
content:'';position:absolute;left:0;top:0;bottom:0;width:14px;
background:repeating-linear-gradient(-45deg,var(--red) 0 10px,var(--paper) 10px 20px,var(--blue) 20px 30px,var(--paper) 30px 40px);
border-right:1px solid var(--ink);
}
.ltr .mark{
position:absolute;right:20px;top:18px;width:56px;height:56px;border:2px solid var(--red);border-radius:50%;
display:grid;place-items:center;text-align:center;transform:rotate(-11deg);
font-size:8.5px;letter-spacing:.08em;text-transform:uppercase;color:var(--red);font-weight:700;line-height:1.25;
}
.ltr h3{font-family:'Big Shoulders Display',sans-serif;font-weight:800;font-size:1.9rem;text-transform:uppercase;line-height:1;padding-right:70px}
.ltr .flag{
display:inline-block;margin-left:9px;vertical-align:5px;font-family:'Karla',sans-serif;font-weight:700;
font-size:9px;letter-spacing:.11em;text-transform:uppercase;background:var(--blue);color:#fff;padding:3px 8px;
}
.ltr .flag.-r{background:var(--red)}
.ltr code{display:block;font-family:ui-monospace,Menlo,monospace;font-size:11.5px;color:var(--blue);margin-top:7px;font-weight:600}
.ltr p{margin-top:12px;color:var(--ink-soft);font-size:.94rem}
.ltr .pen{font-family:'Caveat',cursive;font-size:1.2rem;color:var(--red);margin-top:10px;transform:rotate(-.8deg)}
.ltr ul{list-style:none;margin-top:auto;padding-top:16px}
.ltr li{font-size:.85rem;padding:6px 0 6px 18px;position:relative;border-top:1px dotted rgba(22,35,61,.24)}
.ltr li::before{content:'✈';position:absolute;left:0;color:var(--blue);font-size:9px;top:9px}
/* ── two routes ── */
.routes{background:var(--paper);border-top:2px solid var(--ink);border-bottom:2px solid var(--ink);padding:76px 0;margin-top:56px}
.rc{display:grid;grid-template-columns:1fr 1fr;gap:24px;margin-top:28px}
.rc div{border:2px solid var(--ink);padding:26px 24px}
.rc .-a{background:rgba(211,58,56,.06)}
.rc .-b{background:rgba(36,85,164,.06)}
.rc h4{font-family:'Big Shoulders Display',sans-serif;font-weight:800;font-size:1.5rem;text-transform:uppercase}
.rc ul{list-style:none;margin-top:14px}
.rc li{padding:9px 0 9px 24px;position:relative;font-size:.95rem;border-top:1px solid rgba(22,35,61,.14);color:var(--ink-soft)}
.rc li:first-child{border-top:none}
.rc .-a li::before{content:'✕';position:absolute;left:0;color:var(--red);font-weight:700}
.rc .-b li::before{content:'✓';position:absolute;left:0;color:var(--blue);font-weight:700}
/* ── regulations ── */
.regs{padding:78px 0}
.rg{display:grid;grid-template-columns:repeat(auto-fit,minmax(230px,1fr));gap:0;margin-top:26px;border-top:2px solid var(--ink);border-left:2px solid var(--ink)}
.rg div{padding:22px 20px;border-right:2px solid var(--ink);border-bottom:2px solid var(--ink);background:var(--paper)}
.rg b{font-size:10px;letter-spacing:.15em;text-transform:uppercase;color:var(--red);font-weight:700}
.rg p{margin-top:9px;font-size:.94rem;color:var(--ink-soft)}
.rg p strong{color:var(--ink);font-weight:700}
/* ── install ── */
.install{padding:0 0 82px}
.term{margin-top:20px;background:var(--ink);color:#D6E2F2;padding:24px;
font-family:ui-monospace,Menlo,monospace;font-size:12.5px;line-height:2.05;overflow-x:auto}
.term .c{color:#7488A6}.term .k{color:#F0B06A}.term .p{color:#8FB6E8}
/* ── close ── */
.close{background:var(--ink);color:var(--paper);text-align:center;padding:92px 0}
.close h2{font-family:'Big Shoulders Display',sans-serif;font-weight:800;font-size:clamp(2.4rem,7vw,5.2rem);text-transform:uppercase;line-height:.9}
.close h2 em{font-style:normal;color:#8FB6E8}
.close .hand{display:block;margin-top:14px;font-size:1.5rem;color:#8FB6E8}
.close .btn{margin-top:24px;background:var(--red)}
.close .btn:hover{background:var(--paper);color:var(--ink)}
footer{background:var(--ink);color:rgba(251,250,246,.5);padding:0 0 28px;font-size:10px;letter-spacing:.15em;text-transform:uppercase;font-weight:700}
footer .wrap{display:flex;justify-content:space-between;gap:12px;flex-wrap:wrap;border-top:1px dashed rgba(251,250,246,.24);padding-top:20px}
@media(max-width:700px){:root{--fr:8px}}
@media(max-width:820px){
.wrap{padding:0 20px}
.env{padding:26px 22px 28px}
.addr{grid-template-columns:1fr;gap:22px}
.rc{grid-template-columns:1fr}
.letters{grid-template-columns:1fr}
}
@media(prefers-reduced-motion:reduce){*{transition:none!important}}
</style>
</head>
<body>
<a class="skip" href="#main">Skip to the content</a>
<main id="main">
<div class="frame-t" aria-hidden="true"></div>
<div class="frame-b" aria-hidden="true"></div>
<div class="frame-l" aria-hidden="true"></div>
<div class="frame-r" aria-hidden="true"></div>
<header class="top">
<div class="wrap">
<span class="brand">PLAN<i>2</i>CODE</span>
<span class="avion">Par Avion &middot; Via Air Mail</span>
</div>
</header>
<section class="hero">
<div class="wrap">
<div class="env">
<div class="pstamp"><div class="in"><b>2.0</b><span>Plan2Code</span><span>Postage</span></div></div>
<svg class="cancel" viewBox="0 0 210 110" aria-hidden="true">
<circle class="ring" cx="52" cy="55" r="42"/>
<text x="52" y="46" text-anchor="middle" font-size="10">SHIPPED</text>
<text x="52" y="62" text-anchor="middle" font-size="13" font-weight="700">2026</text>
<text x="52" y="76" text-anchor="middle" font-size="8">NO REWORK</text>
<path class="wv" d="M98,36 C118,28 138,44 158,36 C178,28 194,40 208,34"/>
<path class="wv" d="M98,50 C118,42 138,58 158,50 C178,42 194,54 208,48"/>
<path class="wv" d="M98,64 C118,56 138,72 158,64 C178,56 194,68 208,62"/>
<path class="wv" d="M98,78 C118,70 138,86 158,78 C178,70 194,82 208,76"/>
</svg>
<p class="kick">Spec-driven workflow for AI coding agents</p>
<h1>Send the plan.<br>The build <em>follows</em>.</h1>
<div class="addr">
<div>
<p class="to">Addressed to</p>
<p>Anyone who has watched an AI agent confidently build the wrong thing. You approve a plan, it becomes phase documents in your repo, and the agent works to those documents one phase at a time.</p>
</div>
<div class="acts">
<a class="btn" href="https://github.com/jparkerweb/plan2code">Install Plan2Code</a>
<a class="btn -o" href="#dispatch">Read the letters</a>
</div>
</div>
<section class="hero" aria-labelledby="hero-title">
<div class="narrow">
<svg class="planny" viewBox="0 0 22 30" aria-hidden="true" focusable="false">
<rect class="foot" x="5.7" y="27.1" width="3.6" height="1.7" rx="0.85"/>
<rect class="foot" x="12.7" y="27.1" width="3.6" height="1.7" rx="0.85"/>
<rect class="shell" x="3" y="7" width="16" height="20" rx="3.2"/>
<path class="eye" d="M11 10.95 L11.79 12.91 L13.9 13.06 L12.28 14.42 L12.79 16.47 L11 15.35 L9.21 16.47 L9.72 14.42 L8.1 13.06 L10.21 12.91 Z"/>
<path class="smile" d="M8.3 18.4 Q11 21 13.7 18.4"/>
<g class="arm"><path d="M3 16.5 L1.1 13.7"/><circle cx="1.1" cy="13.7" r="0.85"/></g>
<g class="arm"><path d="M19 16.5 L20.9 13.7"/><circle cx="20.9" cy="13.7" r="0.85"/></g>
</svg>
<h1 id="hero-title">Plan and build with your AI agent, on one page.</h1>
<p class="lede">Plan2Code asks the questions, drafts the plan and builds it one phase at a time. You answer and approve in your browser, and the specs land in your repo.</p>
<div class="cmd">
<pre id="install-cmd"><span aria-hidden="true">$ </span>npx --allow-git=all git+https://github.com/jparkerweb/plan2code.git</pre>
<button class="copy" type="button" aria-label="Copy the install command">
<svg class="clip" viewBox="0 0 24 24" aria-hidden="true"><rect x="9" y="9" width="11" height="12" rx="2"/><path d="M5 15H4a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h9a2 2 0 0 1 2 2v1"/></svg>
<svg class="tick" viewBox="0 0 24 24" aria-hidden="true"><path d="M4 12.5 9.5 18 20 6.5"/></svg>
<span class="label">Copy</span>
</button>
</div>
<p class="then">then run <code>plan2code</code> in any project.</p>
</div>
</section>
<div class="meter">
<div class="wrap">
<div><b>90%</b><span>Confidence gate</span></div>
<div><b>11</b><span>Review dimensions</span></div>
<div><b>1</b><span>Chat per phase</span></div>
<div><b>0</b><span>Context pollution</span></div>
</div>
<div class="stage">
<picture class="shot big">
<source media="(prefers-color-scheme: dark)" srcset="screenshots/dashboard-dark.webp">
<img src="screenshots/dashboard-light.webp" width="2560" height="1600" alt="The Plan2Code dashboard for a task API that is half built: a Set up card, then the steps in order, with the Implement card marked Suggested and reading Phase 3 of 4 is next.">
</picture>
</div>
<section class="dispatch wrap" id="dispatch">
<div class="sect"><h2>Six written letters</h2><span>Two optional</span></div>
<div class="letters">
<article class="ltr">
<span class="mark">Opt<br>00</span>
<h3>Pathfinder<span class="flag">New in 2.0</span></h3>
<code>/plan2code-0-pathfinder</code>
<p>For an idea that arrives too foggy to plan. It charts the idea as a map of decision questions and clears exactly one per session until nothing is left to decide.</p>
<p class="pen">it never answers its own questions — that's the point</p>
<ul>
<li>One decision per file, one file per session</li>
<li>Research questions fan out to parallel agents</li>
<li>Clears at ≥18/25 on four confidence dimensions</li>
<li>Hands a hot plan draft to Step 1</li>
</ul>
</article>
<article class="ltr">
<span class="mark">Req<br>01</span>
<h3>Plan</h3>
<code>/plan2code-1-plan</code>
<p>A real back-and-forth about what gets built. Requirements pulled apart, your codebase actually read, the stack confirmed rather than assumed.</p>
<ul>
<li>Functional and non-functional requirements</li>
<li>Integration points read from real source</li>
<li>Tech stack confirmed by explicit approval</li>
<li>Will not finalize below 90% confidence</li>
</ul>
</article>
<article class="ltr">
<span class="mark">Req<br>02</span>
<h3>Document</h3>
<code>/plan2code-2-document</code>
<p>The plan stops being a conversation and becomes files — an overview plus one self-contained phase document per phase, cut into one-story-point tasks.</p>
<ul>
<li>overview.md and Phase 1…N.md land in your repo</li>
<li>One-point tasks with real checkboxes</li>
<li>Parallel groups marked for concurrent agents</li>
<li>Each phase readable with no prior context</li>
</ul>
</article>
<article class="ltr">
<span class="mark">Req<br>03</span>
<h3>Implement</h3>
<code>/plan2code-3-implement</code>
<p>Build to the drawing, one phase per conversation. Planning context bleeding into implementation is where most agent drift starts.</p>
<p class="pen">reply “approved” — nothing advances on a guess</p>
<ul>
<li>Fresh conversation for every phase</li>
<li>Specs followed exactly as documented</li>
<li>Self-review against the spec before close</li>
<li>[/] claims prevent parallel-agent collisions</li>
</ul>
</article>
<article class="ltr">
<span class="mark">Opt<br>✳</span>
<h3>Review<span class="flag -r">Second opinion</span></h3>
<code>/plan2code-review</code>
<p>An independent senior engineer with fresh eyes and no attachment to the code that just shipped. It reads every changed file completely and reports only what it can prove — file, line, and a concrete fix.</p>
<ul>
<li>Scope adapts: files, branch diff, or subsystem</li>
<li>All 11 dimensions reported, clean ones justified</li>
<li>Ranked Critical / Warning / Suggestion</li>
<li>Read-only until you reply H, A, or S 1,3,5</li>
</ul>
</article>
<article class="ltr">
<span class="mark">Req<br>04</span>
<h3>Finalize</h3>
<code>/plan2code-4-finalize</code>
<p>Close it out. Every task audited against the spec it came from, a summary of what was built, docs checked for drift, and the folder filed away.</p>
<ul>
<li>Audit for tasks quietly left open</li>
<li>Summary of what was built and changed</li>
<li>README and other docs reconciled</li>
<li>Specs archived to specs--completed/</li>
</ul>
</article>
</div>
</section>
<section class="routes">
<section class="s" id="session" aria-labelledby="session-title">
<div class="wrap">
<div class="sect"><h2>Two routes</h2><span>Same destination</span></div>
<div class="rc">
<div class="-a">
<h4>Sent without an address</h4>
<ul>
<li>Coding starts, requirements arrive later</li>
<li>Design problems surface mid-build</li>
<li>Constant backtracking and rework</li>
<li>Context bloated with abandoned attempts</li>
<li>No way to prove requirements were met</li>
<li>Tokens burned on trial and error</li>
</ul>
<h2 id="session-title">How a session goes</h2>
<p class="intro muted">One page, from the first card to the sign-off. Your agent does the work in the terminal; you do your part here.</p>
<div class="row">
<picture class="shot">
<source media="(prefers-color-scheme: dark)" srcset="screenshots/question-dark.webp">
<img src="screenshots/question-light.webp" width="2560" height="1600" loading="lazy" alt="A Plan question card, How tasks are paged, with three options, the first marked Suggested, each with its trade-off underneath.">
</picture>
<div>
<span class="k" aria-hidden="true">1</span>
<h3>Pick a card, answer on the page</h3>
<p>Pick a step on the dashboard and it starts right there. Questions arrive a few at a time, each option with its trade-off written under it.</p>
</div>
<div class="-b">
<h4>Sent with a plan attached</h4>
<ul>
<li>The plan is approved before any code</li>
<li>Problems caught while they are still cheap</li>
<li>Implementation follows the drawing</li>
<li>Clean context, one phase at a time</li>
<li>Checkboxes prove what is done</li>
<li>Tokens spent building the agreed thing</li>
</ul>
</div>
<div class="row flip">
<picture class="shot">
<source media="(prefers-color-scheme: dark)" srcset="screenshots/build-dark.webp">
<img src="screenshots/build-light.webp" width="2560" height="1600" loading="lazy" alt="A phase of a task API being built: 5 of 11 done on the progress bar, and the phase's task list with the first five ticked.">
</picture>
<div>
<span class="k" aria-hidden="true">2</span>
<h3>Watch it build</h3>
<p>Tasks tick off as they land, and the line beside Planny says what is being worked on right now.</p>
</div>
</div>
<div class="row">
<picture class="shot">
<source media="(prefers-color-scheme: dark)" srcset="screenshots/signoff-dark.webp">
<img src="screenshots/signoff-light.webp" width="2560" height="1600" loading="lazy" alt="The Approve this phase card, with a button to read the completion report and three choices: Review the code first, Approve this phase, and I want changes.">
</picture>
<div>
<span class="k" aria-hidden="true">3</span>
<h3>Sign off</h3>
<p>Read the completion report, then approve the phase or ask for changes. In Implement, you can also have the code reviewed first.</p>
</div>
</div>
</div>
</section>
<section class="regs wrap">
<div class="sect"><h2>Postal regulations</h2><span>Four · non-negotiable</span></div>
<div class="rg">
<div><b>Reg 01</b><p>A <strong>fresh conversation</strong> for each step, and each implementation phase.</p></div>
<div><b>Reg 02</b><p>No code until the plan hits <strong>90% confidence</strong>. The gate does not negotiate.</p></div>
<div><b>Reg 03</b><p><strong>Checkboxes are the state</strong>, not the chat. Any session resumes cold.</p></div>
<div><b>Reg 04</b><p>Reply <strong>approved</strong> to close a phase. Nothing advances on a guess.</p></div>
</div>
</section>
<section class="install wrap">
<div class="sect"><h2>Installation</h2><span>Node 14+ · re-run to update</span></div>
<pre class="term"><span class="p">$</span> npx <span class="k">--allow-git=all</span> git+https://github.com/jparkerweb/plan2code.git</pre>
</section>
<section class="close">
<section class="s alt" id="pipeline" aria-labelledby="pipeline-title">
<div class="wrap">
<h2>Post it once.<br><em>Then go outside.</em></h2>
<span class="hand">free, MIT, twenty seconds to install</span>
<br><a class="btn" href="https://github.com/jparkerweb/plan2code">Get Plan2Code</a>
<h2 id="pipeline-title">The pipeline</h2>
<p class="intro muted">Every step is a card on the dashboard. Each one writes files the next one reads, so any session can pick up cold.</p>
<ol class="line">
<li class="opt"><h3>Pathfinder</h3><span class="chip">Step 0 · optional</span><p>A foggy idea becomes a clear list of decisions, settled one question at a time. <code>/plan2code-0-pathfinder</code></p></li>
<li><h3>Plan</h3><span class="chip">Step 1</span><p>Requirements, technical design and your approval: the shape of what gets built. <code>/plan2code-1-plan</code></p></li>
<li><h3>Document</h3><span class="chip">Step 2</span><p>The plan becomes a checklist of small tasks, grouped into phases, that a developer can just do. <code>/plan2code-2-document</code></p></li>
<li><h3>Implement</h3><span class="chip">Step 3 · once per phase</span><p>Builds the next phase, task by task, with a report and your sign-off at the end. <strong>Implement + review</strong> adds a careful review of the changes before the sign-off. <code>/plan2code-3-implement</code></p></li>
<li><h3>Finalize</h3><span class="chip">Step 4</span><p>Checks the work, writes the summary, files the specs away and suggests the commit message. <code>/plan2code-4-finalize</code></p></li>
</ol>
<div class="utils">
<div class="util"><span class="chip">Any time</span><h3>Quick task</h3><p>Too small for the full process? A quick plan, then it just builds it.</p></div>
<div class="util"><span class="chip">Any time</span><h3>Review</h3><p>A second opinion on the changes. Problems ranked, and you pick what gets fixed.</p></div>
<div class="util"><span class="chip">Any time</span><h3>Handoff</h3><p>Pack this conversation into a document a fresh session can pick up cold.</p></div>
</div>
</div>
</section>
<footer><div class="wrap"><span>Plan2Code v2.0.0</span><span>MIT</span><span>plan2code.jparkerweb.com</span></div></footer>
<section class="s" id="features" aria-labelledby="features-title">
<div class="wrap">
<h2 id="features-title">What the page gives you</h2>
<p class="intro muted">The same workflow as the terminal, with more room to think.</p>
<div class="feat">
<div><h3>Trade-offs in plain view</h3><p>Questions arrive a few at a time, each option with its trade-off written under it, so you choose rather than guess.</p></div>
<div><h3>Review before you approve</h3><p>Review the code first, on the sign-off card, reviews exactly the files the phase changed. You tick which findings get fixed, then approve.</p></div>
<div><h3>The Ask tab</h3><p>A chat beside the cards for anything you want to know. It is answered at the agent's next check-in and never holds up your answers.</p></div>
<div><h3>Attachments on notes</h3><p>Paste a screenshot or attach a document to a note on any question, and show instead of describe.</p></div>
<div><h3>Workspace folders</h3><p>Add a shared library or a design folder as read-only context, and point at it by its <code>@name</code>.</p></div>
<div><h3>Session meter</h3><p>A ring in the top bar turns from green to red as a session grows long, and suggests when a fresh one would serve you better.</p></div>
<div><h3>Runs on your machine</h3><p>A small program Plan2Code starts for you serves the page on <code>127.0.0.1</code>. No account, no dependencies, and nothing leaves your computer.</p></div>
</div>
<div class="feat-shots">
<figure>
<picture class="shot">
<source media="(prefers-color-scheme: dark)" srcset="screenshots/ask-dark.webp">
<img src="screenshots/ask-light.webp" width="2560" height="1600" loading="lazy" alt="The Ask tab: the question Where are tasks paged? and Plan2Code's answer naming the file.">
</picture>
<figcaption>The Ask tab, answered between tasks.</figcaption>
</figure>
<figure>
<picture class="shot">
<source media="(prefers-color-scheme: dark)" srcset="screenshots/finish-dark.webp">
<img src="screenshots/finish-light.webp" width="2560" height="1600" loading="lazy" alt="The finished screen after a phase: what happened, the next command with a Copy button, and a Back to the dashboard button.">
</picture>
<figcaption>Every ending says what to run next.</figcaption>
</figure>
</div>
</div>
</section>
<section class="s alt" id="rules" aria-labelledby="rules-title">
<div class="wrap">
<h2 id="rules-title">The rules</h2>
<p class="intro muted">Four rules do most of the work.</p>
<ol class="rules">
<li><div><h3>Keep each session short.</h3><p>Chaining a couple of small skills on the dashboard is fine. Start a fresh console session when the session meter turns red, and always before a big planning or build step. The specs on disk carry everything between sessions.</p></div></li>
<li><div><h3>No code until the plan hits 90% confidence.</h3><p>Step 1 will not finalize below the threshold. Under it, the agent keeps asking and keeps reading your code, and writes every assumption down where you can argue with it.</p></div></li>
<li><div><h3>Checkboxes are the state, not the chat.</h3><p>Progress lives in the spec files. Any agent, any session, resumes cold from them.</p></div></li>
<li><div><h3>Approve a phase to close it.</h3><p>On the sign-off card, or by replying <code>approved</code> in the terminal. Nothing advances on a guess about what you meant.</p></div></li>
</ol>
</div>
</section>
<section class="s" id="terminal" aria-labelledby="terminal-title">
<div class="wrap term-band">
<div>
<h2 id="terminal-title">Prefer the terminal?</h2>
<p class="muted">Every skill also runs right in the terminal. Each one asks at the start whether you want the page or the terminal; add <code>--web</code> to skip the question and go straight to the page.</p>
<p><a href="https://github.com/jparkerweb/plan2code#prefer-the-terminal">See every command in the README</a></p>
</div>
<pre class="term">/plan2code
/plan2code-1-plan --web plan a REST API for tasks</pre>
</div>
</section>
<section class="s alt close" id="install" aria-labelledby="install-title">
<div class="narrow">
<h2 id="install-title">Install</h2>
<p class="muted">Needs Node.js 18 or later. Run it again any time to update.</p>
<div class="cmd">
<pre><span aria-hidden="true">$ </span>npx --allow-git=all git+https://github.com/jparkerweb/plan2code.git</pre>
<button class="copy" type="button" aria-label="Copy the install command">
<svg class="clip" viewBox="0 0 24 24" aria-hidden="true"><rect x="9" y="9" width="11" height="12" rx="2"/><path d="M5 15H4a2 2 0 0 1-2-2V4a2 2 0 0 1 2-2h9a2 2 0 0 1 2 2v1"/></svg>
<svg class="tick" viewBox="0 0 24 24" aria-hidden="true"><path d="M4 12.5 9.5 18 20 6.5"/></svg>
<span class="label">Copy</span>
</button>
</div>
<p class="then">then run <code>plan2code</code> in any project.</p>
</div>
</section>
</main>
<footer>
<div class="wrap">
<a href="https://github.com/jparkerweb/plan2code">Plan2Code on GitHub</a>
<span>MIT license</span>
<span>Version 2.4.1</span>
</div>
</footer>
<script>
(function () {
var CMD = 'npx --allow-git=all git+https://github.com/jparkerweb/plan2code.git';
function wire(btn) {
var label = btn.querySelector('.label');
var timer;
function done() {
btn.classList.add('-done');
label.textContent = 'Copied';
clearTimeout(timer);
timer = setTimeout(function () {
btn.classList.remove('-done');
label.textContent = 'Copy';
}, 1800);
}
function fallback() {
var ta = document.createElement('textarea');
ta.value = CMD;
ta.setAttribute('readonly', '');
ta.style.position = 'fixed';
ta.style.opacity = '0';
document.body.appendChild(ta);
ta.select();
try { document.execCommand('copy'); done(); } catch (e) {}
document.body.removeChild(ta);
}
btn.addEventListener('click', function () {
if (navigator.clipboard && navigator.clipboard.writeText) {
navigator.clipboard.writeText(CMD).then(done, fallback);
} else {
fallback();
}
});
}
var buttons = document.querySelectorAll('.copy');
for (var i = 0; i < buttons.length; i++) wire(buttons[i]);
})();
</script>
</body>
</html>
Binary file not shown.

After

Width:  |  Height:  |  Size: 95 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 104 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 106 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 109 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 80 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 86 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 109 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 115 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 96 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 102 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 103 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 113 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 86 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 95 KiB

+1031 -997
View File
File diff suppressed because it is too large. Load diff
+4 -3
View File
@@ -1,13 +1,14 @@
{
"name": "plan2code",
"version": "2.0.0",
"version": "2.4.2",
"private": true,
"bin": {
"plan2code": "./install.js"
"plan2code-install": "./install.js"
},
"scripts": {
"prepare": "husky",
"test": "node scripts/validate-char-count.js"
"build:skills": "node install.js --build-skills",
"test": "node scripts/validate-char-count.js && node install.js --verify-skills && node scripts/parse-check.mjs && node scripts/check-skill-table.mjs && node --test scripts/test-web-console.mjs scripts/test-launcher.mjs scripts/test-skill-scripts.mjs scripts/test-script-pathfinder.mjs scripts/test-script-finalize.mjs scripts/test-script-handoff.mjs"
},
"devDependencies": {
"husky": "^9.0.0"
+9 -9
View File
@@ -6,25 +6,25 @@ export const MASCOT = {
// Full mascot for headers
full: [
' ╭───╮ ',
' │ ● │ ',
' ╱│ ★ │╲ ',
' │ ◡ │ ',
' ╰───╯ ',
' ╰┬─┬╯ ',
],
// Mini mascot for inline use
mini: '(◉‿◉)',
mini: '╱[★]╲',
// Waving mascot for greetings
wave: [
' ╭───╮ ',
' │ ● │ ',
' │ ◡ │ ',
' ╰───╯ ',
' │ ★ │╱ ',
' ╱│ ◡ │ ',
' ╰┬─┬╯ ',
],
// Celebration mascot for completion
celebrate: [
' ╭───╮ ',
' │ ★ │ ',
' ╲│ ★ │╱ ',
' │ ◡ │ ',
' ╰───╯ ',
' ╰┬─┬╯ ',
],
};
@@ -117,7 +117,7 @@ export const logger = {
console.log(chalk.green.bold('═'.repeat(50)));
console.log();
console.log(chalk.cyan.bold('Next Step:'));
console.log(chalk.white(' Return to your AI Agent and run the'), chalk.yellow.bold('/plan2code-4-finalize'), chalk.white('step.'));
console.log(chalk.white(' Return to your AI Agent and run the'), chalk.yellow.bold('/plan2code-4-finalize --web'), chalk.white('step.'));
console.log(chalk.dim(' This will ensure quality, completeness, and proper documentation.'));
console.log();
},
+1 -1
View File
@@ -226,7 +226,7 @@ Output: `.plan2code-metrics/proposals/<timestamp>-diagnosis.md`
Sends the diagnosis plus current prompt file contents to an AI model. The AI proposes surgical `old_text -> new_text` edits, each validated against:
- The `old_text` actually exists in the target file
- The edited file stays under the 11,000 character limit
- The edited file stays under the 20,000 character limit
Output: `.plan2code-metrics/proposals/prop-<timestamp>.json`
+1 -1
View File
@@ -11,7 +11,7 @@ import chalk from 'chalk';
import type { PromptEdit, PromptProposal } from './types.js';
import { validateEdit } from './improver.js';
const CHAR_LIMIT = 11_000;
const CHAR_LIMIT = 20_000;
// ── Diff display ──────────────────────────────────────────────────────────────
+3 -3
View File
@@ -69,12 +69,12 @@ describe('validateEdit', () => {
expect(result.warnings.some(w => /appears.*times/i.test(w))).toBe(true);
});
it('errors when edit would exceed 11,000 char limit', () => {
const bigText = 'x'.repeat(12_000);
it('errors when edit would exceed 20,000 char limit', () => {
const bigText = 'x'.repeat(21_000);
const edit = makeEdit({ new_text: bigText });
const result = validateEdit(edit, PROMPT_CONTENTS);
expect(result.valid).toBe(false);
expect(result.errors.some(e => /exceed.*11.?000/i.test(e))).toBe(true);
expect(result.errors.some(e => /exceed.*20.?000/i.test(e))).toBe(true);
});
it('warns when reported char counts diverge from actual (>10 chars off)', () => {
+1 -1
View File
@@ -9,7 +9,7 @@ import path from 'path';
import type { PromptEdit, PromptProposal } from './types.js';
import { invokeLLM, type AgentType } from './invoke-llm.js';
const CHAR_LIMIT = 11_000;
const CHAR_LIMIT = 20_000;
const IMPROVE_PROMPT_PATH = new URL('../src/prompts/improve.md', import.meta.url).pathname
.replace(/^\/([A-Za-z]:)/, '$1'); // Fix Windows path
+2 -2
View File
@@ -26,7 +26,7 @@ The following diagnosis was produced by the analysis step:
## Hard Constraints — ALL must be satisfied:
1. **Char limit:** Each target file MUST stay under 11,000 characters after your edit is applied. This is enforced by code — edits that violate it will be automatically rejected.
1. **Char limit:** Each target file MUST stay under 20,000 characters after your edit is applied. This is enforced by code — edits that violate it will be automatically rejected.
2. **Maximum 5 edits per cycle.** Focus on the highest-impact changes only.
3. **Each edit must cite a specific metric** in its `expected_metric_impact` field (e.g., "avg_task_completion_rate", "avg_blocker_count").
4. **`old_text` must be verbatim** from the file. Copy-paste exactly — include surrounding whitespace/newlines as they appear. Edits with mismatched old_text will be automatically rejected.
@@ -85,7 +85,7 @@ Set `char_count_before`, `char_count_after`, and `char_count_delta` to your best
A table with columns: File | Before | Projected After | Delta | Limit | Status (✓/✗)
Verify that NO file exceeds 11,000 characters after your proposed edits.
Verify that NO file exceeds 20,000 characters after your proposed edits.
---
+322
View File
@@ -0,0 +1,322 @@
#!/usr/bin/env node
// Start-up benchmark for the web console. Dev-only: lives in scripts/ so it
// never ships inside a skill.
//
// node scripts/bench-web-console.mjs [--runs 7] [--workflow dashboard]
// [--browser <path> | --browser none] [--cwd <repo>] [--json <out.json>]
//
// Every run starts from nothing, the way a person's first `/plan2code` does:
// no server, a fresh browser context (each session gets a new port, so a new
// origin, so the HTTP cache is always cold). It measures, from the moment the
// agent runs `console.mjs open`:
//
// url the CLI has printed the link (what the agent hands over)
// cliExit the CLI process has exited (when the agent's next step can start)
// response the browser has the page's HTML
// fcp first contentful paint: the person sees something
// ready the session is drawn from real state (the menu, or a question)
//
// plus the bytes and request count the page cost. Needs puppeteer-core in
// node_modules and a local Chrome or Edge for the browser columns; without
// them it still reports url/cliExit and an HTTP-only waterfall.
import fs from "node:fs";
import os from "node:os";
import path from "node:path";
import { spawn, spawnSync } from "node:child_process";
import { fileURLToPath } from "node:url";
const ROOT = path.dirname(path.dirname(fileURLToPath(import.meta.url)));
const CONSOLE_CLI = path.join(ROOT, "src", "web-console", "console.mjs");
const args = {};
for (let i = 2; i < process.argv.length; i++) {
const a = process.argv[i];
if (a.startsWith("--")) args[a.slice(2)] = process.argv[i + 1] && !process.argv[i + 1].startsWith("--") ? process.argv[++i] : true;
}
const RUNS = Number(args.runs || 7);
const WORKFLOW = args.workflow || "dashboard";
const CWD = path.resolve(args.cwd || ROOT);
const HOME = fs.mkdtempSync(path.join(os.tmpdir(), "p2c-bench-"));
const ENV = { ...process.env, PLAN2CODE_CONSOLE_HOME: HOME, PLAN2CODE_NO_BROWSER: "1" };
// A fresh cached failure: the boot-time release check stays off the network,
// so a slow GitHub never colours the timings.
fs.writeFileSync(path.join(HOME, "update-check.json"), JSON.stringify({ checkedAt: Date.now(), latest: null }));
function findBrowser() {
if (args.browser === "none") return null;
if (typeof args.browser === "string") return args.browser;
const candidates = [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
"C:\\Program Files (x86)\\Microsoft\\Edge\\Application\\msedge.exe",
"/Applications/Google Chrome.app/Contents/MacOS/Google Chrome",
"/usr/bin/google-chrome",
"/usr/bin/chromium",
"/usr/bin/chromium-browser",
];
return candidates.find((p) => fs.existsSync(p)) || null;
}
// Resolves at the first stdout line carrying a url, and again at exit.
// --launch hook (default): `open` launches "the browser" the way it would for
// a person, through PLAN2CODE_BROWSER, and the hook just writes the link to
// a file; the bench points its ready-made tab there the instant it appears.
// That measures everything but the real browser's own start-up, which is
// about 900ms on Windows and runs alongside the rest.
// --launch url: navigate only once `open` prints the link. For comparing
// against versions of console.mjs that predate the hook.
const LAUNCH = args.launch || "hook";
// --browser-delay <ms>: how long "the browser" takes to show up after it is
// asked. 0 is a browser that is instant; ~900 is Chrome on Windows through
// `start`, measured on this machine, and is what a person actually waits.
const BROWSER_DELAY = Number(args["browser-delay"] || 0);
const OPEN_LOG = path.join(HOME, "opened.txt");
function runOpen() {
fs.rmSync(OPEN_LOG, { force: true });
const hook = LAUNCH === "hook";
const env = hook
? { ...ENV, PLAN2CODE_NO_BROWSER: "", PLAN2CODE_BROWSER: `echo>>"${OPEN_LOG}"` }
: ENV;
const t0 = Date.now();
const child = spawn(
process.execPath,
[CONSOLE_CLI, "open", "--workflow", WORKFLOW, "--title", "Bench", ...(hook ? [] : ["--no-open"])],
{ cwd: CWD, env, stdio: ["ignore", "pipe", "pipe"] }
);
let out = "";
let err = "";
let urlResolve;
const urlP = new Promise((r) => (urlResolve = r));
child.stdout.on("data", (c) => {
out += c;
const m = out.match(/"url":"([^"]+)"/);
if (m) urlResolve({ url: m[1], at: Date.now() });
});
child.stderr.on("data", (c) => (err += c));
const exitP = new Promise((r) =>
child.on("exit", (code) => {
urlResolve({ url: null, at: Date.now() });
r({ code, at: Date.now(), out, err });
})
);
// The moment "the browser" was asked to open, and with what.
const openP = hook
? new Promise((resolve) => {
const poll = setInterval(() => {
let text = "";
try {
text = fs.readFileSync(OPEN_LOG, "utf8");
} catch {}
const m = text.match(/http:\/\/127\.0\.0\.1:\d+\/s\/[\w-]+\//);
if (m) {
clearInterval(poll);
resolve({ url: m[0], at: Date.now() });
}
}, 2);
exitP.then(() => setTimeout(() => (clearInterval(poll), resolve({ url: null, at: Date.now() })), 500));
})
: urlP;
return { t0, urlP, exitP, openP };
}
function stopAll() {
spawnSync(process.execPath, [CONSOLE_CLI, "stop", "--all"], { cwd: CWD, env: ENV });
}
async function httpWaterfall(url) {
// What the page costs without a browser: the redirect, the HTML, every
// asset it names, then the modules those import. Sequential levels, like a
// browser without preload hints would discover them.
const t = Date.now();
const r1 = await fetch(url, { redirect: "manual" });
const cookie = (r1.headers.get("set-cookie") || "").split(";")[0];
const base = new URL(url).origin;
const get = async (p) => {
const res = await fetch(base + p, { headers: { cookie } });
const body = Buffer.from(await res.arrayBuffer());
return { p, status: res.status, bytes: body.length, text: body.toString("utf8") };
};
const html = await get("/");
let bytes = html.bytes;
let requests = 2;
const seen = new Set();
let level = [...html.text.matchAll(/(?:href|src)="(\/[^"#]+\.(?:css|js))"/g)].map((m) => m[1]);
let levels = 1;
while (level.length) {
const fresh = level.filter((p) => !seen.has(p));
fresh.forEach((p) => seen.add(p));
if (!fresh.length) break;
const got = await Promise.all(fresh.map(get));
requests += got.length;
bytes += got.reduce((n, g) => n + g.bytes, 0);
level = got
.filter((g) => g.p.endsWith(".js"))
.flatMap((g) => [...g.text.matchAll(/from\s+"\.\/([^"]+)"/g)].map((m) => "/" + path.posix.join(path.posix.dirname(g.p.slice(1)), m[1])));
levels++;
}
return { ms: Date.now() - t, bytes, requests, levels };
}
// --roundtrip: the two hops of a live session, rather than the start.
// send the person presses Send -> the agent's `wait` returns the answers
// post the agent runs `post` -> the page's stream delivers the new state
async function roundtrip() {
const payload = path.join(HOME, "rt.json");
const item = { id: "q1", kind: "text", title: "Anything", body: "Say something." };
fs.writeFileSync(payload, JSON.stringify({ workflow: "pathfinder", items: [item], agent: { status: "waiting" } }));
const out = spawnSync(process.execPath, [CONSOLE_CLI, "open", "--file", payload, "--no-open"], { cwd: CWD, env: ENV, encoding: "utf8" });
const opened = JSON.parse(out.stdout.trim().split("\n").pop());
const [, base, token] = opened.url.match(/^(http:\/\/127\.0\.0\.1:\d+)\/s\/([^/]+)\//);
const headers = { cookie: `p2c_console_${new URL(base).port}=${token}`, "content-type": "application/json" };
const sends = [];
const posts = [];
for (let i = -1; i < RUNS; i++) {
// send: a waiter already sitting in its slice, then the press.
const waiter = spawn(process.execPath, [CONSOLE_CLI, "wait", "--session", opened.sid, "--seconds", "30"], { cwd: CWD, env: ENV });
const done = new Promise((r) => waiter.on("exit", () => r(Date.now())));
await new Promise((r) => setTimeout(r, 700));
const t = Date.now();
await fetch(base + "/submit", { method: "POST", headers, body: JSON.stringify({ actions: [{ i: "q1", type: "answer", kind: "text", text: "x" + i }], reply: "x" }) });
const sendMs = (await done) - t;
// post: a page listening on the stream, then the agent's update.
const ctrl = new AbortController();
const res = await fetch(base + "/events", { headers, signal: ctrl.signal });
const reader = res.body.getReader();
const dec = new TextDecoder();
await reader.read();
const patch = path.join(HOME, "rt-patch.json");
fs.writeFileSync(patch, JSON.stringify({ items: [{ id: "q1", status: "open", body: "Again " + i }], agent: { status: "waiting" } }));
const tp = Date.now();
spawnSync(process.execPath, [CONSOLE_CLI, "post", "--session", opened.sid, "--file", patch], { cwd: CWD, env: ENV });
const postCli = Date.now() - tp;
let buf = "";
while (!buf.includes("Again " + i)) buf += dec.decode((await reader.read()).value);
const postMs = Date.now() - tp;
ctrl.abort();
if (i >= 0) {
sends.push(sendMs);
posts.push({ total: postMs, cli: postCli });
}
}
spawnSync(process.execPath, [CONSOLE_CLI, "stop", "--all"], { cwd: CWD, env: ENV });
const med = (v) => v.sort((a, b) => a - b)[Math.floor(v.length / 2)];
console.log(`web console round trips, ${RUNS} runs`);
console.log(`send -> wait returns median ${med(sends)} ms (min ${Math.min(...sends)}, max ${Math.max(...sends)})`);
console.log(`post -> page has it median ${med(posts.map((p) => p.total))} ms (of which the post CLI ${med(posts.map((p) => p.cli))} ms)`);
fs.rmSync(HOME, { recursive: true, force: true });
}
async function main() {
if (args.roundtrip) return roundtrip();
const browserPath = findBrowser();
let puppeteer = null;
if (browserPath) {
try {
puppeteer = (await import("puppeteer-core")).default;
} catch {
console.error("puppeteer-core not installed; browser columns skipped");
}
}
const browser = puppeteer
? await puppeteer.launch({ executablePath: browserPath, headless: true, args: ["--no-first-run", "--no-default-browser-check"] })
: null;
const rows = [];
// One throwaway run so disk caches and the Node binary are warm; the person
// running this has run node before too.
for (let i = -1; i < RUNS; i++) {
stopAll();
// The tab exists before the clock starts: a person's browser is already
// running, and creating a context is the harness's cost, not the page's.
const ctx = browser ? await browser.createBrowserContext() : null;
const page = ctx ? await ctx.newPage() : null;
const { t0, urlP, exitP, openP } = runOpen();
const o = await openP;
if (!o.url) {
const e = await exitP;
throw new Error(`open failed (${e.code}): ${e.err}`);
}
const row = { open: o.at - t0 };
if (browser) {
let bytes = 0;
let requests = 0;
page.on("response", async (res) => {
requests++;
try {
bytes += (await res.buffer()).length;
} catch {}
});
if (BROWSER_DELAY) await new Promise((r) => setTimeout(r, Math.max(0, o.at + BROWSER_DELAY - Date.now())));
await page.goto(o.url, { waitUntil: "domcontentloaded" });
const timing = await page.evaluate(async () => {
// "ready" = drawn from real state: a dashboard card, a question card,
// the starting screen or the waiting pane a real state produces.
const readySel = ".dash-card, .card.question, .waiting-pane, .launching.starting, .handoff";
await new Promise((resolve) => {
const check = () => (document.querySelector(readySel) ? resolve() : requestAnimationFrame(check));
check();
});
const readyAt = performance.timeOrigin + performance.now();
// The real screen can be drawn before the first paint happens at all,
// so wait for the paint rather than reading an entry not yet made.
const fcp = await new Promise((resolve) => {
const got = () => performance.getEntriesByName("first-contentful-paint")[0];
if (got()) return resolve(got());
new PerformanceObserver(() => got() && resolve(got())).observe({ type: "paint", buffered: true });
setTimeout(() => resolve(got() || null), 3000);
});
const nav = performance.getEntriesByType("navigation")[0];
return {
origin: performance.timeOrigin,
fcp: fcp ? performance.timeOrigin + fcp.startTime : null,
response: nav ? performance.timeOrigin + nav.responseEnd : null,
ready: readyAt,
};
});
// The origin here is the redirected document, so the 302 is inside
// `response`, not before it.
row.response = Math.round(timing.response - t0);
row.fcp = timing.fcp ? Math.round(timing.fcp - t0) : null;
row.ready = Math.round(timing.ready - t0);
await new Promise((r) => setTimeout(r, 150));
row.requests = requests;
row.kb = Math.round(bytes / 1024);
await ctx.close();
}
const e = await exitP;
row.url = (await urlP).at - t0;
row.cliExit = e.at - t0;
if (!browser) Object.assign(row, await httpWaterfall(o.url));
if (i >= 0) rows.push(row);
}
stopAll();
if (browser) await browser.close();
const cols = Object.keys(rows[0]);
const stat = (k) => {
const v = rows.map((r) => r[k]).filter((x) => typeof x === "number").sort((a, b) => a - b);
if (!v.length) return null;
return { median: v[Math.floor(v.length / 2)], min: v[0], max: v[v.length - 1] };
};
const summary = Object.fromEntries(cols.map((k) => [k, stat(k)]));
console.log(`web console start-up, ${RUNS} runs, workflow=${WORKFLOW}, cwd=${CWD}`);
console.log(`browser: ${browser ? browserPath : "none (HTTP-only waterfall)"}`);
console.log("metric".padEnd(10) + "median".padStart(9) + "min".padStart(9) + "max".padStart(9));
for (const k of cols) {
const s = summary[k];
if (!s) continue;
const unit = k === "kb" ? " KB" : k === "requests" || k === "levels" ? "" : " ms";
console.log(k.padEnd(10) + `${s.median}${unit}`.padStart(9) + `${s.min}`.padStart(9) + `${s.max}`.padStart(9));
}
if (args.json) fs.writeFileSync(path.resolve(args.json), JSON.stringify({ runs: rows, summary }, null, 2));
fs.rmSync(HOME, { recursive: true, force: true });
}
main().catch((e) => {
stopAll();
console.error(e);
process.exit(1);
});
+368
View File
@@ -0,0 +1,368 @@
#!/usr/bin/env node
// Screenshots of the web console for the docs. Dev-only: lives in scripts/ so
// it never ships inside a skill.
//
// node scripts/capture-screenshots.mjs [--only <shot>] [--browser <path>]
// [--out <dir>]
//
// Every shot is staged with no agent: a throwaway sample project called
// `task-api`, a throwaway console home (PLAN2CODE_CONSOLE_HOME), then
// `console.mjs open` and `post` with the payloads in scripts/screenshots/.
// Each screen is saved light and dark, at 1280x800 and twice the pixel
// density, as `<out>/<shot>-light.webp` and `<out>/<shot>-dark.webp`.
// The shots use the console's Ocean highlight rather than its default orange:
// it is the console's nearest to the Plan2Code blue the site and README use.
//
// Shots: dashboard, question, doc, build, signoff, finish, ask.
//
// Needs puppeteer-core (`npm i --no-save puppeteer-core`, never added to
// package.json) and a local Chrome or Edge. Writes nothing outside its temp
// folders and --out.
import fs from "node:fs";
import os from "node:os";
import path from "node:path";
import { spawnSync } from "node:child_process";
import { fileURLToPath } from "node:url";
const ROOT = path.dirname(path.dirname(fileURLToPath(import.meta.url)));
const CONSOLE_CLI = path.join(ROOT, "src", "web-console", "console.mjs");
const PAYLOADS = path.join(ROOT, "scripts", "screenshots");
const args = {};
for (let i = 2; i < process.argv.length; i++) {
const a = process.argv[i];
if (a.startsWith("--")) args[a.slice(2)] = process.argv[i + 1] && !process.argv[i + 1].startsWith("--") ? process.argv[++i] : true;
}
const OUT = path.resolve(args.out || path.join(ROOT, "docs", "screenshots"));
const QUALITY = 85;
const WARN_KB = 300;
const THEMES = ["light", "dark"];
// Which workflow each shot opens under, which payloads it posts in order,
// which tab it is captured on (by the tab's label; none means the first tab),
// and where to scroll first: `scrollTo` is put at the top of the view and
// `mustShow` has to end up wholly inside it (none means the top).
const SHOTS = {
// The suggested card sits below the fold at 1280x800, and it is the point of
// the shot. Starting at the "Hide unavailable" switch keeps the spec picker's
// cut out of frame and the card in it.
dashboard: {
workflow: "dashboard",
posts: ["dashboard"],
scrollTo: ".dash-toggle",
mustShow: ".dash-card:has(.pill.suggested)",
},
question: { workflow: "plan", posts: ["question"] },
doc: { workflow: "plan", posts: ["doc"], tab: "What we've agreed" },
build: { workflow: "implement", posts: ["build"], tab: "Phase 3 tasks" },
signoff: { workflow: "implement", posts: ["signoff"] },
finish: { workflow: "implement", posts: ["signoff", "finish"] },
ask: { workflow: "plan", posts: ["question"], ask: "Where are tasks paged?" },
};
// On Windows the temp dir sits under the user's profile, and the page can
// show the project's path, so the sample project goes at the drive root
// instead. Elsewhere the temp dir carries no username. Each run makes its own
// folder there (p2c-shots-XXXXXX), so it never deletes anything it did not
// create and two runs never share one.
const SCRATCH_ROOT = process.platform === "win32" ? path.parse(os.tmpdir()).root : os.tmpdir();
let SCRATCH = null;
let HOME = null;
let ENV = null;
const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
function fail(message) {
console.error(message);
process.exit(1);
}
function findBrowser() {
if (typeof args.browser === "string") return fs.existsSync(args.browser) ? args.browser : null;
const candidates = [
"C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
"C:\\Program Files (x86)\\Google\\Chrome\\Application\\chrome.exe",
"C:\\Program Files (x86)\\Microsoft\\Edge\\Application\\msedge.exe",
"/Applications/Google Chrome.app/Contents/MacOS/Google Chrome",
"/Applications/Microsoft Edge.app/Contents/MacOS/Microsoft Edge",
"/usr/bin/google-chrome",
"/usr/bin/chromium",
"/usr/bin/chromium-browser",
];
return candidates.find((p) => fs.existsSync(p)) || null;
}
function phaseFile(n, name, tasks, done) {
const lines = tasks.map((t, i) => `- [${done ? "x" : " "}] **Task ${n}.${i + 1}:** ${t}`).join("\n");
return `# Phase ${n}: ${name}\n\n**Status:** ${done ? "Complete" : "Not Started"}\n\n## Tasks\n\n${lines}\n`;
}
// A neutral project the console scans as mid-build: Phases 1 and 2 done,
// Phases 3 and 4 still to do.
function makeSampleProject() {
const project = path.join(SCRATCH, "task-api");
const spec = path.join(project, "specs", "task-api");
fs.mkdirSync(spec, { recursive: true });
fs.writeFileSync(
path.join(project, "AGENTS.md"),
"# AGENTS.md\n\nA REST API for a team's tasks, on Fastify, Postgres and Zod.\n\n## Development Commands\n\n- `npm start` runs the API\n- `npm test` runs the tests\n"
);
fs.writeFileSync(
path.join(spec, "overview.md"),
[
"# Task API",
"",
"**Status:** In Progress",
"",
"## Summary",
"",
"A REST API for a team's tasks: create, list, update and delete them, with filters by status and assignee and cursor paging.",
"",
"## Phase Checklist",
"",
"- [x] Phase 1: Project setup ([phase-1.md](./phase-1.md))",
"- [x] Phase 2: Data model ([phase-2.md](./phase-2.md))",
"- [ ] Phase 3: API endpoints ([phase-3.md](./phase-3.md))",
"- [ ] Phase 4: Authentication ([phase-4.md](./phase-4.md))",
"",
].join("\n")
);
// The file the Ask shot's answer points at.
fs.mkdirSync(path.join(project, "src", "tasks"), { recursive: true });
fs.writeFileSync(
path.join(project, "src", "tasks", "paginate.js"),
[
"// Cursor paging for GET /tasks: up to 50 tasks a page, oldest first.",
"export function paginate(tasks, cursor, limit = 50) {",
" const start = cursor ? tasks.findIndex((t) => t.id === cursor) + 1 : 0;",
" const page = tasks.slice(start, start + limit);",
" return { tasks: page, nextCursor: page.length === limit ? page[page.length - 1].id : null };",
"}",
"",
].join("\n")
);
fs.writeFileSync(path.join(spec, "phase-1.md"), phaseFile(1, "Project setup", ["Scaffold the Fastify app", "Add the Postgres connection", "Add the test runner"], true));
fs.writeFileSync(
path.join(spec, "phase-2.md"),
phaseFile(2, "Data model", ["Add the tasks table and its migration", "Add the task model", "Add status and assignee indexes", "Seed sample tasks"], true)
);
fs.writeFileSync(
path.join(spec, "phase-3.md"),
phaseFile(3, "API endpoints", ["Add GET /tasks with cursor paging", "Add POST /tasks", "Add PATCH /tasks/:id", "Add DELETE /tasks/:id"], false)
);
fs.writeFileSync(
path.join(spec, "phase-4.md"),
phaseFile(4, "Authentication", ["Add API keys", "Require a key on every route", "Scope tasks to the key's team"], false)
);
return project;
}
function cli(cwd, ...cliArgs) {
const r = spawnSync(process.execPath, [CONSOLE_CLI, ...cliArgs], { cwd, env: ENV, encoding: "utf8" });
const line = (r.stdout || "").trim().split("\n").pop() || "";
let out = null;
try {
out = JSON.parse(line);
} catch {}
return { code: r.status, out, err: r.stderr };
}
function post(project, sid, name) {
const r = cli(project, "post", "--session", sid, "--file", path.join(PAYLOADS, name + ".json"));
if (r.code !== 0) throw new Error(`post ${name}.json failed (${r.code}): ${r.err.trim()}`);
}
// Opens a session for the shot and posts its payloads. Returns { sid, url }.
function stage(project, shot) {
const spec = SHOTS[shot];
const r = cli(project, "open", "--workflow", spec.workflow, "--no-open");
if (r.code !== 0 || !r.out || !r.out.url) throw new Error(`open failed (${r.code}): ${r.err.trim()}`);
for (const name of spec.posts) post(project, r.out.sid, name);
return { sid: r.out.sid, url: r.out.url };
}
// Waits for the page to draw real state, then lets the animations finish
// (the dashboard's wake-up is the longest).
async function settle(page) {
await page.waitForNetworkIdle({ idleTime: 500, timeout: 15000 }).catch(() => {});
await page.waitForSelector(".dash-card, .card, .handoff, .ask-pane, .doc", { timeout: 15000 }).catch(() => {});
await sleep(3500);
}
async function openTab(page, label) {
await page.evaluate((text) => {
const tab = [...document.querySelectorAll("#views .view-tab")].find((b) => b.firstChild && b.firstChild.textContent.trim() === text);
if (tab) tab.click();
}, label);
await sleep(600);
}
async function scrollTo(page, { scrollTo: top, mustShow }) {
const problem = await page.evaluate(
(topSel, showSel) => {
const node = document.querySelector(topSel);
if (!node) return `nothing matched ${topSel}, captured from the top`;
node.scrollIntoView({ block: "start", behavior: "instant" });
if (!showSel) return null;
const show = document.querySelector(showSel);
if (!show) return `nothing matched ${showSel}`;
const r = show.getBoundingClientRect();
return r.top >= 0 && r.bottom <= window.innerHeight ? null : `${showSel} is not wholly in view`;
},
top,
mustShow || null
);
if (problem) console.log(` warning: ${problem}`);
await sleep(300);
}
async function shoot(page, name) {
const written = [];
for (const theme of THEMES) {
await page.emulateMediaFeatures([{ name: "prefers-color-scheme", value: theme }]);
// The theme is applied at load, so reload rather than trust a live switch.
await page.reload({ waitUntil: "domcontentloaded" });
await settle(page);
if (page.__tab) await openTab(page, page.__tab);
if (page.__scroll && page.__scroll.scrollTo) await scrollTo(page, page.__scroll);
const file = path.join(OUT, `${name}-${theme}.webp`);
await page.screenshot({ type: "webp", quality: QUALITY, path: file });
const kb = Math.round(fs.statSync(file).size / 1024);
console.log(`${path.basename(file)} ${kb} KB`);
if (kb > WARN_KB) console.log(` warning: ${path.basename(file)} is over ${WARN_KB} KB`);
written.push(file);
}
return written;
}
// Opens the page for a staged session and saves both themes.
async function capture(browser, url, name, { tab, scrollTo: top, mustShow }) {
const ctx = await browser.createBrowserContext();
try {
const page = await ctx.newPage();
await page.setViewport({ width: 1280, height: 800, deviceScaleFactor: 2 });
await page.goto(url, { waitUntil: "domcontentloaded" });
await settle(page);
page.__tab = tab || null;
page.__scroll = { scrollTo: top, mustShow };
return await shoot(page, name);
} finally {
await ctx.close();
}
}
// The Ask shot: a question typed and sent on the page, collected with `chat`
// the way an agent would, and answered with a posted reply.
async function captureAsk(browser, project, sid, url, question) {
const ctx = await browser.createBrowserContext();
try {
const page = await ctx.newPage();
await page.setViewport({ width: 1280, height: 800, deviceScaleFactor: 2 });
await page.goto(url, { waitUntil: "domcontentloaded" });
await settle(page);
await openTab(page, "Ask");
await page.waitForSelector("#ask-input:not([disabled])", { timeout: 10000 });
await page.type("#ask-input", question);
await page.evaluate(() => {
const send = [...document.querySelectorAll(".ask-composer button")].find((b) => b.textContent.trim() === "Send");
if (send) send.click();
});
let entry = null;
for (let i = 0; i < 40 && !entry; i++) {
const r = cli(project, "chat", "--session", sid);
if (r.code === 0 && r.out && r.out.chat) entry = r.out.chat.find((c) => c.kind === "message");
if (!entry) await sleep(250);
}
if (!entry) throw new Error("the Ask question never reached the console");
const reply = path.join(HOME, "ask-reply.json");
fs.writeFileSync(
reply,
JSON.stringify({
chat: {
replies: [
{
id: "r" + entry.seq,
re: entry.seq,
conversation: entry.conversation,
md: "Paging is in `src/tasks/paginate.js`. It reads the `cursor` query parameter, returns up to 50 tasks and a `nextCursor`, and `GET /tasks` calls it on every request.",
},
],
},
agent: { status: "waiting" },
})
);
const r = cli(project, "post", "--session", sid, "--file", reply);
if (r.code !== 0) throw new Error(`posting the Ask reply failed (${r.code}): ${r.err.trim()}`);
await sleep(1000);
page.__tab = "Ask";
return await shoot(page, "ask");
} finally {
await ctx.close();
}
}
async function main() {
const names = args.only ? [args.only] : Object.keys(SHOTS);
for (const n of names) if (!SHOTS[n]) fail(`Unknown shot "${n}". Shots: ${Object.keys(SHOTS).join(", ")}`);
let puppeteer;
try {
puppeteer = (await import("puppeteer-core")).default;
} catch {
fail("puppeteer-core is not installed. Run: npm i --no-save puppeteer-core");
}
const browserPath = findBrowser();
if (!browserPath) fail("No Chrome or Edge found. Pass one with --browser <path>.");
SCRATCH = fs.mkdtempSync(path.join(SCRATCH_ROOT, "p2c-shots-"));
HOME = path.join(SCRATCH, "console-home");
ENV = { ...process.env, PLAN2CODE_CONSOLE_HOME: HOME, PLAN2CODE_NO_BROWSER: "1" };
fs.mkdirSync(HOME, { recursive: true });
fs.mkdirSync(OUT, { recursive: true });
let browser = null;
let project = null;
try {
project = makeSampleProject();
// A saved role keeps the dashboard's "pick your role" banner out of shot,
// and the Ocean highlight is the console's nearest to the Plan2Code blue
// the landing page uses.
fs.writeFileSync(path.join(HOME, "looks.json"), JSON.stringify({ looks: { role: "engineer", accent: "ocean" } }));
browser = await puppeteer.launch({
executablePath: browserPath,
headless: true,
args: ["--no-first-run", "--no-default-browser-check", "--mute-audio"],
});
for (const name of names) {
const { sid, url } = stage(project, name);
try {
if (SHOTS[name].ask) await captureAsk(browser, project, sid, url, SHOTS[name].ask);
else await capture(browser, url, name, SHOTS[name]);
} finally {
cli(project, "stop", "--session", sid);
}
}
} finally {
if (browser) await browser.close();
if (project) cli(project, "stop", "--all");
removeScratch();
}
}
// `stop` signals the server and returns without waiting for it to exit, and
// on Windows a server still closing holds its log open. Retry the delete, and
// never let a failed cleanup replace the run's own result or error.
function removeScratch() {
try {
fs.rmSync(SCRATCH, { recursive: true, force: true, maxRetries: 10, retryDelay: 200 });
} catch (e) {
console.log(`warning: could not remove ${SCRATCH} (${e.code || e.message}); delete it by hand`);
}
}
main().catch((e) => {
console.error(e && e.message ? e.message : e);
process.exit(1);
});
+91
View File
@@ -0,0 +1,91 @@
// Check the dashboard's "What each skill reads and writes" table against the
// skills it describes.
//
// The table is what the dashboard and the Ask tab treat as ground truth for a
// skill's files. Every backticked name in a row must appear in that skill's own
// prompt or its references, so a rename in a skill fails here instead of the
// table quietly going stale. `<placeholder>` segments match any placeholder,
// so `PLAN-DRAFT-<date>.md` accepts the skill's `PLAN-DRAFT-<YYYYMMDD>.md`.
import fs from "node:fs";
import path from "node:path";
import { fileURLToPath } from "node:url";
const ROOT = path.dirname(path.dirname(fileURLToPath(import.meta.url)));
const SRC = path.join(ROOT, "src");
const HEADING = "## What each skill reads and writes";
function readTree(dir) {
if (!fs.existsSync(dir)) return "";
return fs
.readdirSync(dir, { withFileTypes: true })
.map((entry) => {
const full = path.join(dir, entry.name);
if (entry.isDirectory()) return readTree(full);
return entry.name.endsWith(".md") ? fs.readFileSync(full, "utf8") : "";
})
.join("\n");
}
function skillCorpus(skill) {
const file = path.join(SRC, `plan2code-${skill}.md`);
if (!fs.existsSync(file)) return null;
return fs.readFileSync(file, "utf8") + "\n" + readTree(path.join(SRC, `plan2code-${skill}-references`));
}
// "3-implement(-review)" names both 3-implement and 3-implement-review.
function expandSkills(cell) {
const optional = cell.match(/^(.*)\((.+)\)$/);
return optional ? [optional[1], optional[1] + optional[2]] : [cell];
}
function tokenPattern(token) {
const escaped = token.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
return new RegExp(escaped.replace(/<[^>]+>/g, "<[^>]+>"));
}
function mentions(corpus, token) {
if (tokenPattern(token).test(corpus)) return true;
const leaf = token.slice(token.lastIndexOf("/") + 1);
return leaf !== "" && leaf !== token && tokenPattern(leaf).test(corpus);
}
const prompt = fs.readFileSync(path.join(SRC, "plan2code.md"), "utf8");
const start = prompt.indexOf(HEADING);
if (start === -1) {
console.error(` FAIL src/plan2code.md has no "${HEADING}" section.`);
process.exit(1);
}
const section = prompt.slice(start + HEADING.length).split(/\n## /)[0];
const rows = section
.split("\n")
.filter((line) => line.startsWith("|") && !/^\|\s*(skill|---)/.test(line))
.map((line) => line.split("|").slice(1, -1).map((cell) => cell.trim()));
if (rows.length === 0) {
console.error(` FAIL "${HEADING}" has no table rows.`);
process.exit(1);
}
let bad = 0;
for (const [skillCell, ...cells] of rows) {
const skills = expandSkills(skillCell);
const corpora = skills.map(skillCorpus);
if (corpora.some((corpus) => corpus === null)) {
bad++;
console.error(` FAIL ${skillCell}: no src/plan2code-<skill>.md for ${skills.join(" / ")}`);
continue;
}
const corpus = corpora.join("\n");
const tokens = cells.join(" ").match(/`[^`]+`/g) ?? [];
const missing = tokens.map((t) => t.slice(1, -1)).filter((token) => !mentions(corpus, token));
if (missing.length) {
bad++;
console.error(` FAIL ${skillCell}: not in the skill's source: ${missing.join(", ")}`);
} else {
console.log(` ok ${skillCell} (${tokens.length} name(s))`);
}
}
if (bad) {
console.error(`${bad} row(s) of "${HEADING}" disagree with their skill.`);
process.exit(1);
}
+36
View File
@@ -0,0 +1,36 @@
// Parse every shipped browser module as a real ES module.
//
// `node --check` is not enough: it parses a .js file as a CommonJS script, and
// it let a string literal containing a raw newline through to the browser,
// where the page died at import with nothing rendered. Importing the source as
// a data: module is the same parse the browser does.
import fs from "node:fs";
import path from "node:path";
import { fileURLToPath } from "node:url";
const ROOT = path.dirname(path.dirname(fileURLToPath(import.meta.url)));
const DIR = path.join(ROOT, "src", "web-console", "public");
const FILES = ["app.js", "answers.js", "render.js", "palette.js", "starters.js", "favicon.js"];
let bad = 0;
for (const name of FILES) {
const src = fs.readFileSync(path.join(DIR, name), "utf8");
try {
await import("data:text/javascript," + encodeURIComponent(src));
console.log(` ok ${name}`);
} catch (err) {
// A module that parses but cannot run here (app.js touches `document` at
// import) is fine; only a parse failure is a real problem.
const parseFailure = err instanceof SyntaxError;
if (parseFailure) {
bad++;
console.error(` FAIL ${name}: ${err.message}`);
} else {
console.log(` ok ${name} (parsed; not runnable outside a browser)`);
}
}
}
if (bad) {
console.error(`${bad} file(s) would not parse in a browser.`);
process.exit(1);
}
+18
View File
@@ -0,0 +1,18 @@
# Screenshot payloads
Each JSON file here stages one web console screen for the docs, with no agent
involved. `node scripts/capture-screenshots.mjs` opens a console session in a
throwaway sample project, posts these files, and saves the screens as light
and dark WebP images in `docs/screenshots/`.
| File | Stages |
| --- | --- |
| `dashboard.json` | The dashboard menu, suggesting the next build phase |
| `question.json` | A Plan session with three questions open (also the base of the Ask shot) |
| `doc.json` | A living document with settled, assumed and draft sections, and its review card |
| `build.json` | A phase being built: the progress bar, the task list and the current task |
| `signoff.json` | The completion report and the sign-off card |
| `finish.json` | The finished screen after sign-off (posted after `signoff.json`) |
Refresh every image with `node scripts/capture-screenshots.mjs`, or one with
`--only <shot>`.
+40
View File
@@ -0,0 +1,40 @@
{
"workflow": "implement",
"title": "task-api: Phase 3",
"headline": {
"stage": "Phase 3: API endpoints",
"cleared": 5,
"total": 11
},
"docs": [
{
"id": "phase",
"title": "Phase 3 tasks",
"version": 6,
"blocks": [
{
"id": "tasks",
"state": "settled",
"md": "- [x] **Task 3.1:** Register the /tasks routes plugin\n- [x] **Task 3.2:** Add GET /tasks with cursor paging\n- [x] **Task 3.3:** Add GET /tasks/:id\n- [x] **Task 3.4:** Add POST /tasks with a Zod schema\n- [x] **Task 3.5:** Add PATCH /tasks/:id for status and fields\n- [ ] **Task 3.6:** Add DELETE /tasks/:id as a soft delete\n- [ ] **Task 3.7:** Filter GET /tasks by status and assignee\n- [ ] **Task 3.8:** Return one error shape for 400, 404 and 409\n- [ ] **Task 3.9:** Document the endpoints in openapi.yaml\n- [ ] **Task 3.10:** Write tests for paging, filters and validation\n- [ ] **Task 3.11:** Run the test suite and record the results"
}
]
},
{
"id": "phasefile",
"title": "phase-3.md",
"version": 6,
"blocks": [
{
"id": "body",
"state": "settled",
"md": "# Phase 3: API endpoints\n\n**Status:** In Progress\n**Estimated Tasks:** 11\n\n## Overview\n\nThe /tasks endpoints: list with paging and filters, read, create, update and soft delete, with one error shape for every failure.\n\n## Prerequisites\n\n- Phase 2 complete (the tasks table and model exist) VERIFIED\n- Postgres runs locally VERIFIED\n\n## Tasks\n\n- [x] **Task 3.1:** Register the /tasks routes plugin\n- [x] **Task 3.2:** Add GET /tasks with cursor paging\n- [x] **Task 3.3:** Add GET /tasks/:id\n- [x] **Task 3.4:** Add POST /tasks with a Zod schema\n- [x] **Task 3.5:** Add PATCH /tasks/:id for status and fields\n- [ ] **Task 3.6:** Add DELETE /tasks/:id as a soft delete\n- [ ] **Task 3.7:** Filter GET /tasks by status and assignee\n- [ ] **Task 3.8:** Return one error shape for 400, 404 and 409\n- [ ] **Task 3.9:** Document the endpoints in openapi.yaml\n- [ ] **Task 3.10:** Write tests for paging, filters and validation\n- [ ] **Task 3.11:** Run the test suite and record the results\n\n## Acceptance Criteria\n\n- Every endpoint validates its input and answers errors in one shape\n- Paging stays stable while tasks are added and removed\n"
}
]
}
],
"agent": {
"status": "working",
"activity": "Task 6 of 11: deleting a task",
"quietMinutes": 15
}
}
+12
View File
@@ -0,0 +1,12 @@
{
"menu": {
"note": "task-api is built up to Phase 2 of 4. Phase 3 is next.",
"recommend": "plan2code-3-implement",
"details": {
"plan2code-3-implement": "Phase 3 of 4 is next"
}
},
"agent": {
"status": "waiting"
}
}
+63
View File
@@ -0,0 +1,63 @@
{
"workflow": "plan",
"title": "Task API",
"headline": {
"stage": "Step 3 of 7: What we've agreed",
"cleared": 4,
"total": 7,
"confidence": "Most of it is settled. Paging is assumed for now."
},
"docs": [
{
"id": "map",
"title": "What we've agreed",
"version": 2,
"note": "Added paging",
"blocks": [
{
"id": "b1",
"state": "settled",
"md": "## Where we're headed\n\nA REST API for a team's tasks: create, list, update and delete them, with filters by status and assignee. Built on Fastify, Postgres and Zod."
},
{
"id": "b2",
"state": "assumed",
"md": "### Paging\n\nAssuming cursor paging, 50 tasks per page. Still open: the web app may want page numbers."
},
{
"id": "b3",
"state": "draft",
"md": "### Still to work out\n\n- How clients sign in\n- Whether a deleted task can be restored"
}
]
}
],
"items": [
{
"id": "agree",
"kind": "review",
"title": "Check the plan so far",
"doc": "map",
"body": "Read what we've agreed, then approve it or say what to change. Paging is marked as assumed.",
"verdicts": [
{
"id": "approve",
"label": "Looks right"
},
{
"id": "changes",
"label": "I want changes"
}
],
"token": {
"approve": "yes",
"changes": "needs changes"
},
"nothingYet": "Nothing is written to disk until you approve.",
"fallbackText": "Does what we've agreed look right? Reply yes, or describe what to change."
}
],
"agent": {
"status": "waiting"
}
}
+21
View File
@@ -0,0 +1,21 @@
{
"items": [
{
"id": "signoff",
"status": "answered",
"answer": {
"verdict": "approve"
}
}
],
"finish": {
"headline": "Phase 3 is built",
"body": "11 of 11 tasks, all on disk. Phase 4 (authentication) is next.",
"command": "/plan2code-3-implement specs/task-api/overview.md",
"doc": "report",
"dashboard": true
},
"agent": {
"status": "waiting"
}
}
+117
View File
@@ -0,0 +1,117 @@
{
"workflow": "plan",
"title": "Task API",
"headline": {
"stage": "Step 1 of 7: What we're building",
"cleared": 1,
"total": 7,
"confidence": "A clear idea so far. Who calls the API and how it signs in are still open."
},
"topics": [
{
"id": "t1",
"title": "What we're building",
"status": "current"
}
],
"items": [
{
"id": "q2",
"topic": "t1",
"kind": "choice",
"title": "How tasks are paged",
"body": "A team can have thousands of tasks. How should the list endpoint hand them back?",
"options": [
{
"k": "A",
"text": "Cursor paging",
"detail": "Stable while tasks are added and removed. No jumping to page 7.",
"recommended": true
},
{
"k": "B",
"text": "Page numbers",
"detail": "Familiar and easy to link to, but rows shift when tasks change mid-scroll."
},
{
"k": "C",
"text": "Everything at once",
"detail": "Simplest to build; slow and heavy once a team has a few thousand tasks."
}
],
"token": {
"A": "a",
"B": "b",
"C": "c"
},
"fallbackText": "How tasks are paged: a) cursor paging b) page numbers c) everything at once. My guess: a."
},
{
"id": "q3",
"topic": "t1",
"kind": "multi",
"title": "Day one endpoints",
"body": "Which of these should be there on the first day? Anything left unticked waits for a later version.",
"options": [
{
"k": "1",
"text": "List and read tasks",
"detail": "GET /tasks and GET /tasks/:id, with filters by status and assignee."
},
{
"k": "2",
"text": "Create and update tasks",
"detail": "POST /tasks and PATCH /tasks/:id, validated."
},
{
"k": "3",
"text": "Delete tasks",
"detail": "A soft delete, so a task can come back."
},
{
"k": "4",
"text": "Comments on a task",
"detail": "A thread under each task. Doubles the data model."
}
],
"presets": [
{
"label": "The essentials",
"ks": [
"1",
"2"
]
},
{
"label": "Everything",
"ks": [
"1",
"2",
"3",
"4"
]
}
],
"token": {
"1": "1",
"2": "2",
"3": "3",
"4": "4"
},
"fallbackText": "Day one endpoints (pick any): 1) list and read 2) create and update 3) delete 4) comments on a task."
},
{
"id": "q4",
"topic": "t1",
"kind": "text",
"title": "Who calls it",
"body": "Which apps will call this API, and roughly how many people use them?",
"placeholder": "For example: our web app and a mobile app, about 2,000 people across 40 teams.",
"token": {},
"fallbackText": "Who calls it: which apps will call this API, and roughly how many people use them?"
}
],
"agent": {
"status": "waiting"
}
}
+88
View File
@@ -0,0 +1,88 @@
{
"workflow": "implement",
"title": "task-api: Phase 3",
"headline": {
"stage": "Phase 3: API endpoints",
"cleared": 11,
"total": 11
},
"docs": [
{
"id": "phase",
"title": "Phase 3 tasks",
"version": 12,
"blocks": [
{
"id": "tasks",
"state": "settled",
"md": "- [x] **Task 3.1:** Register the /tasks routes plugin\n- [x] **Task 3.2:** Add GET /tasks with cursor paging\n- [x] **Task 3.3:** Add GET /tasks/:id\n- [x] **Task 3.4:** Add POST /tasks with a Zod schema\n- [x] **Task 3.5:** Add PATCH /tasks/:id for status and fields\n- [x] **Task 3.6:** Add DELETE /tasks/:id as a soft delete\n- [x] **Task 3.7:** Filter GET /tasks by status and assignee\n- [x] **Task 3.8:** Return one error shape for 400, 404 and 409\n- [x] **Task 3.9:** Document the endpoints in openapi.yaml\n- [x] **Task 3.10:** Write tests for paging, filters and validation\n- [x] **Task 3.11:** Run the test suite and record the results"
}
]
},
{
"id": "report",
"title": "Completion report",
"version": 1,
"blocks": [
{
"id": "summary",
"state": "settled",
"md": "## Summary\n\nThe /tasks endpoints work end to end. Clients can list tasks with cursor paging and filters, read one, create, update and soft delete, and every failure comes back in the same error shape.\n\n**Tasks:** 11 of 11 done, none blocked."
},
{
"id": "tests",
"state": "settled",
"md": "## Tests\n\n| Suite | Passed | Failed |\n| --- | --- | --- |\n| tasks | 31 | 0 |\n| paging | 8 | 0 |"
},
{
"id": "files",
"state": "settled",
"md": "## Files\n\n**Created:** `src/tasks/routes.js`, `src/tasks/schema.js`, `src/tasks/paginate.js`, `test/tasks.test.js`\n\n**Modified:** `src/app.js`, `openapi.yaml`"
},
{
"id": "verify",
"state": "settled",
"md": "## How to check it\n\n1. Run `npm test`: 39 passing.\n2. Create two tasks, then call `GET /tasks?limit=1`: one task and a `nextCursor`.\n3. `PATCH /tasks/:id` with an unknown status answers 400 in the shared error shape."
}
]
}
],
"items": [
{
"id": "signoff",
"kind": "review",
"title": "Approve this phase",
"doc": "report",
"body": "Phase 3 is built. Read the report, try the checks, then approve it, review the code first, or say what to change.",
"verdicts": [
{
"id": "review",
"label": "Review the code first"
},
{
"id": "approve",
"label": "Approve this phase"
},
{
"id": "changes",
"label": "I want changes"
}
],
"token": {
"review": "review",
"approve": "approved",
"changes": "needs changes"
},
"nothingYet": "Nothing is marked complete until you approve.",
"gate": {
"id": "signoff",
"authoritative": true
},
"fallbackText": "Reply \"approved\" to mark this phase complete, \"review\" to run the code review first, or describe any issues."
}
],
"agent": {
"status": "waiting",
"quietMinutes": null
}
}
+61
View File
@@ -0,0 +1,61 @@
// Shared helpers for the skill-script tests (scripts/test-*script*.mjs).
//
// The tests run the scripts the way an agent does: as `node <skill>/scripts/x.mjs`
// from the built skills/ tree, against throwaway fixtures in the OS temp dir,
// with the clock pinned by PLAN2CODE_NOW. They live in scripts/ so nothing
// here ships.
import fs from "node:fs";
import os from "node:os";
import path from "node:path";
import { spawnSync, execFileSync } from "node:child_process";
import { fileURLToPath } from "node:url";
export const ROOT = path.dirname(path.dirname(fileURLToPath(import.meta.url)));
export const SKILLS = path.join(ROOT, "skills");
export const NOW = "2026-08-03 14:05";
// Rebuild skills/ on every import, not only when it is missing: a test file run
// directly after an edit to src/ must not test a stale tree. The build is
// idempotent and takes well under a second.
execFileSync(process.execPath, [path.join(ROOT, "install.js"), "--build-skills"], { stdio: "ignore" });
export function runScript(skill, script, args, { cwd, env } = {}) {
const r = spawnSync(process.execPath, [path.join(SKILLS, skill, "scripts", script), ...args], {
cwd: cwd || ROOT,
encoding: "utf8",
env: { ...process.env, PLAN2CODE_NOW: NOW, GIT_CONFIG_NOSYSTEM: "1", ...env },
});
let json = null;
try {
json = JSON.parse(r.stdout);
} catch {}
return { code: r.status, json, stdout: r.stdout, stderr: r.stderr };
}
export function tmp(name) {
return fs.mkdtempSync(path.join(fs.realpathSync.native(os.tmpdir()), `p2c-${name}-`));
}
export function write(root, rel, content) {
const p = path.join(root, rel);
fs.mkdirSync(path.dirname(p), { recursive: true });
fs.writeFileSync(p, content);
return p;
}
export const read = (root, rel) => fs.readFileSync(path.join(root, rel), "utf8");
export function gitInit(dir, branch = "main") {
const g = (...a) => execFileSync("git", a, { cwd: dir, stdio: "pipe", encoding: "utf8" });
g("init", "-q", "-b", branch);
g("config", "user.email", "t@example.com");
g("config", "user.name", "Test");
g("config", "commit.gpgsign", "false");
return g;
}
/** A phase-N.md in the Documentation template's shape. tasks: [[marker, id], ...] */
export function phase(n, status, tasks, extra = "") {
return `# Phase ${n}: Name ${n}\n\n**Status:** ${status}\n**Estimated Tasks:** ${tasks.length}\n\n## Overview\n\nPhase ${n} builds the thing. More words here.\n\n## Prerequisites\n\n- Something exists\n\n## Tasks\n\n${tasks.map(([m, id]) => `- [${m}] **Task ${id}:** Do ${id}`).join("\n")}\n${extra}\n## Acceptance Criteria\n\n- It works\n`;
}
+208
View File
@@ -0,0 +1,208 @@
// Tests for the global `plan2code` launcher. Run with: node --test scripts/test-launcher.mjs
//
// The launcher is a script that runs on load, so each test spawns it for real: a temp home
// for ~/.plan2code/launcher.json, and stand-in `claude` / `devin` commands on PATH that
// record the arguments they were started with. stdin is a pipe, never a TTY, so the launcher
// takes its no-prompt path: no menu and no wait at the model notice.
import test from "node:test";
import assert from "node:assert/strict";
import fs from "node:fs";
import os from "node:os";
import path from "node:path";
import { spawnSync } from "node:child_process";
import { createRequire } from "node:module";
import { fileURLToPath } from "node:url";
const require = createRequire(import.meta.url);
const ROOT = path.resolve(path.dirname(fileURLToPath(import.meta.url)), "..");
const LAUNCHER = path.join(ROOT, "src", "launcher", "plan2code.js");
const IS_WINDOWS = process.platform === "win32";
const NOTICE = "will use the model you last used in it";
/** A temp home and a bin dir holding stand-ins for `clis`; each writes its args to <bin>/<cli>.json. */
function sandbox(clis) {
const dir = fs.mkdtempSync(path.join(os.tmpdir(), "p2c-launcher-"));
const home = path.join(dir, "home");
const bin = path.join(dir, "bin");
fs.mkdirSync(home);
fs.mkdirSync(bin);
const recorder = path.join(bin, "record.js");
fs.writeFileSync(
recorder,
'require("fs").writeFileSync(require("path").join(__dirname, process.argv[2] + ".json"), JSON.stringify(process.argv.slice(3)));\n' +
"process.exitCode = Number(process.env.P2C_TEST_EXIT || 0);\n"
);
for (const cli of clis) {
if (IS_WINDOWS) {
fs.writeFileSync(path.join(bin, `${cli}.cmd`), `@"${process.execPath}" "${recorder}" ${cli} %*\r\n`);
} else {
const file = path.join(bin, cli);
fs.writeFileSync(file, `#!/bin/sh\nexec "${process.execPath}" "${recorder}" ${cli} "$@"\n`);
fs.chmodSync(file, 0o755);
}
}
return {
home,
stateFile: path.join(home, ".plan2code", "launcher.json"),
run: (args, env = {}) =>
spawnSync(process.execPath, [LAUNCHER, ...args], {
cwd: dir,
input: "",
encoding: "utf8",
env: { ...process.env, PATH: bin, HOME: home, USERPROFILE: home, ...env },
}),
argsOf: (cli) => {
const file = path.join(bin, `${cli}.json`);
return fs.existsSync(file) ? JSON.parse(fs.readFileSync(file, "utf8")) : null;
},
cleanup: () => fs.rmSync(dir, { recursive: true, force: true }),
};
}
test("launcher: Claude Code gets the prompt first, then bypassPermissions and the forwarded args", (t) => {
const box = sandbox(["claude"]);
t.after(box.cleanup);
const result = box.run(["--verbose"]);
assert.equal(result.status, 0, result.stderr);
assert.deepEqual(box.argsOf("claude"), ["/plan2code", "--permission-mode", "bypassPermissions", "--verbose"]);
assert.match(result.stdout, new RegExp(`Claude Code ${NOTICE}`));
});
test("launcher: Devin keeps its args ahead of the -- and the prompt", (t) => {
const box = sandbox(["devin"]);
t.after(box.cleanup);
const result = box.run(["--verbose"]);
assert.equal(result.status, 0, result.stderr);
assert.deepEqual(box.argsOf("devin"), ["--permission-mode", "bypass", "--verbose", "--", "/plan2code"]);
});
test("launcher: a forwarded --model skips the model notice", (t) => {
const box = sandbox(["claude"]);
t.after(box.cleanup);
for (const args of [["--model", "opus"], ["--model=opus"]]) {
const result = box.run(args);
assert.equal(result.status, 0, result.stderr);
assert.doesNotMatch(result.stdout, new RegExp(NOTICE));
assert.deepEqual(box.argsOf("claude"), ["/plan2code", "--permission-mode", "bypassPermissions", ...args]);
}
});
test("launcher: --cli picks the CLI and is never forwarded; the pick is merged into launcher.json", (t) => {
const box = sandbox(["claude", "devin"]);
t.after(box.cleanup);
fs.mkdirSync(path.dirname(box.stateFile), { recursive: true });
fs.writeFileSync(box.stateFile, JSON.stringify({ lastFolder: box.home }));
for (const args of [["--cli", "devin"], ["--cli=DEVIN"]]) {
const result = box.run(args);
assert.equal(result.status, 0, result.stderr);
assert.deepEqual(box.argsOf("devin"), ["--permission-mode", "bypass", "--", "/plan2code"]);
}
assert.equal(box.argsOf("claude"), null);
assert.deepEqual(JSON.parse(fs.readFileSync(box.stateFile, "utf8")), { lastFolder: box.home, lastCli: "devin" });
});
test("launcher: with both installed and no terminal to ask in, the last pick wins", (t) => {
const box = sandbox(["claude", "devin"]);
t.after(box.cleanup);
fs.mkdirSync(path.dirname(box.stateFile), { recursive: true });
fs.writeFileSync(box.stateFile, JSON.stringify({ lastCli: "devin" }));
assert.equal(box.run([]).status, 0);
assert.notEqual(box.argsOf("devin"), null);
assert.equal(box.argsOf("claude"), null);
});
test("launcher: an unknown --cli exits 2, a missing CLI 127, and nothing is started", (t) => {
const box = sandbox(["claude"]);
t.after(box.cleanup);
const unknown = box.run(["--cli", "bogus"]);
assert.equal(unknown.status, 2);
assert.match(unknown.stderr, /--cli takes claude or devin, not "bogus"/);
const missing = box.run(["--cli", "devin"]);
assert.equal(missing.status, 127);
assert.match(missing.stderr, /the `devin` CLI was not found/);
assert.equal(box.argsOf("claude"), null);
});
test("launcher: mergeModels keeps the shipped list first and drops user duplicates", () => {
const { mergeModels, modelEntries } = require(LAUNCHER);
const shipped = [{ id: "a", label: "A" }, { id: "b", label: "B" }];
const user = [{ id: "b", label: "B old" }, { id: "c", label: "C" }, { id: "c", label: "C again" }];
assert.deepEqual(mergeModels(shipped, user), [...shipped, { id: "c", label: "C" }]);
assert.deepEqual(mergeModels(shipped, []), shipped);
// A missing or corrupt user file yields no additions rather than an error.
assert.deepEqual(modelEntries(null, "devin"), []);
assert.deepEqual(modelEntries({ devin: "nope" }, "devin"), []);
assert.deepEqual(modelEntries({ devin: [{ id: 1 }, null, { id: "ok", label: "Ok" }] }, "devin"), [{ id: "ok", label: "Ok" }]);
});
test("launcher: the shipped models.json is curated Devin at low/medium/high only", () => {
const shipped = JSON.parse(fs.readFileSync(path.join(path.dirname(LAUNCHER), "models.json"), "utf8"));
assert.ok(shipped.claude.length > 0);
assert.ok(shipped.devin.length > 0 && shipped.devin.length <= 30);
for (const { id, label } of shipped.devin) {
assert.ok(id && label);
assert.doesNotMatch(id, /(xhigh|max|fast|priority)$/);
}
});
test("launcher: with neither CLI installed it exits 127 naming both", (t) => {
const box = sandbox([]);
t.after(box.cleanup);
const result = box.run([]);
assert.equal(result.status, 127);
assert.match(result.stderr, /neither `claude` nor `devin` was found/);
});
test("launcher: a forwarded argument with a space and a cmd metacharacter arrives unchanged", (t) => {
// On Windows the stand-in is a .cmd, so this goes through the cmd.exe quoting in quoteForCmd.
const box = sandbox(["claude"]);
t.after(box.cleanup);
const result = box.run(["--append-system-prompt", "a b&c"]);
assert.equal(result.status, 0, result.stderr);
assert.deepEqual(box.argsOf("claude"), ["/plan2code", "--permission-mode", "bypassPermissions", "--append-system-prompt", "a b&c"]);
});
test("launcher: the CLI's non-zero exit code is passed through", (t) => {
const box = sandbox(["claude"]);
t.after(box.cleanup);
const result = box.run([], { P2C_TEST_EXIT: "7" });
assert.equal(result.status, 7, result.stderr);
assert.notEqual(box.argsOf("claude"), null);
});
test("launcher: renderBanner draws Planny and the name, colored only through paint", () => {
const { renderBanner, paint } = require(LAUNCHER);
const plain = renderBanner(paint(false));
assert.doesNotMatch(plain, /\x1b/);
for (const part of ["╭───╮", "╰┬─┬╯", "★", "Plan2Code"]) assert.ok(plain.includes(part), part);
assert.ok(plain.split("\n").length >= 4);
const colored = renderBanner(paint(true));
assert.ok(colored.includes("\x1b["));
assert.ok(colored.includes("\x1b[0m"));
});
test("launcher: paint wraps text in the escape codes only when color is on", () => {
const { paint } = require(LAUNCHER);
assert.equal(paint(false).red("x"), "x");
assert.equal(paint(true).red("x"), "\x1b[31mx\x1b[0m");
});
test("launcher: a piped run prints no escape codes, with or without NO_COLOR", (t) => {
const box = sandbox(["claude"]);
t.after(box.cleanup);
const none = sandbox([]);
t.after(none.cleanup);
for (const env of [{}, { NO_COLOR: "1" }]) {
const runs = [box.run([], env), box.run(["--cli", "bogus"], env), none.run([], env)];
assert.deepEqual(runs.map((r) => r.status), [0, 2, 127]);
for (const r of runs) {
assert.doesNotMatch(r.stdout, /\x1b/);
assert.doesNotMatch(r.stderr, /\x1b/);
}
}
});
+174
View File
@@ -0,0 +1,174 @@
// Tests for plan2code-4-finalize's scripts/feedback-payload.mjs.
//
// See skill-script-helpers.mjs for how these run.
import { test } from "node:test";
import assert from "node:assert/strict";
import fs from "node:fs";
import path from "node:path";
import { spawnSync } from "node:child_process";
import crypto from "node:crypto";
import { ROOT, SKILLS, NOW, runScript, tmp, write, read, gitInit, phase } from "./skill-script-helpers.mjs";
/* ========================================================= feedback-payload.mjs */
test("feedback-payload: payload from the archive, real prompt hashes, size and truncation", () => {
const root = tmp("fb");
const spec = "specs--completed/lunch-vote";
write(root, `${spec}/PLAN-DRAFT-20260801.md`, '# Plan\n\n<!-- METRICS_JSON {"confidence": 92, "clarification_rounds": 1, "functional_requirements_count": 5, "non_functional_requirements_count": 2, "risk_count": 3, "phase_count": 2, "verification_gaps_found": 0, "confidence_breakdown": {"requirements": 24, "feasibility": 23, "integration": 23, "risk": 22}} -->\n');
write(root, `${spec}/overview.md`, '# O\n\n<!-- METRICS_JSON {"step": "document", "total_tasks": 3, "tasks_per_phase": [2, 1], "phase_count": 2, "parallel_groups_identified": 0, "verification_items_added": 1} -->\n\n## User Feedback\n| Field | Value |\n|-------|-------|\n| Rating | 8 |\n| Reason | Good \\| fast |\n| Went Well | Specs |\n| Went Poorly | Review |\n');
write(root, `${spec}/phase-1.md`, phase(1, "Complete", [["x", "1.1"], ["x", "1.T"]]));
write(root, `${spec}/phase-2.md`, phase(2, "Complete", [["!", "2.1"]]));
const outDir = path.join(root, "out");
const F = (...a) => runScript("plan2code-4-finalize", "feedback-payload.mjs", [...a, "--root", root]);
assert.equal(F("build", "lunch-vote", "--out", outDir).code, 2, "the two judgment counts are required");
const b = F("build", "lunch-vote", "--out", outDir, "--set", "verification_failures_found=1", "--set", "documentation_updates_needed=2");
assert.equal(b.code, 0, b.stderr);
const p = b.json.payload;
assert.match(p.run_id, /^run-20260803-140500-[0-9a-f]{4}$/);
assert.equal(p.plan2code_version, JSON.parse(fs.readFileSync(path.join(ROOT, "version.json"), "utf8")).version);
const srcHash = crypto.createHash("sha256").update(fs.readFileSync(path.join(ROOT, "src", "plan2code-1-plan.md"), "utf8")).digest("hex").slice(0, 12);
assert.equal(p.prompt_versions_short.plan, srcHash, "the hash is the src prompt's, recovered from the installed SKILL.md");
assert.ok(Object.values(p.prompt_versions_short).every((v) => /^[0-9a-f]{12}$/.test(v)));
assert.deepEqual(p.step1.confidence_breakdown, { requirements: 24, feasibility: 23, integration: 23, risk: 22 });
assert.equal(p.step1.final_confidence, 92);
assert.equal(p.step2.total_tasks, 3);
assert.deepEqual(p.step3, { task_completion_rate: 0.67, tasks_completed: 2, tasks_total: 3, blocker_count: 1 });
assert.deepEqual(p.step4, { completion_rate_at_audit: 0.67, verification_failures_found: 1, documentation_updates_needed: 2, archival_succeeded: true });
assert.deepEqual(p.user_feedback, { overall_rating: 8, rating_reason: "Good | fast", what_went_well: "Specs", what_went_poorly: "Review" });
assert.equal(b.json.title, `[Feedback] v${p.plan2code_version} - rating 8/10`);
assert.match(b.json.body, /<!-- METRICS_JSON \{"schema_version":"1\.0"/);
assert.ok(!/[—]/.test(b.json.title), "no em dash in the issue title");
assert.equal(b.json.oversize, false);
assert.ok(fs.existsSync(path.join(outDir, "body.md")));
// The ingest side parses the body back.
const parsed = JSON.parse(b.json.body.match(/<!--\s*METRICS_JSON\s+(\{[\s\S]*?\})\s*-->/)[1]);
assert.equal(parsed.schema_version, "1.0");
// Oversize, then truncate.
const ov = path.join(root, spec, "overview.md");
fs.writeFileSync(ov, fs.readFileSync(ov, "utf8").replace("| Good \\| fast |", `| ${"long reason ".repeat(500)} |`));
const big = F("build", "lunch-vote", "--out", outDir, "--set", "verification_failures_found=1", "--set", "documentation_updates_needed=2");
assert.equal(big.json.oversize, true);
assert.match(big.json.next, /--truncate/);
const cut = F("build", "lunch-vote", "--out", outDir, "--set", "verification_failures_found=1", "--set", "documentation_updates_needed=2", "--truncate");
assert.equal(cut.json.oversize, false);
assert.deepEqual(cut.json.truncated, ["rating_reason"]);
const dry = runScript("plan2code-4-finalize", "feedback-payload.mjs", ["submit", "--from", outDir, "--dry-run"]);
assert.equal(dry.code, 0);
assert.match(dry.json.tiers[0], /^gh issue create --repo jparkerweb\/plan2code .*--label community-feedback$/);
assert.equal(dry.json.oversize, false);
assert.equal(dry.json.urlBytes, cut.json.urlBytes);
fs.writeFileSync(ov, "# O\n");
assert.equal(F("build", "lunch-vote", "--set", "verification_failures_found=1", "--set", "documentation_updates_needed=2").code, 3, "no feedback table, nothing to submit");
});
/* ================================================= verification regressions */
test("feedback-payload: truncation always ends, '-->' survives ingest, lenient rating, absent steps are null", () => {
const root = tmp("fb2");
const spec = "specs--completed/x";
const long = (w) => `${w} `.repeat(400).trim();
// The visible answers carry markup that must stay text: a stray comment
// opener, and a fake payload comment an ingest regex could match first.
const fake = '<!-- METRICS_JSON {"schema_version":"1.0"} -->';
const inputs = { rating_reason: `see {a} --> b <!-- x ${long("reason")}`, what_went_well: `${fake} ${long("well")}`, what_went_poorly: long("poorly") };
write(root, `${spec}/overview.md`, `# O\n\n## User Feedback\n| field | value |\n|---|---|\n| rating | 8/10 |\n| reason | ${inputs.rating_reason} |\n| went well | ${inputs.what_went_well} |\n| went poorly | ${inputs.what_went_poorly} |\n`);
write(root, `${spec}/phase-1.md`, phase(1, "Complete", [["x", "1.1"]]));
const F = (...a) => runScript("plan2code-4-finalize", "feedback-payload.mjs", [...a, "--root", root]);
const sets = ["--set", "verification_failures_found=0", "--set", "documentation_updates_needed=0"];
assert.equal(F("build", "x", ...sets, "--set", "bogus=1").code, 2);
// Untruncated, the first METRICS_JSON match is the real payload: its answers
// come back exactly as typed, not the fake's bare schema_version.
const whole = F("build", "x", ...sets, "--out", path.join(root, "whole"));
assert.equal(whole.code, 0, whole.stderr);
const first = JSON.parse(whole.json.body.match(/<!--\s*METRICS_JSON\s+(\{[\s\S]*?\})\s*-->/)[1]);
assert.deepEqual(first.user_feedback, { overall_rating: 8, ...inputs });
assert.equal(whole.json.body.split("<!--").length - 1, 1, "the payload's is the only comment that opens");
assert.ok(!/<!--(?! METRICS_JSON \{"schema_version":"1\.0","run_id")/.test(whole.json.body), "no comment opens before the payload's");
const r = spawnSync(process.execPath, [path.join(SKILLS, "plan2code-4-finalize", "scripts", "feedback-payload.mjs"), "build", "x", ...sets, "--truncate", "--out", path.join(root, "out"), "--root", root], { encoding: "utf8", timeout: 20000, env: { ...process.env, PLAN2CODE_NOW: NOW } });
assert.equal(r.status, 0, "build --truncate finishes with every field long");
const j = JSON.parse(r.stdout);
assert.equal(j.oversize, false);
assert.deepEqual(j.truncated, ["rating_reason", "what_went_well", "what_went_poorly"]);
assert.equal(j.payload.user_feedback.overall_rating, 8, "8/10 reads as 8, field names in any case");
assert.equal(j.payload.step1, null);
assert.equal(j.payload.step2, null);
// The ingest side's own extraction regex (plan2code-metrics collector.ts).
const m = j.body.match(/<!--\s*METRICS_JSON\s+(\{[\s\S]*?\})\s*-->/);
const back = JSON.parse(m[1]);
assert.match(back.user_feedback.rating_reason, /^see \{a\} --> b <!-- x/);
assert.equal(back.user_feedback.what_went_well, j.payload.user_feedback.what_went_well);
assert.match(back.user_feedback.what_went_well, /^<!-- METRICS_JSON/, "the fake comment is still just text in the real payload");
assert.ok(!/<!--(?! METRICS_JSON \{"schema_version":"1\.0","run_id")/.test(j.body), "no comment opens before the payload's");
});
test("feedback-payload: --truncate never splits an emoji", () => {
const root = tmp("fb3");
const spec = "specs--completed/x";
const reason = "a" + "\u{1F44D}".repeat(400);
write(root, `${spec}/overview.md`, `# O\n\n## User Feedback\n| Field | Value |\n|---|---|\n| Rating | 9 |\n| Reason | ${reason} |\n| Went Well | ok |\n| Went Poorly | none |\n`);
write(root, `${spec}/phase-1.md`, phase(1, "Complete", [["x", "1.1"]]));
const r = runScript("plan2code-4-finalize", "feedback-payload.mjs", ["build", "x", "--set", "verification_failures_found=0", "--set", "documentation_updates_needed=0", "--truncate", "--out", path.join(root, "out"), "--root", root]);
assert.equal(r.code, 0, r.stderr);
assert.equal(r.json.oversize, false);
assert.deepEqual(r.json.truncated, ["rating_reason"]);
const back = JSON.parse(r.json.body.match(/<!--\s*METRICS_JSON\s+(\{[\s\S]*?\})\s*-->/)[1]);
assert.equal(back.user_feedback.rating_reason, r.json.payload.user_feedback.rating_reason);
assert.ok(!/[\uD800-\uDBFF](?![\uDC00-\uDFFF])/.test(back.user_feedback.rating_reason), "no lone high surrogate");
assert.match(back.user_feedback.rating_reason, /^a(\u{1F44D})+\.\.\.$/u);
});
test("feedback-payload: task counts read the same phase files specs.mjs does, never a stale sibling", () => {
const root = tmp("fb5");
const spec = "specs--completed/x";
write(root, `${spec}/overview.md`, "# O\n\n## User Feedback\n| Field | Value |\n|---|---|\n| Rating | 7 |\n| Reason | r |\n| Went Well | w |\n| Went Poorly | p |\n");
write(root, `${spec}/phase-1.md`, phase(1, "Complete", [["x", "1.1"], ["x", "1.2"]]));
write(root, `${spec}/phase-1-old.md`, phase(1, "Not Started", [[" ", "1.1"], [" ", "1.2"], [" ", "1.3"]]));
const r = runScript("plan2code-4-finalize", "feedback-payload.mjs", ["build", "x", "--set", "verification_failures_found=0", "--set", "documentation_updates_needed=0", "--out", path.join(root, "out"), "--root", root]);
assert.equal(r.code, 0, r.stderr);
assert.deepEqual(r.json.payload.step3, { task_completion_rate: 1, tasks_completed: 2, tasks_total: 2, blocker_count: 0 });
const m = runScript("plan2code-4-finalize", "specs.mjs", ["metrics", spec, "--step", "finalize", "--set", "verification_failures_found=0", "--set", "documentation_updates_needed=0", "--root", root]);
assert.equal(m.code, 0, m.stderr);
assert.equal(m.json.metrics.tasks_total, r.json.payload.step3.tasks_total);
assert.equal(m.json.metrics.tasks_completed, r.json.payload.step3.tasks_completed);
});
test("feedback-payload: a bare name in both specs/ and specs--completed/ is ambiguous", () => {
const root = tmp("fb4");
const table = (rating) => `# O\n\n## User Feedback\n| Field | Value |\n|---|---|\n| Rating | ${rating} |\n| Reason | r |\n| Went Well | w |\n| Went Poorly | p |\n`;
write(root, "specs--completed/x/overview.md", table(3));
write(root, "specs/x/overview.md", table(7));
const sets = ["--set", "verification_failures_found=0", "--set", "documentation_updates_needed=0"];
const F = (...a) => runScript("plan2code-4-finalize", "feedback-payload.mjs", [...a, ...sets, "--out", path.join(root, "out"), "--root", root]);
const bare = F("build", "x");
assert.equal(bare.code, 5);
assert.equal(bare.json.error, "ambiguous-spec");
assert.match(bare.json.next, /specs\/x/);
const live = F("build", "specs/x");
assert.equal(live.code, 0, live.stderr);
assert.equal(live.json.payload.user_feedback.overall_rating, 7, "the exact path reads this run, not the old archive");
assert.equal(live.json.payload.step4.archival_succeeded, false);
});
test("feedback-payload: submit refuses an oversize URL when gh fails, and dry-run says so", () => {
const root = tmp("fb5");
write(root, "specs--completed/x/overview.md", `# O\n\n## User Feedback\n| Field | Value |\n|---|---|\n| Rating | 6 |\n| Reason | ${"long reason ".repeat(800)} |\n| Went Well | w |\n| Went Poorly | p |\n`);
const outDir = path.join(root, "out");
const b = runScript("plan2code-4-finalize", "feedback-payload.mjs", ["build", "x", "--set", "verification_failures_found=0", "--set", "documentation_updates_needed=0", "--out", outDir, "--root", root]);
assert.equal(b.json.oversize, true);
const dry = runScript("plan2code-4-finalize", "feedback-payload.mjs", ["submit", "--from", outDir, "--dry-run"]);
assert.equal(dry.code, 0);
assert.equal(dry.json.oversize, true);
assert.equal(dry.json.urlBytes, b.json.urlBytes);
// gh must never succeed here: no PATH but node's own folder (so gh is not
// found), and should one turn up anyway, an empty config, no token and a
// host that does not resolve.
const env = Object.fromEntries(Object.entries(process.env).filter(([k]) => !/^(path|gh_|github_token$)/i.test(k)));
Object.assign(env, { PATH: path.dirname(process.execPath), GH_CONFIG_DIR: tmp("fb5-gh"), GH_HOST: "gh.invalid", PLAN2CODE_NOW: NOW });
const r = spawnSync(process.execPath, [path.join(SKILLS, "plan2code-4-finalize", "scripts", "feedback-payload.mjs"), "submit", "--from", outDir], { cwd: root, encoding: "utf8", timeout: 70000, env });
assert.equal(r.status, 4, r.stderr);
const j = JSON.parse(r.stdout);
assert.equal(j.error, "oversize");
assert.match(j.next, /--truncate/);
assert.match(j.next, /gh/);
});
+55
View File
@@ -0,0 +1,55 @@
// Tests for plan2code-handoff's scripts/handoff-path.mjs.
//
// See skill-script-helpers.mjs for how these run.
import { test } from "node:test";
import assert from "node:assert/strict";
import fs from "node:fs";
import path from "node:path";
import { ROOT, SKILLS, NOW, runScript, tmp, write, read, gitInit, phase } from "./skill-script-helpers.mjs";
/* =========================================================== handoff-path.mjs */
test("handoff-path: timestamped name, no overwrite, git ignore check", () => {
const repo = tmp("handoff");
gitInit(repo);
const H = (...a) => runScript("plan2code-handoff", "handoff-path.mjs", a, { cwd: repo });
const first = H("--dir", "handoffs");
assert.equal(first.code, 0, first.stderr);
assert.equal(path.basename(first.json.path), "2026-08-03-1405-handoff.md");
assert.equal(first.json.created, true);
assert.equal(first.json.git.inRepo, true);
assert.equal(first.json.git.ignored, false);
assert.match(first.json.warning, /handoffs\//);
fs.writeFileSync(first.json.path, "x");
assert.equal(path.basename(H("--dir", "handoffs").json.path), "2026-08-03-1405-2-handoff.md");
write(repo, ".gitignore", "handoffs/\n");
assert.equal(H("--dir", "handoffs").json.git.ignored, true);
const temp = H();
assert.equal(temp.json.git.inRepo, false);
assert.equal(temp.json.git.ignored, null);
const dry = runScript("plan2code-handoff", "handoff-path.mjs", ["--dir", "later", "--dry-run"], { cwd: repo });
assert.equal(dry.json.created, false);
assert.ok(!fs.existsSync(path.join(repo, "later")));
});
/* ================================================= verification regressions */
test("handoff-path: the repo root itself is never the line to ignore", () => {
const repo = tmp("handoff2");
gitInit(repo);
const j = runScript("plan2code-handoff", "handoff-path.mjs", ["--dir", "."], { cwd: repo }).json;
assert.equal(j.git.inRepo, true);
assert.equal(j.git.ignoreLine, "*-handoff.md");
});
test("handoff-path: --dir naming an existing file is refused", () => {
const repo = tmp("handoff3");
write(repo, "notes.txt", "x");
const r = runScript("plan2code-handoff", "handoff-path.mjs", ["--dir", "notes.txt"], { cwd: repo });
assert.equal(r.code, 5);
assert.equal(r.json.ok, false);
assert.equal(r.json.error, "not-a-folder");
assert.match(r.json.next, /--dir temp/);
assert.equal(runScript("plan2code-handoff", "handoff-path.mjs", ["--dir", "notes.txt", "--dry-run"], { cwd: repo }).code, 5, "dry-run refuses it too");
});
+545
View File
@@ -0,0 +1,545 @@
// Tests for plan2code-0-pathfinder's scripts/pathfinder.mjs.
//
// See skill-script-helpers.mjs for how these run.
import { test } from "node:test";
import assert from "node:assert/strict";
import fs from "node:fs";
import path from "node:path";
import { ROOT, SKILLS, NOW, runScript, tmp, write, read, gitInit, phase } from "./skill-script-helpers.mjs";
/* ============================================================ pathfinder.mjs */
const B = "> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.";
function q(name, { type = "grill · HITL", state = "open", blocked = "none", claimed = "none", resolved = "none", locked = "no", answer = null, evidence = null, skipResolved = false } = {}) {
return [
B, "", `# ${name}`, "", `Type: ${type}`, `State: ${state}`, `Blocked by: ${blocked}`, `Claimed: ${claimed}`,
...(skipResolved ? [] : [`Resolved: ${resolved}`]), `Locked: ${locked}`, "", "## Question", "", `What about ${name}?`, "",
...(answer ? ["## Answer", "", answer, ""] : []),
...(evidence ? ["## Evidence", "", evidence, ""] : []),
].join("\n");
}
const MAP = `${B}
# Map: audit-export
*Decisions live in \`questions/\` — one file each. This file is just the index.*
**Status:** Working
**Updated:** 2026-08-01
**Confidence:** Requirements-clarity 19/25 · Feasibility-technical 14/25 · Integration-points 20/25 · Risk-assessment 18/25
## Destination
A locked plan for exporting audit events.
## Ground rules
- \`AGENTS.md\` exists and governs.
## Question Checklist
<!-- Rebuilt from questions/ every session. -->
- [x] [Codebase context](./questions/00-codebase-context.md) — Node and Postgres; a long gist
that wraps onto a second line. *(2026-08-01)*
- [ ] [Export format](./questions/01-export-format.md)
- [ ] [Row-count
ceiling](./questions/02-row-count-ceiling.md)
- [ ] [Testing posture](./questions/03-testing-posture.md)
- [ ] [Delivery channel](./questions/04-delivery-channel.md)
- [ ] [Gone question](./questions/09-gone.md)
## Not yet specified
- How far back an export may reach.
## Out of scope
- Redesign of the schema.
`;
function mapFixture() {
const root = tmp("pf");
write(root, "AGENTS.md", "# AGENTS.md\n");
const pf = "specs/audit-export/pathfinder";
write(root, `${pf}/map.md`, MAP);
write(root, `${pf}/questions/00-codebase-context.md`, q("Codebase context", { type: "legwork · AFK", state: "resolved", claimed: "2026-08-01 09:00", resolved: "2026-08-01", answer: "Node.\n\n**Gist:** Node and Postgres." }));
// An answer under State: open (a crash after writing the answer): reconcile resolves it and backfills the date.
write(root, `${pf}/questions/01-export-format.md`, q("Export format", { claimed: "2026-08-02 10:00", locked: "yes", answer: "**Decision.** CSV.\n\n**Rejected**\n\n- *JSONL* - recipients use Excel.\n\n**Gist:** CSV with a BOM.", skipResolved: true }));
// A stale claim with no answer: reset.
write(root, `${pf}/questions/02-row-count-ceiling.md`, q("Row-count ceiling", { type: "research · AFK", state: "claimed", claimed: "2026-08-02 11:00", evidence: "- half done" }));
write(root, `${pf}/questions/03-testing-posture.md`, q("Testing posture"));
write(root, `${pf}/questions/04-delivery-channel.md`, q("Delivery channel", { blocked: "02" }));
write(root, `${pf}/questions/05-siem-push.md`, q("SIEM push", { state: "out-of-scope" }));
return { root, pf };
}
test("pathfinder reconcile: files win, dates backfill, stale claims reset, rows keep their words", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
const dry = P("reconcile", "audit-export", "--dry-run");
assert.equal(dry.code, 0, dry.stderr);
assert.equal(read(root, `${pf}/map.md`), MAP, "dry-run writes nothing");
const r = P("reconcile", "audit-export");
assert.deepEqual(r.json.repairs.map((x) => [x.nn, x.repair]), [["01", "state-resolved"], ["01", "resolved-backfilled"], ["02", "stale-claim-reset"]]);
assert.match(r.json.repairs[1].detail, /2026-08-02 \(from Claimed:\)/);
assert.deepEqual(r.json.refire.map((x) => x.nn), ["02"], "research with no Research complete line is re-fired");
assert.equal(r.json.refire[0].partialEvidence, true);
const map = read(root, `${pf}/map.md`);
assert.match(map, /- \[x\] \[Codebase context\]\(\.\/questions\/00-codebase-context\.md\) — Node and Postgres; a long gist\n that wraps onto a second line\. \*\(2026-08-01\)\*/, "an unchanged row keeps its own wrapped text");
assert.match(map, /- \[x\] \[Export format\]\(\.\/questions\/01-export-format\.md\) — CSV with a BOM\. \*\(2026-08-02\)\*/);
assert.match(map, /- \[ \] \[Row-count\n ceiling\]\(\.\/questions\/02-row-count-ceiling\.md\)/, "a wrapped link is still one row");
assert.match(map, /- \[!\] \[Delivery channel\]\(\.\/questions\/04-delivery-channel\.md\) — Blocked by 02/);
assert.match(map, /- \[-\] \[SIEM push\]\(\.\/questions\/05-siem-push\.md\) — out of scope, see below/);
assert.ok(!map.includes("09-gone"), "a row with no file is dropped");
assert.match(map, /<!-- Rebuilt from questions\/ every session\. -->/);
const q1 = read(root, `${pf}/questions/01-export-format.md`);
assert.match(q1, /State: resolved\nBlocked by: none\nClaimed: 2026-08-02 10:00\nResolved: 2026-08-02\nLocked: yes/, "Resolved: inserted in schema order");
assert.match(read(root, `${pf}/questions/02-row-count-ceiling.md`), /State: open\nBlocked by: none\nClaimed: none/);
const again = P("reconcile", "audit-export");
assert.deepEqual([again.json.repairs.length, again.json.mapChanges.length, again.json.mapWritten], [0, 0, false], "idempotent");
});
test("pathfinder frontier, claim, resolve, rule-out", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
P("reconcile", "audit-export");
let f = P("frontier", `${pf}`).json;
assert.deepEqual(f.frontier.map((x) => x.nn), ["02", "03"]);
assert.deepEqual(f.blocked.map((x) => [x.nn, x.blockedBy.map((b) => b.nn)]), [["04", ["02"]]]);
assert.equal(f.nextNN, "10", "09 was linked from the map before reconcile dropped its row: never reissued");
assert.match(read(root, `${pf}/map.md`), /^\*\*Confidence:\*\*.*\n\*\*Next NN:\*\* 10$/m, "reconcile keeps the high-water mark at the end of the header block");
assert.deepEqual(f.counts, { resolved: 2, total: 5, outOfScope: 1, fog: 1 });
assert.deepEqual(f.drift, []);
assert.equal(P("claim", "audit-export", "04").code, 5, "blocked");
assert.equal(P("claim", "audit-export", "05").code, 5, "out of scope");
assert.equal(P("claim", "audit-export", "77").code, 3);
const c = P("claim", "audit-export", "2");
assert.equal(c.code, 0, c.stderr);
assert.equal(c.json.claimed, NOW);
assert.match(read(root, `${pf}/questions/02-row-count-ceiling.md`), /State: claimed\nBlocked by: none\nClaimed: 2026-08-03 14:05/);
assert.match(read(root, `${pf}/map.md`), /- \[\/\] \[Row-count\n ceiling\]/);
assert.equal(P("claim", "audit-export", "03").json.error, "claim-open", "one claim at a time");
assert.equal(P("resolve", "audit-export", "02").json.error, "no-answer");
const qp = path.join(root, pf, "questions/02-row-count-ceiling.md");
fs.appendFileSync(qp, "\n## Answer\n\nFacts.\n");
assert.equal(P("resolve", "audit-export", "02").json.error, "no-gist");
fs.appendFileSync(qp, "\n**Gist:** 2M rows at the 99th percentile.\n");
const res = P("resolve", "audit-export", "02");
assert.equal(res.code, 0, res.stderr);
assert.deepEqual(res.json.unblocked, [{ nn: "04", name: "Delivery channel" }]);
const map = read(root, `${pf}/map.md`);
assert.match(map, /- \[x\] \[Row-count ceiling\]\(\.\/questions\/02-row-count-ceiling\.md\) — 2M rows at the 99th percentile\. \*\(2026-08-03\)\*/);
assert.match(map, /- \[ \] \[Delivery channel\]\(\.\/questions\/04-delivery-channel\.md\)\n/, "the Blocked by tail goes with the [!] marker");
assert.match(map, /^\*\*Updated:\*\* 2026-08-03$/m);
const ro = P("rule-out", "audit-export", "03", "--reason", "testing is owned elsewhere");
assert.equal(ro.code, 0, ro.stderr);
assert.match(read(root, `${pf}/questions/03-testing-posture.md`), /State: out-of-scope[\s\S]*## Question\n\nOut of scope: testing is owned elsewhere/);
assert.match(read(root, `${pf}/map.md`), /## Out of scope\n\n- Redesign of the schema\.\n- \[Testing posture\]\(\.\/questions\/03-testing-posture\.md\) — testing is owned elsewhere/);
assert.equal(P("rule-out", "audit-export", "04").code, 2, "--reason is required");
assert.equal(P("rule-out", "audit-export", "00", "--reason", "x").code, 5, "a resolved decision is not ruled out");
});
test("pathfinder trail: exact footer for forms A, B and C", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
P("reconcile", "audit-export");
P("claim", "audit-export", "03");
const b = P("trail", "audit-export", "--form", "B", "--item", "Q1 Types", "--item", "Q2 Cadence", "--text");
assert.equal(b.code, 0, b.stderr);
assert.equal(b.stdout, [
"🧭 audit-export · Working · 2/5 cleared",
" START ●━●··○━◉··⊘··⊝ ····⚑",
" ● done · ◉ here · ○ open · ⊘ blocked · ⊝ out of scope · ⚑ destination · ~1 fog",
" 1 Codebase context ✔ · 2 Export format ✔ · 3 Row-count ceiling · 4 Testing posture ◀ here",
" 5 Delivery channel (blocked:3) · 6 SIEM push (out of scope)",
" Confidence: not yet — Feasibility still needs work.",
"",
"WAITING ON YOU · answer here, in this conversation:",
"Q1 Types · Q2 Cadence",
"",
].join("\n"));
assert.equal(P("trail", "audit-export", "--form", "B").code, 2, "Form B names its items");
assert.equal(P("trail", "audit-export", "--form", "B", "--item", "1", "--item", "2", "--item", "3", "--item", "4").code, 5, "three-probe cap");
assert.equal(P("trail", "audit-export", "--form", "C").json.error, "claim-open", "no menu while a claim is open");
const a = P("trail", "audit-export", "--form", "A", "--text").stdout;
assert.ok(a.endsWith("NEXT STEP · start a new conversation and run:\n`/plan2code-0-pathfinder specs/audit-export/pathfinder --web`\n"), a);
fs.appendFileSync(path.join(root, pf, "questions/03-testing-posture.md"), "\n## Answer\n\nUnit; `Run after each phase`; `Critical paths`.\n\n**Gist:** Unit tests, run after each phase.\n");
P("resolve", "audit-export", "03");
const c = P("trail", "audit-export", "--form", "C", "--resolved-this-session", "3");
assert.equal(c.code, 0, c.stderr);
assert.ok(c.json.text.endsWith("OR START FRESH · new conversation, paste:\n`/plan2code-0-pathfinder specs/audit-export/pathfinder --web`"));
assert.deepEqual(c.json.menu.options.map((o) => o.label), ["Row-count ceiling", "Start fresh"]);
assert.equal(c.json.menu.recommended, "Start fresh", "the checkpoint after ~3 decisions");
assert.equal(P("trail", "audit-export", "--form", "C").json.menu.recommended, "Row-count ceiling");
});
test("pathfinder trail: a cleared map arrives; init first only while AGENTS.md is still missing", () => {
const root = tmp("pfc");
const pf = "specs/done/pathfinder";
write(root, `${pf}/map.md`, MAP.replace("**Status:** Working", "**Status:** Cleared").replace("- `AGENTS.md` exists and governs.", "- `AGENTS.md` is absent; continued without it.").replace(/## Not yet specified\n\n- How far back an export may reach\.\n/, "## Not yet specified\n\n").replace("Feasibility-technical 14/25", "Feasibility-technical 22/25"));
write(root, `${pf}/questions/00-codebase-context.md`, q("Codebase context", { state: "resolved", resolved: "2026-08-01", answer: "**Gist:** Node." }));
write(root, `${pf}/questions/01-export-format.md`, q("Export format", { state: "resolved", resolved: "2026-08-02", answer: "**Gist:** CSV." }));
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
P("reconcile", "done");
assert.equal(P("trail", "done", "--form", "A", "--text").stdout, [
"🧭 done · Cleared · 2/2 cleared",
" START ●━●━⚑ arrived",
" Confidence: solid, but Requirements, Risk are borderline.",
"",
"NEXT STEP · start a new conversation and run:",
"`/plan2code-init`, then `/plan2code-1-plan --web`",
"",
].join("\n"));
write(root, "AGENTS.md", "# AGENTS.md\n");
assert.match(P("trail", "done", "--form", "A", "--text").stdout, /run:\n`\/plan2code-1-plan --web`\n$/);
assert.equal(P("trail", "done", "--form", "C").code, 5, "no menu on a cleared map");
});
test("pathfinder gate: mechanical checks and hard caps", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
P("reconcile", "audit-export");
let g = P("gate", "audit-export").json;
assert.equal(g.mechanicalPass, false);
const failed = g.checks.filter((c) => !c.pass).map((c) => c.id);
assert.deepEqual(failed, ["no-open-claimed-blocked", "fog-empty", "hard-caps", "scores-at-least-18"]);
assert.ok(g.caps.some((c) => c.dimension === "Requirements" && /not resolved/.test(c.why)));
assert.ok(g.caps.some((c) => c.dimension === "Risk" && /Export format records no consequences/.test(c.why)));
assert.equal(g.planDraftPath, "specs/audit-export/PLAN-DRAFT-20260803.md");
// Settle everything.
const qd = path.join(root, pf, "questions");
fs.writeFileSync(path.join(qd, "01-export-format.md"), q("Export format", { state: "resolved", resolved: "2026-08-02", locked: "yes", answer: "CSV.\n\n**Consequences**\n\n- Excel.\n\n**Gist:** CSV." }));
fs.writeFileSync(path.join(qd, "02-row-count-ceiling.md"), q("Row-count ceiling", { type: "research · AFK", state: "resolved", resolved: "2026-08-02", answer: "**Gist:** 2M." }));
fs.writeFileSync(path.join(qd, "03-testing-posture.md"), q("Testing posture", { state: "resolved", resolved: "2026-08-02", answer: "Unit and integration. `Dedicated phase only`. `Moderate (~60-80%)`.\n\n**Gist:** Unit." }));
fs.writeFileSync(path.join(qd, "04-delivery-channel.md"), q("Delivery channel", { state: "resolved", resolved: "2026-08-03", blocked: "02", answer: "**Gist:** Link." }));
const mp = path.join(root, pf, "map.md");
fs.writeFileSync(mp, fs.readFileSync(mp, "utf8").replace("- How far back an export may reach.\n", "").replace("Feasibility-technical 14/25", "Feasibility-technical 20/25"));
P("reconcile", "audit-export");
g = P("gate", "audit-export").json;
assert.deepEqual(g.checks.filter((c) => !c.pass).map((c) => c.id), ["hard-caps"]);
assert.match(g.caps[0].why, /research 02 Row-count ceiling is resolved with an empty ## Evidence/);
fs.appendFileSync(path.join(qd, "02-row-count-ceiling.md"), "\n## Evidence\n\n- Query plan.\n\n**Research complete:** 2026-08-02\n");
g = P("gate", "audit-export").json;
assert.equal(g.mechanicalPass, true, JSON.stringify(g.checks));
assert.ok(g.judgment.length >= 2, "the gate names what stays judgment");
});
test("pathfinder lint: loop tokens, checkboxes, reserved names, schema, cycles, scraper bait", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
P("reconcile", "audit-export");
let l = P("lint", "audit-export");
assert.equal(l.code, 0, JSON.stringify(l.json.problems));
const qd = path.join(root, pf, "questions");
fs.appendFileSync(path.join(qd, "03-testing-posture.md"), "\n- [ ] a checkbox\nTASK_COMPLETE\n");
fs.writeFileSync(path.join(qd, "06-loop-a.md"), q("Loop A", { blocked: "07" }));
fs.writeFileSync(path.join(qd, "07-loop-b.md"), q("Loop B", { blocked: "06", type: "chat · HITL" }));
write(root, `${pf}/overview.md`, "nope");
write(root, `${pf}/briefs/brief-20260803.md`, "Requirements 18 and 90%\n");
write(root, "specs/audit-export/PLAN-DRAFT-20260803.md", "**Charted by:** `/plan2code-0-pathfinder`\n**Confidence:** 85%\n| Requirements | 11 |\n");
l = P("lint", "audit-export");
assert.equal(l.code, 4);
const msgs = l.json.problems.map((p) => `${p.file}: ${p.message}`).join("\n");
for (const re of [/03-testing-posture\.md: has a checkbox/, /03-testing-posture\.md: contains the loop completion token TASK_COMPLETE/, /cycle 06 -> 07 -> 06|cycle 07 -> 06 -> 07/, /Type: "chat · HITL"/, /overview\.md is a reserved name/, /brief-20260803\.md: metrics-scraper bait \(bare Requirements/, /brief-20260803\.md: contains %/, /PLAN-DRAFT-20260803\.md: metrics-scraper bait \(overall confidence with %/, /PLAN-DRAFT-20260803\.md: metrics-scraper bait \(bare Requirements/]) {
assert.match(msgs, re);
}
});
test("pathfinder brief-data: range filtering, frontier notes, plain sections", () => {
const { root } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
P("reconcile", "audit-export");
const today = P("brief-data", "audit-export").json;
assert.equal(today.file, "specs/audit-export/pathfinder/briefs/brief-20260803.md");
assert.deepEqual(today.decided, []);
assert.equal(today.note, "No decisions were recorded in this period.");
const full = P("brief-data", "audit-export", "--range", "full").json;
assert.deepEqual(full.decided.map((d) => d.name), ["Codebase context", "Export format"]);
assert.deepEqual(full.decided[1].rejected, ["*JSONL* - recipients use Excel."]);
assert.equal(full.decided[1].hardToReverse, true);
assert.equal(full.progress, "2 of 5 decisions made");
assert.deepEqual(full.waiting, [{ name: "Delivery channel", waitingOn: ["Row-count ceiling"] }]);
assert.deepEqual(full.fog, ["How far back an export may reach."]);
assert.deepEqual(full.ruledOut, ["Redesign of the schema."]);
assert.equal(full.confidence, "Confidence: not yet — Feasibility still needs work.");
const since = P("brief-data", "audit-export", "--range", "since:2026-08-02").json;
assert.deepEqual(since.decided.map((d) => d.name), ["Export format"]);
assert.deepEqual(P("brief-data", "audit-export", "--range", "2026-08-01..2026-08-01").json.decided.map((d) => d.name), ["Codebase context"]);
assert.equal(P("brief-data", "audit-export", "--range", "2026-08-05..2026-08-01").code, 2);
assert.equal(P("brief-data", "audit-export", "--range", "yesterday").code, 2);
assert.equal(P("brief-data", "nope").code, 3);
});
/* ================================================= verification regressions */
test("pathfinder: a ## Answer inside a fenced sketch is content, not an answer", () => {
const { root, pf } = mapFixture();
const qd = path.join(root, pf, "questions");
fs.writeFileSync(path.join(qd, "03-testing-posture.md"), q("Testing posture", { type: "sketch · HITL", evidence: "```markdown\n## Answer\n\n**Gist:** fake\n```" }));
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
const r = P("reconcile", "audit-export");
assert.ok(!r.json.repairs.some((x) => x.nn === "03"), JSON.stringify(r.json.repairs));
assert.match(read(root, `${pf}/questions/03-testing-posture.md`), /State: open/);
});
test("pathfinder reconcile: an unrepairable file is skipped and named, nothing half-written, its row kept", () => {
const { root, pf } = mapFixture();
write(root, `${pf}/questions/07-bare.md`, `${B}\n\n# Bare\n\n## Answer\n\nYes.\n`);
const mp = path.join(root, pf, "map.md");
fs.writeFileSync(mp, fs.readFileSync(mp, "utf8").replace("- [ ] [Gone question]", "- [ ] [Bare](./questions/07-bare.md)\n- [ ] [Gone question]"));
const r = runScript("plan2code-0-pathfinder", "pathfinder.mjs", ["reconcile", "audit-export", "--root", root]);
assert.equal(r.code, 0, r.stderr);
assert.deepEqual(r.json.findings.map((f) => [f.file, f.finding]), [["07-bare.md", "unrepairable"]]);
assert.match(read(root, `${pf}/questions/01-export-format.md`), /State: resolved/, "the other repairs still landed");
assert.equal(read(root, `${pf}/questions/07-bare.md`), `${B}\n\n# Bare\n\n## Answer\n\nYes.\n`, "left untouched");
assert.ok(!r.json.mapChanges.some((c) => c.file === "07-bare.md"), JSON.stringify(r.json.mapChanges));
assert.match(read(root, `${pf}/map.md`), /^- \[.\] \[Bare\]\(\.\/questions\/07-bare\.md\)$/m, "its row survives");
});
test("pathfinder reconcile: rows stay in place among grouping lines, a re-opened row loses its gist, stray claims clear", () => {
const { root, pf } = mapFixture();
const mp = path.join(root, pf, "map.md");
fs.writeFileSync(mp, fs.readFileSync(mp, "utf8").replace("- [ ] [Export format]", "### Group two\n\n- [x] [Export format]").replace("- [ ] [Export format](./questions/01-export-format.md)", "- [x] [Export format](./questions/01-export-format.md) — old gist *(2026-07-01)*"));
fs.writeFileSync(path.join(root, pf, "questions/01-export-format.md"), q("Export format", { claimed: "2026-08-02 10:00" }));
const r = runScript("plan2code-0-pathfinder", "pathfinder.mjs", ["reconcile", "audit-export", "--root", root]);
assert.ok(r.json.repairs.some((x) => x.nn === "01" && x.repair === "stray-claim-cleared"));
const map = read(root, `${pf}/map.md`);
assert.match(map, /### Group two\n\n- \[ \] \[Export format\]\(\.\/questions\/01-export-format\.md\)\n/, "the group heading keeps its place and the row its old tail is gone");
});
test("pathfinder gate: testing posture by keyword (never from Rejected), research-style consequences count", () => {
const { root, pf } = mapFixture();
const qd = path.join(root, pf, "questions");
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
fs.writeFileSync(path.join(qd, "03-testing-posture.md"), q("Testing posture", { state: "resolved", resolved: "2026-08-02", answer: "Unit + integration, run after each phase, moderate coverage.\n\n**Gist:** Unit." }));
fs.writeFileSync(path.join(qd, "01-export-format.md"), q("Export format", { state: "resolved", resolved: "2026-08-02", locked: "yes", answer: "### What this implies\n\n- Excel only.\n\n**Gist:** CSV." }));
P("reconcile", "audit-export");
let caps = P("gate", "audit-export").json.caps;
assert.ok(!caps.some((c) => c.dimension === "Requirements"), JSON.stringify(caps));
assert.ok(!caps.some((c) => c.dimension === "Risk"), JSON.stringify(caps));
fs.writeFileSync(path.join(qd, "03-testing-posture.md"), q("Testing posture", { state: "resolved", resolved: "2026-08-02", answer: "Unit tests.\n\n**Rejected**\n\n- Run after each phase with Moderate coverage: too slow.\n\n**Gist:** Unit." }));
caps = P("gate", "audit-export").json.caps;
assert.ok(caps.some((c) => c.dimension === "Requirements" && /cadence/.test(c.why)), "a cadence named only as rejected does not count");
});
test("pathfinder: Form C needs a frontier; names come from the rows; NN is never handed out twice", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
P("reconcile", "audit-export");
const mp = path.join(root, pf, "map.md");
fs.writeFileSync(mp, fs.readFileSync(mp, "utf8").replace("[Testing posture](", "[How we test]("));
const c = P("trail", "audit-export", "--form", "C").json;
assert.deepEqual(c.menu.options.map((o) => o.label), ["Row-count ceiling", "How we test", "Start fresh"]);
fs.rmSync(path.join(root, pf, "questions/05-siem-push.md"));
// Without the high-water line (an older map), the links alone still count.
fs.writeFileSync(mp, fs.readFileSync(mp, "utf8").replace(/^\*\*Next NN:\*\*.*\n/m, ""));
assert.equal(P("frontier", "audit-export").json.nextNN, "06", "05 is still linked from the map");
const qd = path.join(root, pf, "questions");
fs.writeFileSync(path.join(qd, "02-row-count-ceiling.md"), q("Row-count ceiling", { blocked: "04" }));
fs.writeFileSync(path.join(qd, "03-testing-posture.md"), q("Testing posture", { blocked: "04 (needs the channel)" }));
P("reconcile", "audit-export");
const none = P("trail", "audit-export", "--form", "C");
assert.equal(none.code, 5);
assert.equal(none.json.error, "no-frontier");
});
test("pathfinder lint: map %, brief scores and checkboxes, Planning Metrics in a pathfinder draft", () => {
const { root, pf } = mapFixture();
runScript("plan2code-0-pathfinder", "pathfinder.mjs", ["reconcile", "audit-export", "--root", root]);
const mp = path.join(root, pf, "map.md");
fs.writeFileSync(mp, fs.readFileSync(mp, "utf8") + "\n80% sure.\n");
write(root, `${pf}/briefs/brief-20260803.md`, "Solid at 22/25.\n- [ ] todo\n");
write(root, "specs/audit-export/PLAN-DRAFT-20260803.md", "**Charted by:** `/plan2code-0-pathfinder`\n\n## Planning Metrics\n");
const l = runScript("plan2code-0-pathfinder", "pathfinder.mjs", ["lint", "audit-export", "--root", root]).json;
const msgs = l.problems.map((p) => `${p.level} ${p.message}`).join("\n");
assert.match(msgs, /warn contains %; the map/);
assert.match(msgs, /error contains a raw NN\/25 score/);
assert.match(msgs, /error has a checkbox line/);
assert.match(msgs, /error has a ## Planning Metrics section/);
});
test("pathfinder brief-data: calendar dates only, no future since, no NNs or links in what a brief prints", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
P("reconcile", "audit-export");
assert.equal(P("brief-data", "audit-export", "--range", "since:2026-13-99").json.error, "bad-date");
assert.equal(P("brief-data", "audit-export", "--range", "since:2026-09-01").code, 2);
const mp = path.join(root, pf, "map.md");
fs.writeFileSync(mp, fs.readFileSync(mp, "utf8").replace("- Redesign of the schema.", "- [SIEM push](./questions/05-siem-push.md) — not ours.\n- Nothing yet."));
fs.writeFileSync(path.join(root, pf, "questions/04-delivery-channel.md"), q("Delivery channel", { blocked: "09" }));
const j = P("brief-data", "audit-export", "--range", "full").json;
assert.deepEqual(j.ruledOut, ["SIEM push — not ours."]);
assert.deepEqual(j.waiting.find((w) => w.name === "Delivery channel").waitingOn, ["a question that no longer exists"]);
});
/* ===================================================== review-fix regressions */
test("pathfinder: claim, resolve and rule-out honour --dry-run", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
P("reconcile", "audit-export");
const snap = () => fs.readdirSync(path.join(root, pf, "questions")).map((n) => read(root, `${pf}/questions/${n}`)).concat(read(root, `${pf}/map.md`));
const before = snap();
const c = P("claim", "audit-export", "03", "--dry-run");
assert.equal(c.code, 0, c.stderr);
assert.deepEqual([c.json.dryRun, c.json.mapWritten], [true, false]);
assert.deepEqual(snap(), before, "claim --dry-run leaves every file unchanged");
fs.appendFileSync(path.join(root, pf, "questions/03-testing-posture.md"), "\n## Answer\n\nUnit.\n\n**Gist:** Unit.\n");
const withAnswer = snap();
const r = P("resolve", "audit-export", "03", "--dry-run");
assert.equal(r.code, 0, r.stderr);
assert.deepEqual([r.json.dryRun, r.json.mapWritten], [true, false]);
assert.deepEqual(snap(), withAnswer, "resolve --dry-run writes nothing");
const o = P("rule-out", "audit-export", "02", "--reason", "not ours", "--dry-run");
assert.equal(o.code, 0, o.stderr);
assert.deepEqual([o.json.dryRun, o.json.mapWritten], [true, false]);
assert.deepEqual(snap(), withAnswer, "rule-out --dry-run writes nothing");
const real = P("claim", "audit-export", "02");
assert.deepEqual([real.json.dryRun, real.json.mapWritten], [false, true]);
});
test("pathfinder: a deleted top NN is never reissued, linked from an answer or not", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
const qd = path.join(root, pf, "questions");
const mp = path.join(root, pf, "map.md");
// Start from a map with no high-water line and no stale 09 link: reconcile adds the line.
fs.writeFileSync(mp, MAP.replace("- [ ] [Gone question](./questions/09-gone.md)\n", ""));
fs.writeFileSync(path.join(qd, "06-subsumed.md"), q("Subsumed"));
let r = P("reconcile", "audit-export");
assert.deepEqual(r.json.mapChanges.find((c) => c.change === "next-nn"), { change: "next-nn", from: null, to: "07" });
assert.match(read(root, `${pf}/map.md`), /^\*\*Next NN:\*\* 07$/m);
// resolve.md: delete the subsumed file, reconcile drops its row, the absorbing answer links to it.
fs.rmSync(path.join(qd, "06-subsumed.md"));
fs.appendFileSync(path.join(qd, "04-delivery-channel.md"), "\n## Answer\n\nAbsorbed [Subsumed](./questions/06-subsumed.md).\n\n**Gist:** Link.\n");
r = P("reconcile", "audit-export");
assert.ok(r.json.mapChanges.some((c) => c.change === "row-dropped" && c.file === "06-subsumed.md"));
assert.ok(!read(root, `${pf}/map.md`).includes("06-subsumed"), "the map no longer links it");
assert.equal(P("frontier", "audit-export").json.nextNN, "07", "linked from an answer");
// Even with the answer's link gone, the high-water mark holds.
fs.writeFileSync(path.join(qd, "04-delivery-channel.md"), q("Delivery channel", { blocked: "02" }));
assert.equal(P("frontier", "audit-export").json.nextNN, "07", "the **Next NN:** line");
assert.equal(P("reconcile", "audit-export").json.nextNN, "07", "reconcile never lowers it");
assert.equal(P("lint", "audit-export").code, 0, "lint accepts the line");
});
test("pathfinder: a name with brackets round-trips; reconcile is stable", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
write(root, `${pf}/questions/06-retention.md`, q("Retention [legal] hold"));
for (let i = 0; i < 3; i++) assert.equal(P("reconcile", "audit-export").code, 0);
const map = read(root, `${pf}/map.md`);
assert.equal(map.split("\n").filter((l) => l.includes("06-retention.md")).length, 1, map);
assert.match(map, /^- \[ \] \[Retention \[legal\] hold\]\(\.\/questions\/06-retention\.md\)$/m);
assert.deepEqual(P("frontier", "audit-export").json.drift, []);
assert.equal(P("reconcile", "audit-export").json.mapWritten, false);
});
test("pathfinder: an author's sub-bullet under a row survives resolve, rule-out and a marker change", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
const mp = path.join(root, pf, "map.md");
fs.writeFileSync(mp, fs.readFileSync(mp, "utf8")
.replace("- [ ] [Testing posture](./questions/03-testing-posture.md)", "- [ ] [Testing posture](./questions/03-testing-posture.md)\n - note: ask Dana first")
.replace("- [ ] [Delivery channel](./questions/04-delivery-channel.md)", "- [ ] [Delivery channel](./questions/04-delivery-channel.md)\n - note: depends on the volume\n and wraps"));
P("reconcile", "audit-export");
let map = read(root, `${pf}/map.md`);
assert.match(map, /- \[!\] \[Delivery channel\]\(\.\/questions\/04-delivery-channel\.md\) — Blocked by 02\n - note: depends on the volume\n and wraps\n/, "kept under a regenerated [!] row");
P("claim", "audit-export", "03");
fs.appendFileSync(path.join(root, pf, "questions/03-testing-posture.md"), "\n## Answer\n\nUnit.\n\n**Gist:** Unit tests only.\n");
assert.equal(P("resolve", "audit-export", "03").code, 0);
map = read(root, `${pf}/map.md`);
assert.match(map, /- \[x\] \[Testing posture\]\(\.\/questions\/03-testing-posture\.md\) — Unit tests only\. \*\(2026-08-03\)\*\n - note: ask Dana first\n/, "kept on resolve");
assert.equal(P("rule-out", "audit-export", "04", "--reason", "not ours").code, 0);
map = read(root, `${pf}/map.md`);
assert.match(map, /- \[-\] \[Delivery channel\]\(\.\/questions\/04-delivery-channel\.md\) — out of scope, see below\n - note: depends on the volume\n and wraps\n/, "kept on rule-out");
assert.deepEqual(P("frontier", "audit-export").json.drift, []);
});
test("pathfinder: AGENTS.md absent reads in the canonical and the skill's own wording", () => {
for (const line of ["- `AGENTS.md` is absent.", "- No `AGENTS.md`.", "- No AGENTS.md; init skipped."]) {
const { root, pf } = mapFixture();
fs.rmSync(path.join(root, "AGENTS.md"));
const mp = path.join(root, pf, "map.md");
fs.writeFileSync(mp, fs.readFileSync(mp, "utf8").replace("- `AGENTS.md` exists and governs.", line));
const g = runScript("plan2code-0-pathfinder", "pathfinder.mjs", ["gate", "audit-export", "--root", root]).json;
assert.deepEqual([g.agentsAbsent, g.initFirst], [true, true], line);
}
});
test("pathfinder lint: a failure carries error, message and next, on stderr too", () => {
const { root, pf } = mapFixture();
runScript("plan2code-0-pathfinder", "pathfinder.mjs", ["reconcile", "audit-export", "--root", root]);
fs.appendFileSync(path.join(root, pf, "questions/03-testing-posture.md"), "\n- [ ] a checkbox\n");
const l = runScript("plan2code-0-pathfinder", "pathfinder.mjs", ["lint", "audit-export", "--root", root]);
assert.equal(l.code, 4);
assert.deepEqual([l.json.ok, l.json.error, l.json.message, l.json.next], [false, "lint-failed", "lint found 1 error.", "Fix each error in problems, then rerun lint."]);
assert.equal(l.json.errors, 1);
assert.equal(l.json.problems.filter((p) => p.level === "error").length, 1);
assert.match(l.stderr, /^error: lint found 1 error\.\nnext: Fix each error in problems, then rerun lint\.\n$/);
});
test("pathfinder lint: densely blocked questions check in linear time", () => {
const root = tmp("pfd");
const pf = "specs/dense/pathfinder";
write(root, `${pf}/map.md`, MAP);
write(root, `${pf}/questions/00-codebase-context.md`, q("Codebase context", { type: "legwork · AFK", state: "resolved", resolved: "2026-08-01", answer: "**Gist:** Node." }));
for (let i = 1; i < 40; i++) {
const nn = String(i).padStart(2, "0");
const blocked = i === 1 ? "00" : [i - 1, i - 2].map((n) => String(n).padStart(2, "0")).join(", ");
write(root, `${pf}/questions/${nn}-q${nn}.md`, q(`Q ${nn}`, { blocked }));
}
const t0 = Date.now();
const l = runScript("plan2code-0-pathfinder", "pathfinder.mjs", ["lint", "dense", "--root", root]);
const ms = Date.now() - t0;
assert.ok(!(l.json.problems || []).some((p) => /cycle/.test(p.message)), JSON.stringify(l.json.problems));
assert.ok(ms < 2000, `lint took ${ms} ms`);
// A cycle is still found, once, in the same format.
write(root, `${pf}/questions/01-q01.md`, q("Q 01", { blocked: "39" }));
const c = runScript("plan2code-0-pathfinder", "pathfinder.mjs", ["lint", "dense", "--root", root]).json;
const cycles = c.problems.filter((p) => /cycle/.test(p.message));
assert.ok(cycles.length >= 1, JSON.stringify(c.problems));
assert.match(cycles[0].message, /^Blocked by: cycle (\d\d)( -> \d\d)+ -> \1;/);
});
test("pathfinder: a one-digit NN is question 0N everywhere, and lint asks for the rename", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
write(root, `${pf}/questions/6-short.md`, q("Short"));
write(root, `${pf}/questions/07-after.md`, q("After", { blocked: "06" }));
P("reconcile", "audit-export");
const f = P("frontier", "audit-export").json;
assert.ok(f.frontier.some((x) => x.nn === "06" && x.name === "Short"), JSON.stringify(f.frontier));
assert.deepEqual(f.blocked.find((x) => x.nn === "07").blockedBy, [{ nn: "06", name: "Short", state: "open" }]);
assert.equal(P("claim", "audit-export", "6").json.nn, "06");
const l = P("lint", "audit-export").json;
assert.ok(l.problems.some((p) => p.level === "error" && /6-short\.md$/.test(p.file) && /rename it to 06-short\.md/.test(p.message)), JSON.stringify(l.problems));
});
test("pathfinder: a duplicate map row is reported as drift and dropped by name", () => {
const { root, pf } = mapFixture();
const P = (...a) => runScript("plan2code-0-pathfinder", "pathfinder.mjs", [...a, "--root", root]);
P("reconcile", "audit-export");
const mp = path.join(root, pf, "map.md");
fs.writeFileSync(mp, fs.readFileSync(mp, "utf8").replace("- [ ] [Testing posture](./questions/03-testing-posture.md)", "- [ ] [Testing posture](./questions/03-testing-posture.md)\n- [ ] [Testing again](./questions/03-testing-posture.md)"));
assert.ok(P("frontier", "audit-export").json.drift.includes("map row 03-testing-posture.md appears 2 times"));
const r = P("reconcile", "audit-export");
assert.deepEqual(r.json.mapChanges, [{ change: "row-dropped", file: "03-testing-posture.md", name: "Testing again", why: "duplicate row" }]);
assert.equal(read(root, `${pf}/map.md`).split("\n").filter((l) => l.includes("03-testing-posture.md")).length, 1);
assert.deepEqual(P("frontier", "audit-export").json.drift, []);
});
test("pathfinder lint: the chart.md map template lints clean", () => {
const chart = fs.readFileSync(path.join(SKILLS, "plan2code-0-pathfinder", "references", "chart.md"), "utf8");
const template = chart.split("## The `map.md` template")[1].match(/```markdown\r?\n([\s\S]*?)\r?\n```/)[1];
assert.match(template, /^\*\*Next NN:\*\* \d+$/m, "the template carries the high-water line");
const root = tmp("pft");
write(root, "specs/audit-log-export/pathfinder/map.md", template + "\n");
const l = runScript("plan2code-0-pathfinder", "pathfinder.mjs", ["lint", "audit-log-export", "--root", root]);
assert.equal(l.code, 0, JSON.stringify(l.json.problems));
assert.deepEqual([l.json.errors, l.json.warnings], [0, 0], JSON.stringify(l.json.problems));
});
+625
View File
@@ -0,0 +1,625 @@
// Tests for the skill-script plumbing (installer) and the shared scripts in src/skill-scripts/.
//
// See skill-script-helpers.mjs for how these run.
import { test } from "node:test";
import assert from "node:assert/strict";
import fs from "node:fs";
import path from "node:path";
import { spawnSync } from "node:child_process";
import { ROOT, SKILLS, NOW, runScript, tmp, write, read, gitInit, phase } from "./skill-script-helpers.mjs";
/* ================================================================ install */
test("every skill with scripts ships common.mjs and version.json; dashboard ships none", () => {
const withScripts = fs.readdirSync(SKILLS).filter((s) => fs.existsSync(path.join(SKILLS, s, "scripts")));
assert.ok(withScripts.length >= 11, `expected most skills to carry scripts, got ${withScripts.join(", ")}`);
for (const s of withScripts) {
for (const f of ["common.mjs", "version.json"]) assert.ok(fs.existsSync(path.join(SKILLS, s, "scripts", f)), `${s}/scripts/${f}`);
}
assert.ok(!fs.existsSync(path.join(SKILLS, "plan2code", "scripts")), "the dashboard needs no scripts");
const v = JSON.parse(fs.readFileSync(path.join(ROOT, "version.json"), "utf8")).version;
assert.equal(JSON.parse(fs.readFileSync(path.join(SKILLS, "plan2code-4-finalize", "scripts", "version.json"), "utf8")).version, v);
});
test("every script a SKILL.md (or a reference it ships) runs exists beside it, and answers --help", () => {
for (const skill of fs.readdirSync(SKILLS)) {
const dir = path.join(SKILLS, skill);
const texts = [];
const walk = (d) => {
for (const e of fs.readdirSync(d, { withFileTypes: true })) {
const p = path.join(d, e.name);
if (e.isDirectory()) {
if (e.name !== "web-console" && e.name !== "scripts") walk(p);
} else if (e.name.endsWith(".md")) texts.push(fs.readFileSync(p, "utf8"));
}
};
walk(dir);
const named = new Set();
for (const t of texts) for (const m of t.matchAll(/<S>\/([a-z-]+\.mjs)/g)) named.add(m[1]);
for (const script of named) {
const p = path.join(dir, "scripts", script);
assert.ok(fs.existsSync(p), `${skill} names <S>/${script} but scripts/${script} is not shipped`);
const r = spawnSync(process.execPath, [p, "--help"], { encoding: "utf8" });
assert.equal(r.status, 0, `${skill}/${script} --help`);
assert.match(r.stdout, /Exit:/, `${skill}/${script} --help documents its exit codes`);
}
}
});
test("scripts are self-contained: built-ins and siblings only", () => {
for (const dir of [path.join(ROOT, "src", "skill-scripts"), ...fs.readdirSync(path.join(ROOT, "src")).filter((n) => n.endsWith("-scripts")).map((n) => path.join(ROOT, "src", n))]) {
for (const f of fs.readdirSync(dir).filter((n) => n.endsWith(".mjs"))) {
const t = fs.readFileSync(path.join(dir, f), "utf8");
for (const m of t.matchAll(/^import .* from "([^"]+)";$/gm)) {
assert.ok(m[1].startsWith("node:") || m[1] === "./common.mjs", `${f} imports ${m[1]}`);
}
assert.ok(!/[\u0000-\u0008]/.test(t), `${f} contains a control character (an escape lost in an edit?)`);
}
}
});
/* ================================================================ specs.mjs */
const OVERVIEW = `# Lunch vote
## Phase Checklist
- [x] **[Phase 1: Data model](./phase-1.md)** - 3 tasks.
- [/] [Phase 2: API](./phase-2.md) _(2 tasks)_
- [ ] **Phase 3 — Web UI** — see [phase-3.md](./phase-3.md)
- [ ] [Phase 4: Docs](./phase-4.md): readme **[Milestone 1]**
## Parallel Execution Groups
| Group | Phases | Reason |
|-------|--------|--------|
| A | 3, 4 | Separate files |
| B (sequential) | 1, 2 | Not parallel |
## Quick Reference
`;
function specFixture() {
const root = tmp("specs");
write(root, "AGENTS.md", "# AGENTS.md\n");
write(root, "specs/lunch-vote/overview.md", OVERVIEW);
write(root, "specs/lunch-vote/PLAN-DRAFT-20260801.md", "# Plan\n\n**Status:** Complete\n");
write(root, "specs/lunch-vote/phase-1.md", phase(1, "Complete", [["x", "1.1"], ["x", "1.2"], ["x", "1.T"]]));
write(root, "specs/lunch-vote/phase-2.md", phase(2, "In Progress", [["x", "2.1"], [" ", "2.2"]]));
write(root, "specs/lunch-vote/phase-3.md", phase(3, "Not Started", [[" ", "3.1"], [" ", "3.2"], [" ", "3.2a"], [" ", "3.4"]], "\n## Phase Testing\n\n- [ ] Run unit tests\n"));
write(root, "specs/lunch-vote/phase-4.md", phase(4, "Not Started", [["!", "4.1"], ["?", "4.2"]]));
write(root, "specs/idea-only/pathfinder/map.md", "# Map\n\n**Status:** Working\n");
write(root, "specs--completed/old-one/overview.md", "# Old\n");
return root;
}
test("specs list: active specs with state, drafts and archives, never specs--completed", () => {
const root = specFixture();
const r = runScript("plan2code-1-plan", "specs.mjs", ["list", "--root", root]);
assert.equal(r.code, 0, r.stderr);
const names = r.json.specs.map((s) => s.name).sort();
assert.deepEqual(names, ["idea-only", "lunch-vote"]);
const lv = r.json.specs.find((s) => s.name === "lunch-vote");
assert.equal(lv.state, "building");
assert.deepEqual(lv.planDrafts, [{ file: "PLAN-DRAFT-20260801.md", archived: false, status: "Complete" }]);
assert.equal(lv.phaseFiles, 4);
assert.deepEqual(r.json.completed, ["old-one"]);
assert.equal(r.json.hasAgents, true);
});
test("specs status: checklist formats, groups, task markers, numbering, totals, checks", () => {
const root = specFixture();
const r = runScript("plan2code-3-implement", "specs.mjs", ["status", "specs/lunch-vote/overview.md", "--root", root]);
assert.equal(r.code, 0, r.stderr);
const j = r.json;
assert.deepEqual(j.phases.map((p) => [p.number, p.name, p.checklist]), [
[1, "Data model", "complete"],
[2, "API", "in-progress"],
[3, "Web UI", "pending"],
[4, "Docs", "pending"],
]);
assert.deepEqual(j.parallelGroups.map((g) => [g.id, g.phases]), [["A", [3, 4]]], "a group labelled sequential is not a parallel group");
assert.deepEqual(j.totals, { total: 11, complete: 4, assumed: 1, pending: 5, "in-progress": 0, blocked: 1, unknown: 0, completed: 5, incomplete: 5 });
assert.equal(j.completionPercent, 45);
assert.equal(j.phases[0].tasks.total, 3, "a named testing task (1.T) counts");
assert.equal(j.phases[0].goal, "Phase 1 builds the thing.");
const codes = j.checks.map((c) => c.code);
assert.ok(j.checks.some((c) => c.code === "task-numbering" && /expected 3\.3, found 3\.4/.test(c.message)), JSON.stringify(j.checks));
assert.ok(!codes.includes("criteria-checkbox"), "a Phase Testing checkbox block is the template's own");
assert.ok(j.checks.some((c) => c.code === "unchecked-with-done-tasks" && /Phase 4/.test(c.message)), "an assumed [?] task counts as done, so a [ ] phase holding one is flagged");
assert.equal(j.next.action, "ask-resume");
assert.equal(j.next.phase.number, 2);
assert.deepEqual(j.pending.map((p) => p.number), [2, 3, 4]);
});
test("specs status: the implement selection table", () => {
const root = specFixture();
const ov = (m1, m2, m3, m4) => OVERVIEW.replace("- [x] **[Phase 1", `- [${m1}] **[Phase 1`).replace("- [/] [Phase 2", `- [${m2}] [Phase 2`).replace("- [ ] **Phase 3", `- [${m3}] **Phase 3`).replace("- [ ] [Phase 4", `- [${m4}] [Phase 4`);
const next = (...m) => {
write(root, "specs/lunch-vote/overview.md", ov(...m));
return runScript("plan2code-3-implement", "specs.mjs", ["status", "lunch-vote", "--root", root]).json.next;
};
assert.deepEqual(next("x", "x", " ", " ").action, "choose");
assert.deepEqual(next("x", "x", " ", " ").options.map((o) => o.number), [3, 4]);
assert.ok(next("x", "x", "/", "/").warning, "all options already [/] warns about duplicated work");
assert.equal(next("x", " ", " ", " ").action, "auto-start");
assert.equal(next("x", "x", "x", " ").action, "auto-start");
assert.equal(next("x", "x", "x", "x").action, "none");
assert.equal(next("x", "x", "x", "x").reason, "every phase is complete");
});
test("specs status: drift between the checklist and the phase files is an error", () => {
const root = specFixture();
write(root, "specs/lunch-vote/overview.md", OVERVIEW.replace("- [/] [Phase 2", "- [x] [Phase 2"));
write(root, "specs/lunch-vote/phase-5.md", phase(5, "Not Started", [[" ", "5.1"]]));
const j = runScript("plan2code-1b-revise-plan", "specs.mjs", ["status", "lunch-vote", "--root", root]).json;
assert.ok(j.checks.some((c) => c.code === "checked-with-open-tasks" && /Phase 2/.test(c.message)));
assert.ok(j.checks.some((c) => c.code === "missing-checklist-row" && /phase-5/.test(c.message)));
});
test("specs status: errors are actionable and exit-coded", () => {
const root = specFixture();
const miss = runScript("plan2code-1-plan", "specs.mjs", ["status", "nope", "--root", root]);
assert.equal(miss.code, 3);
assert.equal(miss.json.error, "spec-not-found");
assert.match(miss.json.next, /specs\.mjs list/);
assert.equal(runScript("plan2code-1-plan", "specs.mjs", ["frobnicate"]).code, 2);
assert.equal(runScript("plan2code-1-plan", "specs.mjs", ["status", "lunch-vote", "--bogus"]).code, 2);
const archived = runScript("plan2code-4-finalize", "specs.mjs", ["mark", "specs--completed/old-one", "--phase", "1", "--to", "done", "--root", root]);
assert.equal(archived.code, 5);
});
test("specs mark: transitions, the no-reset rule, dry-run and idempotence", () => {
const root = specFixture();
const S = (...a) => runScript("plan2code-3-implement", "specs.mjs", [...a, "--root", root]);
const dry = S("mark", "lunch-vote", "--phase", "3", "--to", "in-progress", "--dry-run");
assert.equal(dry.code, 0);
assert.equal(dry.json.changed, true);
assert.match(read(root, "specs/lunch-vote/overview.md"), /- \[ \] \*\*Phase 3/, "dry-run writes nothing");
assert.equal(S("mark", "lunch-vote", "--phase", "3", "--to", "in-progress").code, 0);
assert.match(read(root, "specs/lunch-vote/overview.md"), /- \[\/\] \*\*Phase 3 — Web UI\*\* — see/, "only the marker changes");
assert.equal(S("mark", "lunch-vote", "--phase", "3", "--to", "in-progress").json.changed, false);
const reset = S("mark", "lunch-vote", "--phase", "3", "--to", "open");
assert.equal(reset.code, 5);
assert.equal(reset.json.error, "no-reset");
const done = S("mark", "lunch-vote", "--phase", "2", "--to", "done");
assert.equal(done.code, 0);
assert.match(read(root, "specs/lunch-vote/overview.md"), /- \[x\] \[Phase 2: API\]/);
assert.match(read(root, "specs/lunch-vote/phase-2.md"), /^\*\*Status:\*\* Complete$/m);
assert.equal(S("mark", "lunch-vote", "--phase", "9", "--to", "done").code, 3);
assert.equal(S("mark", "lunch-vote", "--phase", "1", "--to", "open").code, 0, "re-opening a complete phase (revise-plan) is allowed");
});
test("specs metrics: document and finalize comments, judgment fields from --set", () => {
const root = specFixture();
const S = (...a) => runScript("plan2code-2-document", "specs.mjs", [...a, "--root", root]);
assert.equal(S("metrics", "lunch-vote", "--step", "document").code, 2, "verification_items_added is the agent's number");
write(root, "specs/lunch-vote/phase-3.md", read(root, "specs/lunch-vote/phase-3.md") + "\n<!-- VERIFICATION: Added (FR-2) -->\n");
const doc = S("metrics", "lunch-vote", "--step", "document", "--set", "verification_items_added=1");
assert.equal(doc.code, 0, doc.stderr);
assert.equal(doc.json.comment, '<!-- METRICS_JSON {"step": "document", "total_tasks": 11, "tasks_per_phase": [3, 2, 4, 2], "phase_count": 4, "parallel_groups_identified": 1, "verification_items_added": 1} -->');
assert.equal(doc.json.verificationCommentsFound, 1);
assert.equal(S("metrics", "lunch-vote", "--step", "finalize").code, 2);
const fin = S("metrics", "lunch-vote", "--step", "finalize", "--set", "verification_failures_found=1", "--set", "documentation_updates_needed=2");
assert.equal(fin.json.comment, '<!-- METRICS_JSON {"step": "finalize", "completion_rate_at_audit": 0.45, "tasks_completed": 5, "tasks_total": 11, "verification_failures_found": 1, "documentation_updates_needed": 2} -->');
});
test("specs archive: dry-run, move, refuse to clobber", () => {
const root = specFixture();
const S = (...a) => runScript("plan2code-4-finalize", "specs.mjs", [...a, "--root", root]);
const dry = S("archive", "lunch-vote", "--dry-run");
assert.equal(dry.code, 0);
assert.ok(dry.json.files.includes("overview.md"));
assert.ok(fs.existsSync(path.join(root, "specs/lunch-vote")));
const moved = S("archive", "lunch-vote");
assert.equal(moved.code, 0, moved.stdout);
assert.equal(moved.json.sourceRemoved, true);
assert.equal(moved.json.filesAtTarget, dry.json.files.length);
assert.ok(fs.existsSync(path.join(root, "specs--completed/lunch-vote/phase-4.md")));
write(root, "specs/idea-only/x.md", "x");
fs.mkdirSync(path.join(root, "specs--completed/idea-only"));
const clash = S("archive", "idea-only");
assert.equal(clash.code, 5);
assert.ok(fs.existsSync(path.join(root, "specs/idea-only/x.md")), "nothing moved on refusal");
const after = S("status", "specs--completed/lunch-vote");
assert.equal(after.code, 0, "status reads an archived spec (Finalize after Step 6)");
});
/* ============================================================ commit-msg.mjs */
test("commit-msg: ticket from the branch, the three -m flags, subject rules", () => {
const repo = tmp("commit");
const g = gitInit(repo, "feature/ABC-10968-export-csv");
write(repo, "a.txt", "a");
g("add", "-A");
g("commit", "-q", "-m", "init");
const C = (...a) => runScript("plan2code-review", "commit-msg.mjs", a, { cwd: repo });
const ok = C("--subject", "Add the export endpoint", "--add-all");
assert.equal(ok.code, 0, ok.stderr);
assert.equal(ok.json.ticket, "ABC-10968");
assert.equal(ok.json.command, 'git add -A && git commit -m "Add the export endpoint" -m "ABC-10968" -m "AI Assisted"');
assert.equal(ok.json.message, "Add the export endpoint\n\nABC-10968\nAI Assisted");
assert.equal(C("--subject", "x".repeat(101)).code, 4);
assert.equal(C("--subject", 'say "hi"').code, 4);
assert.equal(C("--subject", "costs $5").code, 4);
assert.equal(C().code, 2);
assert.match(C("--subject", "Fix it", "--files", "src/a.ts", "--files", "my file.md").json.command, /^git add -- src\/a\.ts 'my file\.md' && git commit/);
g("checkout", "-q", "-b", "chore/no-ticket-here");
const none = C("--subject", "Fix it");
assert.equal(none.code, 4);
assert.equal(none.json.error, "no-ticket");
assert.equal(C("--subject", "Fix it", "--ticket", "ABC-1").json.ticket, "ABC-1");
assert.equal(C("--subject", "Fix it", "--ticket", "abc-1").code, 4);
g("checkout", "-q", "-b", "PROJ2-7_underscore");
assert.equal(C("--subject", "Fix it").json.ticket, "PROJ2-7");
const notRepo = runScript("plan2code-review", "commit-msg.mjs", ["--subject", "x"], { cwd: tmp("norepo") });
assert.equal(notRepo.code, 3);
});
/* ============================================================ agent-files.mjs */
test("agent-files: detect, apply only what was chosen, rules folders, idempotent", () => {
const root = tmp("agentfiles");
write(root, "AGENTS.md", "# AGENTS.md\n");
write(root, "CLAUDE.md", Array.from({ length: 12 }, (_, i) => `line ${i}`).join("\n") + "\n");
write(root, "GEMINI.md", "# Gemini\nshort\n");
write(root, ".github/copilot-instructions.md", "copilot\n");
write(root, ".cursor/rules/a.md", "a\n");
write(root, ".cursor/rules/b.md", "b\n");
const A = (...a) => runScript("plan2code-init", "agent-files.mjs", [...a, "--root", root]);
const d = A("detect");
assert.equal(d.code, 0);
assert.deepEqual(d.json.found.map((f) => f.id), ["CLAUDE.md", "GEMINI.md", ".github/copilot-instructions.md", ".cursor/rules"]);
assert.equal(d.json.found[0].lines, 12);
assert.equal(d.json.found[0].custom, true);
assert.equal(d.json.found[1].custom, false);
assert.match(d.json.warnings[0], /CLAUDE\.md has custom content/);
const dry = A("apply", "CLAUDE.md", "--dry-run");
assert.equal(dry.code, 0);
assert.match(read(root, "CLAUDE.md"), /^line 0/, "dry-run writes nothing");
assert.equal(A("apply", "CLAUDE.md", ".cursor/rules/", ".github/copilot-instructions.md").code, 0);
assert.match(read(root, "CLAUDE.md"), /CRITICAL — MANDATORY FIRST STEP: You MUST read \[AGENTS\.md\]\(\.\/AGENTS\.md\)/);
assert.match(read(root, ".github/copilot-instructions.md"), /^# Copilot Instructions\n\nSee \[AGENTS\.md\]\(\.\.\/AGENTS\.md\)/);
for (const f of ["CLAUDE.md", ".github/copilot-instructions.md"]) {
assert.match(read(root, f), /for complete project documentation including:\n- Development commands and setup\n[\s\S]*- Keeping this file current \/ Failure log\n- Section details in \.agents-docs\//, `${f} uses the bullet-list pointer`);
}
assert.deepEqual(fs.readdirSync(path.join(root, ".cursor/rules")), ["reference.md"]);
assert.match(read(root, ".cursor/rules/reference.md"), /\]\(\.\.\/\.\.\/AGENTS\.md\)/);
assert.match(read(root, "GEMINI.md"), /^# Gemini/, "unchosen files are untouched");
const again = A("apply", "CLAUDE.md");
assert.deepEqual(again.json.actions, [{ action: "unchanged", path: "CLAUDE.md" }]);
assert.equal(A("detect").json.found.find((f) => f.id === "CLAUDE.md").alreadyPointer, true);
fs.writeFileSync(path.join(root, "CLAUDE.md"), read(root, "CLAUDE.md").replace(/\n/g, "\r\n"));
assert.equal(A("detect").json.found.find((f) => f.id === "CLAUDE.md").alreadyPointer, true, "a CRLF copy of the pointer still counts");
assert.equal(A("apply", "GEMINI.md", "nope.md").code, 3);
fs.rmSync(path.join(root, "AGENTS.md"));
assert.equal(A("apply", "--all").code, 3);
const empty = runScript("plan2code-init", "agent-files.mjs", ["detect", "--root", tmp("none")]);
assert.deepEqual(empty.json.found, []);
assert.match(empty.json.note, /skip this step silently/);
});
/* ============================================================== agents-md.mjs */
test("agents-md: structure, budget, orphans, broken links, inline rules", () => {
const root = tmp("agentsmd");
write(root, "AGENTS.md", [
"# AGENTS.md", "", "## Project Overview", "", "Thing.", "",
"## Architecture", "", "Summary.", "", "Details: [Architecture](./.agents-docs/AGENTS-architecture.md)", "",
"## Testing", "", "Details: [Testing](./.agents-docs/AGENTS-testing.md)", "",
"## Failure log", "", "Details: [Git](./.agents-docs/AGENTS-git.md)", "",
"```", "## not a heading inside a fence", "```", "",
].join("\n"));
write(root, ".agents-docs/AGENTS-architecture.md", "# Architecture\n> Part of [AGENTS.md](../AGENTS.md) — project guidance for AI coding agents.\n");
write(root, ".agents-docs/AGENTS-git.md", "# Git\nno breadcrumb\n");
write(root, ".agents-docs/AGENTS-orphan.md", "# Orphan\n> Part of [AGENTS.md](../AGENTS.md) — project guidance for AI coding agents.\n");
const r = runScript("plan2code-init-update", "agents-md.mjs", ["inspect", "--root", root]);
assert.equal(r.code, 0, r.stderr);
const j = r.json;
assert.equal(j.structure, "progressive");
assert.equal(j.budget, `${j.lines}/500`);
assert.deepEqual(j.sections.map((s) => s.name), ["Project Overview", "Architecture", "Testing", "Failure log"]);
assert.deepEqual(j.brokenLinks, [".agents-docs/AGENTS-testing.md"]);
assert.deepEqual(j.orphans, [".agents-docs/AGENTS-orphan.md"]);
assert.ok(j.problems.some((p) => /Failure log must stay inline/.test(p)));
assert.ok(j.problems.some((p) => /AGENTS-git\.md lacks the/.test(p)));
const missing = runScript("plan2code-init", "agents-md.mjs", ["inspect", "--root", tmp("noagents")]);
assert.equal(missing.json.exists, false);
const fn = runScript("plan2code-init", "agents-md.mjs", ["filename", "Development Commands"]);
assert.equal(fn.json.file, ".agents-docs/AGENTS-development-commands.md");
assert.equal(fn.json.link, "Details: [Development Commands](./.agents-docs/AGENTS-development-commands.md)");
});
/* ============================================================ review-scope.mjs */
test("review-scope: base detection, committed + staged + untracked, doc share", () => {
const repo = tmp("scope");
const g = gitInit(repo, "main");
write(repo, "src/a.js", "1\n2\n");
g("add", "-A");
g("commit", "-q", "-m", "init");
g("checkout", "-q", "-b", "feature/X-1-thing");
write(repo, "src/a.js", "1\n2\n3\n");
write(repo, "README.md", "# r\n");
g("add", "-A");
g("commit", "-q", "-m", "work");
write(repo, "docs/new.md", "a\nb\n");
write(repo, "src/b.js", "x\n");
g("add", "src/b.js");
const r = runScript("plan2code-review", "review-scope.mjs", [], { cwd: repo });
assert.equal(r.code, 0, r.stderr);
const j = r.json;
assert.equal(j.base, "main");
assert.equal(j.branch, "feature/X-1-thing");
assert.deepEqual(j.files.map((f) => [f.path, f.status.join("+")]), [["docs/new.md", "untracked"], ["README.md", "committed"], ["src/a.js", "committed"], ["src/b.js", "staged"]]);
assert.deepEqual(j.counts, { files: 4, lines: 5, added: 5, deleted: 0, docs: 2, code: 2 });
assert.equal(j.mix, "mixed");
assert.equal(j.commits.length, 1);
assert.equal(j.empty, false);
assert.equal(runScript("plan2code-review", "review-scope.mjs", ["--base", "nope"], { cwd: repo }).code, 3);
assert.equal(runScript("plan2code-review", "review-scope.mjs", [], { cwd: tmp("plain") }).code, 3);
});
/* ================================================= verification regressions */
test("specs: duplicate phase files and checklist rows are reported, and mark will not guess", () => {
const root = specFixture();
write(root, "specs/lunch-vote/phase-1-old.md", "# Old\n");
write(root, "specs/lunch-vote/overview.md", OVERVIEW.replace("- [/] [Phase 2: API](./phase-2.md) _(2 tasks)_", "- [/] [Phase 2: API](./phase-2.md) _(2 tasks)_\n- [ ] Phase 2 again"));
const j = runScript("plan2code-3-implement", "specs.mjs", ["status", "lunch-vote", "--root", root]).json;
assert.equal(j.phases[0].file, "phase-1.md", "the exact phase-N.md wins over phase-1-old.md");
assert.ok(j.checks.some((c) => c.code === "duplicate-phase-file" && /phase-1-old\.md/.test(c.message)));
assert.ok(j.checks.some((c) => c.code === "duplicate-checklist-row" && /Phase 2/.test(c.message)));
const m = runScript("plan2code-3-implement", "specs.mjs", ["mark", "lunch-vote", "--phase", "2", "--to", "done", "--root", root]);
assert.equal(m.code, 5);
assert.equal(m.json.error, "duplicate-row");
});
test("specs: group ranges expand, chains are not parallel, [?] phases count as done, a blocked phase is skipped aloud", () => {
const root = specFixture();
const groups = (cells) => OVERVIEW.replace("| A | 3, 4 | Separate files |", cells);
write(root, "specs/lunch-vote/overview.md", groups("| A | 2-4 | Separate files |"));
assert.deepEqual(runScript("plan2code-3-implement", "specs.mjs", ["status", "lunch-vote", "--root", root]).json.parallelGroups[0].phases, [2, 3, 4]);
write(root, "specs/lunch-vote/overview.md", groups("| A | 3 → 4 | 4 needs 3 |"));
assert.deepEqual(runScript("plan2code-3-implement", "specs.mjs", ["status", "lunch-vote", "--root", root]).json.parallelGroups, []);
write(root, "specs/lunch-vote/overview.md", OVERVIEW.replace("- [/] [Phase 2", "- [?] [Phase 2").replace("- [ ] **Phase 3", "- [?] **Phase 3").replace("- [ ] [Phase 4", "- [?] [Phase 4"));
const done = runScript("plan2code-3-implement", "specs.mjs", ["status", "lunch-vote", "--root", root]).json;
assert.equal(done.next.reason, "every phase is complete");
assert.deepEqual(done.pending, []);
write(root, "specs/lunch-vote/overview.md", OVERVIEW.replace("- [/] [Phase 2", "- [!] [Phase 2"));
const skip = runScript("plan2code-3-implement", "specs.mjs", ["status", "lunch-vote", "--root", root]).json.next;
assert.equal(skip.action, "choose");
assert.deepEqual(skip.skipped.map((p) => p.number), [2]);
assert.match(skip.warning, /skipping Phase 2 \[blocked\]/);
});
test("specs: blocked tasks may stand at approval; a re-opened phase is normal; done warns on unfinished work", () => {
const root = specFixture();
write(root, "specs/lunch-vote/phase-1.md", phase(1, "Complete", [["x", "1.1"], ["!", "1.2"]]));
let j = runScript("plan2code-1b-revise-plan", "specs.mjs", ["status", "lunch-vote", "--root", root]).json;
assert.ok(!j.checks.some((c) => c.code === "checked-with-open-tasks"), JSON.stringify(j.checks));
assert.ok(j.checks.some((c) => c.code === "approved-with-blocked"));
write(root, "specs/lunch-vote/phase-1.md", phase(1, "Complete", [["x", "1.1"], ["x", "1.2"], [" ", "1.3"]]));
const reopen = runScript("plan2code-1b-revise-plan", "specs.mjs", ["mark", "lunch-vote", "--phase", "1", "--to", "open", "--root", root]);
assert.deepEqual(reopen.json.changes.map((c) => c.to), ["[ ]", "In Progress"]);
j = runScript("plan2code-1b-revise-plan", "specs.mjs", ["status", "lunch-vote", "--root", root]).json;
assert.ok(j.checks.some((c) => c.code === "partly-done-unchecked" && c.level === "info"));
const done = runScript("plan2code-3-implement", "specs.mjs", ["mark", "lunch-vote", "--phase", "2", "--to", "done", "--root", root]);
assert.deepEqual(done.json.unfinished, ["2.2"]);
assert.match(done.json.warning, /unfinished/);
});
test("specs: --set cannot override computed fields; empty overview exists; uncounted task lines are flagged", () => {
const root = specFixture();
assert.equal(runScript("plan2code-2-document", "specs.mjs", ["metrics", "lunch-vote", "--step", "document", "--set", "verification_items_added=1", "--set", "total_tasks=99", "--root", root]).code, 2);
write(root, "specs/lunch-vote/phase-2.md", read(root, "specs/lunch-vote/phase-2.md") + "- [ ] **Task 2.4A:** odd\n1. [ ] **Task 2.5:** numbered\n");
const j = runScript("plan2code-3-implement", "specs.mjs", ["status", "lunch-vote", "--root", root]).json;
assert.equal(j.checks.filter((c) => c.code === "uncounted-task").length, 2);
write(root, "specs/lunch-vote/overview.md", "");
const e = runScript("plan2code-3-implement", "specs.mjs", ["status", "lunch-vote", "--root", root]).json;
assert.equal(e.overview, "overview.md");
assert.ok(e.checks.some((c) => c.code === "no-checklist"));
assert.ok(!e.checks.some((c) => c.code === "no-overview"));
});
test("commit-msg: a repo with no commits, stray arguments, and paths quoted for every shell", () => {
const repo = tmp("commit2");
gitInit(repo, "feature/ABC-12-x");
const C = (...a) => runScript("plan2code-review", "commit-msg.mjs", a, { cwd: repo });
assert.equal(C("--subject", "Add x").json.ticket, "ABC-12", "no commits yet");
const stray = C("--subject", "Fix", "--files", "a.js", "b.js");
assert.equal(stray.code, 2);
assert.equal(stray.json.error, "stray-argument");
assert.equal(C("--subject", "Fix", "--files", "src\\a.js", "--files", "my $file.md").json.command.split(" && ")[0], "git add -- src/a.js 'my $file.md'");
assert.equal(C("--subject", "Fix", "--files", "it's.md").code, 4);
});
test("review-scope: non-ASCII paths, empty untracked files, .txt is not a doc", () => {
const repo = tmp("scope2");
const g = gitInit(repo, "main");
write(repo, "a.js", "1\n");
g("add", "-A");
g("commit", "-q", "-m", "init");
write(repo, "café.md", "a\nb\n");
write(repo, "empty.js", "");
write(repo, "requirements.txt", "x\n");
const j = runScript("plan2code-review", "review-scope.mjs", [], { cwd: repo }).json;
const byPath = Object.fromEntries(j.files.map((f) => [f.path, f]));
assert.equal(byPath["café.md"].added, 2);
assert.equal(byPath["empty.js"].added, 0);
assert.equal(j.counts.docs, 1, "only café.md is a doc");
});
test("agents-md and agent-files: indirect links are not orphans, accents fold, Cursor .mdc rules are replaced", () => {
const root = tmp("agents2");
write(root, "AGENTS.md", "# AGENTS.md\n\n## Architecture\n\nDetails: [A](./.agents-docs/AGENTS-architecture.md)\n");
write(root, ".agents-docs/AGENTS-architecture.md", "# Architecture\n> Part of [AGENTS.md](../AGENTS.md) — project guidance for AI coding agents.\n\nSee [deep](./AGENTS-deep.md).\n");
write(root, ".agents-docs/AGENTS-deep.md", "# Deep\n> Part of [AGENTS.md](../AGENTS.md) — project guidance for AI coding agents.\n");
const j = runScript("plan2code-init-update", "agents-md.mjs", ["inspect", "--root", root]).json;
assert.deepEqual(j.orphans, []);
assert.deepEqual(j.indirect, [".agents-docs/AGENTS-deep.md"]);
assert.equal(runScript("plan2code-init", "agents-md.mjs", ["filename", "Café Ünïcode"]).json.file, ".agents-docs/AGENTS-cafe-unicode.md");
write(root, ".cursor/rules/style.mdc", "---\nalwaysApply: true\n---\nrule\n");
const d = runScript("plan2code-init", "agent-files.mjs", ["detect", "--root", root]).json;
assert.deepEqual(d.found.find((f) => f.id === ".cursor/rules").files.map((f) => f.path), [".cursor/rules/style.mdc"]);
runScript("plan2code-init", "agent-files.mjs", ["apply", ".cursor/rules", "--root", root]);
assert.deepEqual(fs.readdirSync(path.join(root, ".cursor/rules")), ["reference.md"]);
});
/* ================================================== code-review regressions */
test("agent-files: an entry that is AGENTS.md through a link is never written through", () => {
const root = tmp("agentlink");
const agents = "# AGENTS.md\n\nThe real instructions.\n";
write(root, "AGENTS.md", agents);
fs.linkSync(path.join(root, "AGENTS.md"), path.join(root, "CLAUDE.md"));
write(root, "GEMINI.md", Array.from({ length: 12 }, (_, i) => `line ${i}`).join("\n") + "\n");
const A = (...a) => runScript("plan2code-init", "agent-files.mjs", [...a, "--root", root]);
const d = A("detect").json;
assert.equal(d.found.find((f) => f.id === "CLAUDE.md").linkedToAgents, true, "a hard link to AGENTS.md is AGENTS.md");
assert.equal(d.found.find((f) => f.id === "GEMINI.md").linkedToAgents, false);
assert.ok(!d.warnings.some((w) => /CLAUDE\.md/.test(w)), "a linked entry is not warned as custom content");
const refused = A("apply", "CLAUDE.md", "GEMINI.md");
assert.equal(refused.code, 5);
assert.equal(refused.json.error, "linked-to-agents");
assert.match(refused.json.next, /already serves AGENTS\.md/);
assert.equal(read(root, "AGENTS.md"), agents, "refused before anything was written");
assert.match(read(root, "GEMINI.md"), /^line 0/, "a refusal writes nothing at all");
const all = A("apply", "--all");
assert.equal(all.code, 0, all.stderr);
assert.ok(all.json.actions.some((a) => a.action === "skipped-linked" && a.path === "CLAUDE.md"));
assert.deepEqual(all.json.updated, ["GEMINI.md"]);
assert.equal(read(root, "AGENTS.md"), agents, "--all skips the linked entry");
assert.match(read(root, "GEMINI.md"), /^# GEMINI\.md\n\nSee \[AGENTS\.md\]/);
// Symlinks need admin (or Developer Mode) on Windows: run this part only where they work.
const sroot = tmp("agentsym");
write(sroot, "AGENTS.md", agents);
write(sroot, "shared/other.md", "someone else's file\n");
try {
fs.symlinkSync(path.join(sroot, "AGENTS.md"), path.join(sroot, "CLAUDE.md"));
fs.symlinkSync(path.join(sroot, "shared/other.md"), path.join(sroot, "GEMINI.md"));
} catch (err) {
if (err.code === "EPERM" || err.code === "EACCES") return;
throw err;
}
const S = (...a) => runScript("plan2code-init", "agent-files.mjs", [...a, "--root", sroot]);
assert.equal(S("detect").json.found.find((f) => f.id === "CLAUDE.md").linkedToAgents, true);
assert.equal(S("apply", "CLAUDE.md").code, 5);
assert.equal(S("apply", "GEMINI.md").code, 0);
assert.equal(read(sroot, "shared/other.md"), "someone else's file\n", "a symlink to another file is replaced, not written through");
assert.equal(fs.lstatSync(path.join(sroot, "GEMINI.md")).isSymbolicLink(), false);
assert.equal(read(sroot, "AGENTS.md"), agents);
});
test("agent-files: a CRLF checkout of a pointer is still the pointer", () => {
const root = tmp("agentcrlf");
write(root, "AGENTS.md", "# AGENTS.md\n");
write(root, "GEMINI.md", "x\n");
write(root, ".windsurf/rules/a.md", "a\n");
const A = (...a) => runScript("plan2code-init", "agent-files.mjs", [...a, "--root", root]);
assert.equal(A("apply", "--all").code, 0);
for (const f of ["GEMINI.md", ".windsurf/rules/reference.md"]) write(root, f, read(root, f).replace(/\n/g, "\r\n"));
const d = A("detect").json;
assert.equal(d.found.find((f) => f.id === "GEMINI.md").alreadyPointer, true);
assert.equal(d.found.find((f) => f.id === ".windsurf/rules").alreadyPointer, true);
assert.deepEqual(A("apply", "GEMINI.md", ".windsurf/rules").json.actions.map((a) => a.action), ["unchanged", "unchanged"]);
assert.match(read(root, "GEMINI.md"), /\r\n/, "an unchanged CRLF pointer is left as it is");
});
test("specs mark: a plain or bold Status: line is updated; a missing one is reported", () => {
const root = specFixture();
const S = (...a) => runScript("plan2code-3-implement", "specs.mjs", [...a, "--root", root]);
write(root, "specs/lunch-vote/phase-3.md", phase(3, "x", [["x", "3.1"]]).replace("**Status:** x", "Status: In Progress"));
const plain = S("mark", "lunch-vote", "--phase", "3", "--to", "done");
assert.equal(plain.code, 0, plain.stderr);
assert.match(read(root, "specs/lunch-vote/phase-3.md"), /^Status: Complete$/m, "the plain form is kept as written");
assert.ok(!plain.json.warning);
write(root, "specs/lunch-vote/phase-4.md", phase(4, "x", [["x", "4.1"]]).replace("**Status:** x", "- **Status**: Not Started"));
assert.equal(S("status", "lunch-vote").json.phases[3].status, "Not Started");
assert.equal(S("mark", "lunch-vote", "--phase", "4", "--to", "done").code, 0);
assert.match(read(root, "specs/lunch-vote/phase-4.md"), /^- \*\*Status\*\*: Complete$/m);
write(root, "specs/lunch-vote/phase-2.md", phase(2, "x", [["x", "2.1"]]).replace("**Status:** x\n", ""));
const none = S("mark", "lunch-vote", "--phase", "2", "--to", "done");
assert.equal(none.code, 0);
assert.match(none.json.warning, /phase-2\.md has no Status line, so only the Phase Checklist checkbox changed/);
assert.match(read(root, "specs/lunch-vote/overview.md"), /- \[x\] \[Phase 2: API\]/);
});
test("specs: only a folder directly under specs/ is a spec; archive verification fails loudly", () => {
const root = specFixture();
write(root, "src/app.js", "x\n");
const S = (...a) => runScript("plan2code-4-finalize", "specs.mjs", [...a, "--root", root]);
for (const arg of [".", "./src", "./specs", "specs/lunch-vote/.."]) {
const r = S("archive", arg, "--dry-run");
assert.equal(r.code, 5, `archive ${arg}: ${r.stdout}`);
assert.equal(r.json.error, "not-a-spec");
assert.match(r.json.next, /specs\/<name>/);
}
assert.ok(fs.existsSync(path.join(root, "specs/lunch-vote/overview.md")));
assert.equal(S("archive", "./specs/lunch-vote", "--dry-run").code, 0);
assert.equal(S("archive", path.join(root, "specs", "lunch-vote"), "--dry-run").code, 0, "an absolute path under specs/ is fine");
assert.equal(S("status", "specs--completed/old-one").code, 0, "a read may name an archived spec");
assert.equal(S("status", "./src").json.error, "not-a-spec");
});
test("specs: an em dash in a group range expands like a hyphen", () => {
const root = specFixture();
write(root, "specs/lunch-vote/overview.md", OVERVIEW.replace("| A | 3, 4 | Separate files |", "| A | 2\u20144 | Separate files |"));
assert.deepEqual(runScript("plan2code-3-implement", "specs.mjs", ["status", "lunch-vote", "--root", root]).json.parallelGroups[0].phases, [2, 3, 4]);
});
test("commit-msg: @ and - paths are quoted, git add ends its options, risky subjects are refused, add and commit apart", () => {
const repo = tmp("commit3");
gitInit(repo, "feature/ABC-12-x");
const C = (...a) => runScript("plan2code-review", "commit-msg.mjs", a, { cwd: repo });
const files = C("--subject", "Fix", "--files", "@types/x.d.ts", "--files", "-weird.md", "--files", "src/a@b.js");
assert.equal(files.code, 0, files.stderr);
assert.equal(files.json.add, "git add -- '@types/x.d.ts' '-weird.md' src/a@b.js");
assert.equal(files.json.commit, 'git commit -m "Fix" -m "ABC-12" -m "AI Assisted"');
assert.equal(files.json.command, `${files.json.add} && ${files.json.commit}`);
assert.equal(C("--subject", "Fix", "--add-all").json.add, "git add -A");
const bare = C("--subject", "Fix").json;
assert.equal(bare.add, null);
assert.equal(bare.command, bare.commit);
for (const bad of ["back\\slash", "wow!", "a\rb", "say \u201Chi\u201D"]) {
assert.equal(C("--subject", bad).code, 4, JSON.stringify(bad));
}
assert.equal(C("--subject", "wow!").json.error, "unsafe-subject");
assert.equal(C("--subject", "Fix", "--files", "it\u2019s.md").json.error, "unsafe-path");
});
test("review-scope: no main or master is no-base unless the repo has no commits; a move up a level keeps one slash", () => {
const repo = tmp("scope3");
const g = gitInit(repo, "develop");
write(repo, "d/sub/f.txt", "1\n");
write(repo, "top/h.txt", "3\n");
g("add", "-A");
g("commit", "-q", "-m", "init");
g("checkout", "-q", "-b", "feature/X-2-thing");
write(repo, "src/new.js", "x\n");
g("add", "-A");
g("commit", "-q", "-m", "feature work");
const R = (...a) => runScript("plan2code-review", "review-scope.mjs", a, { cwd: repo });
const none = R();
assert.equal(none.code, 3);
assert.equal(none.json.error, "no-base");
assert.match(none.json.next, /--base <trunk>/);
const dev = R("--base", "develop");
assert.equal(dev.code, 0, dev.stderr);
assert.deepEqual(dev.json.files.map((f) => f.path), ["src/new.js"]);
g("mv", "d/sub/f.txt", "d/f.txt");
g("mv", "top/h.txt", "h.txt");
const moved = R("--base", "develop").json.files.map((f) => f.path);
assert.ok(moved.includes("d/f.txt") && moved.includes("h.txt"), JSON.stringify(moved));
assert.ok(!moved.some((p) => p.includes("//") || p.startsWith("/")), JSON.stringify(moved));
const fresh = tmp("scope4");
gitInit(fresh, "develop");
write(fresh, "a.js", "1\n");
const unborn = runScript("plan2code-review", "review-scope.mjs", [], { cwd: fresh });
assert.equal(unborn.code, 0, unborn.stderr);
assert.equal(unborn.json.base, null);
assert.match(unborn.json.note, /No commits yet/);
assert.deepEqual(unborn.json.files.map((f) => f.path), ["a.js"]);
});
File diff suppressed because it is too large. Load diff
+2 -2
View File
@@ -3,14 +3,14 @@
const fs = require('fs');
const path = require('path');
const CHAR_LIMIT = 11000;
const CHAR_LIMIT = 20000;
const srcDir = path.join(__dirname, '..', 'src');
// Note: readdirSync is non-recursive, so files in subdirectories like
// plan2code-review-references/ are automatically excluded from
// the character limit check. Reference files have no char limit.
const files = fs.readdirSync(srcDir)
.filter(f => f.startsWith('plan2code-') && f.endsWith('.md'))
.filter(f => /^plan2code(-.*)?\.md$/.test(f))
.sort();
const failures = [];
+174
View File
@@ -0,0 +1,174 @@
---
name: plan2code-0-pathfinder
description: "Plan2Code Step 0: Pathfinder Mode - user-initiated workflow step. Do not invoke autonomously."
disable-model-invocation: true
---
# 🧭 PATHFINDER MODE
Start all PATHFINDER MODE responses with '🧭 [PATHFINDER: Chart - Step X: Name]' or '🧭 [PATHFINDER: Work - Step X: Name]'.
## Role
Pathfinder, not architect. An idea has arrived too big or unclear to plan. Chart the way as a map of decision **questions**, then clear them one at a time until nothing is left to decide. Hand off to `/plan2code-1-plan`.
Read references/grilling.md
> Fallback: ≤3 independent probes/turn, each with a recommendation, re-ask any skipped; AskUserQuestion for every question, prose only if unavailable; plain English; facts you look up, decisions are the human's.
## Interface
The FIRST thing you do, before anything else in this file — before Step 0: ask **web console** (suggested) or **terminal**? Console → run `node "<D>/console.mjs" open --workflow pathfinder` before reading (<D>: references/web-console/ beside this SKILL.md, or the dir in ~/.plan2code/console/console-dir), then <D>/console.md → Pathfinder notes. Switchable anytime. `--web` or `Use the web console for this session.` in the argument answers it; drop the flag. Dashboard-launched? Resume its open session (console.md → Launches), then Step 0.
## Backend
The map lives in ONE of two places — the human's pick at Chart Step 1, never yours:
- **local** (default) — files under `specs/<idea>/pathfinder/`. Private, gitignored, solo.
- **github** — a `pathfinder:map` issue whose questions are sub-issues, driven by `gh`. Shared, visible in the tracker, parallel.
Read references/github-issues.md — REQUIRED on `github`, skip on `local`.
> Fallback: map = a `pathfinder:map` issue titled `Map: <idea>`; questions = its sub-issues, labelled `pathfinder:<type>-<mode>`; blocking = native dependencies; claim = assign `@me`; resolve = `## Answer` comment, then close.
Recorded as the first `## Ground rules` bullet (`**Backend:** local|github`), never re-asked or switched. Either way the PLAN-DRAFT lands in local `specs/<idea>/` — downstream steps read files, not issues.
## Scripts
On `local`, the map's bookkeeping is a script; the decisions are yours. `node "<S>/pathfinder.mjs" <command> specs/<idea>/pathfinder`, run from the project root, where `<S>` is `scripts/` beside this SKILL.md (`~/.agents/skills/plan2code-0-pathfinder/scripts/` globally). It prints one JSON object. Exit 0 is success; anything else carries `message` and `next`, so do what `next` says. Exit 5 (refused) means a rule would break: never work around it by editing the files by hand. `--help` lists every command; they are named at the steps below. It reads local files only: on `github`, the backend playbook does each step with `gh` (except `lint`, which checks the local briefs and PLAN-DRAFT once one exists), and a `no-map` result there is expected, never a cue to chart.
## Project Context
Load `./AGENTS.md` if it exists — its conventions govern; never re-ask what it answers. If missing, fold it into the Step 0 gate batch: *"No `AGENTS.md`. Pathfinder can chart without it. Continue, or run `plan2code-init` first?"* Record the outcome as the first line of `## Ground rules`: ``- `AGENTS.md` exists and governs.`` or, when missing, exactly ``- `AGENTS.md` is absent.`` (the handoff reads it).
## Rules
- **Plan, don't do.** Every question resolves a DECISION. The pull to just build it is the edge of the map — hand off.
- **Confirm before creating anything.** No files, no issues, until the Intent Gate (Step 0) and backend pick (Step 1) return.
- **One question at a time** (`research` excepted — parallel); fork-menu between decisions, fresh session after ~3.
- **Refer by name.** "[Export format](<link>)", never "02" or "#42" in prose. Bare ids belong on `Blocked by:` lines and in commands.
- **HITL questions are never self-answered.** Ask and wait; an agent that answers its own grill has broken the skill.
- **Questions are ground truth; the map is a rebuildable index.** A filled `## Answer` beats any state marker; detail lives in one place.
- **Never write implementation code** into the project. Sketches are throwaway, living only under `pathfinder/sketch-NN/`.
- **Reserved names — never create inside `pathfinder/`:** `overview.md`, `phase-<N>.md`, `PLAN-DRAFT-*.md`, `PLAN-CONVERSATION-*.md`.
- **Never emit the loop's completion tokens under `specs/`** — `TASK_COMPLETE`, `PHASE_COMPLETE`, `ALL_TASKS_COMPLETE`, `IMPLEMENTATION_COMPLETE`, `SPEC_COMPLETE`, `WORK_COMPLETE`. It scans for them.
- **No `- [ ]` checkboxes inside question files**, and no `METRICS_JSON` anywhere.
- **Lint before every Session End** (on `github`, once a brief or PLAN-DRAFT is on local disk; it checks those): `lint` checks the rules above, the schema lines, `Blocked by:` targets and cycles, and metrics-scraper bait. Fix every error it lists (exit 4) before you stop.
## Auto-Discovery and Mode Selection
⚠️ `specs/` is gitignored — NEVER use Glob. Shell only: `ls specs/` (Bash) or `Get-ChildItem specs/` (PS).
**Brief/recap/minutes asked?** → read references/brief.md: `reconcile`, then `brief-data --range <range>` (`github`: its brief section), write plain-English `briefs/brief-<YYYYMMDD>.md` from the output, STOP. Never resolves.
**Identify the idea FIRST** (from the argument, or ask), then evaluate for THAT idea — first match wins. An issue URL or number means `github`; else look for a local map, then `gh issue list --label pathfinder:map` for `Map: <idea>`.
| Condition | Route |
|---|---|
| No map, but `specs/<idea>/overview.md` exists | Documented — offer `/plan2code-3-implement`. STOP |
| No map in either backend | MODE A, Step 0 (Intent Gate) |
| Map `**Status:** Charting` | MODE A, resume at Step 6 |
| Map `**Status:** Working` | MODE B |
| Map `**Status:** Cleared` | Point at the PLAN-DRAFT and `/plan2code-1-plan`. STOP |
| Map exists, `**Status:**` unreadable | MODE B — Step 2 rebuilds and sets it |
One idea per session, each with its own map.
## Questions
Read references/questions.md — the `local` format. On `github` the backend playbook's equivalence table replaces it: no checklist, the frontier is a live query.
> Fallback (`local`): `map.md` indexes; `questions/NN-<slug>.md` hold the decisions, `00` is codebase context, six `Key: value` lines each, `Resolved:` dating it. Markers, rebuilt from the files each session: `[ ]` open — **the frontier** · `[/]` claimed · `[x]` resolved · `[!]` blocked · `[-]` out of scope.
## MODE A: Chart
Read references/chart.md
> Fallback: confirm the outcome with the human FIRST; only then grill the destination, then breadth-first; write the map and one question per sharp decision.
0. `[Chart - Step 0: Intent Gate]` **Before creating anything**, use AskUserQuestion for the outcome and WAIT: **chart a map** (foggy — Step 1), **`/plan2code-1-plan`** (clear — STOP), **`/plan2code-quick-task`** (tiny — STOP). HITL, never self-select "chart".
1. `[Chart - Step 1: Name and backend]` Only after the gate says "chart." Confirm the kebab-case idea name, then ask — HITL, never self-picked — **local files or GitHub Issues?** Recommend `local` for solo work; offer `github` only if preflight passes, naming the repo's visibility. THEN the first write.
2. `[Chart - Step 2: Destination]` Grill until it is one or two lines. It fixes scope — settle it first.
3. `[Chart - Step 3: Recon]` Explore the codebase; record codebase context, resolved on the spot, `legwork · AFK`. On `github` hold it until Step 6 so a Step 4 off-ramp leaves no litter.
4. `[Chart - Step 4: Map the frontier]` Grill again **breadth-first**: fan out, never deep on one thread. Surface the open decisions and what is takeable now.
5. `[Chart - Step 5: Create the map]` `**Status:** Charting`, Destination, Ground rules (backend first), an empty index, the fog in `## Not yet specified`. Say once where it lives and who sees it.
6. `[Chart - Step 6: Write the questions]` One per decision you can phrase sharply NOW, dependency order, `Blocked by:` filled the same pass — on `github`, create all first, wire edges second. The rest stays fog. Include a `grill · HITL` testing-posture question; `/plan2code-1-plan` Phase 1 needs it.
7. `[Chart - Step 7: Index]` `local`: `reconcile` fills `## Question Checklist` from the files (every missing row, with its marker). Either backend: set `**Status:** Working`. `local`: then `lint`.
8. `[Chart - Step 8: Fire research]` One subagent per `research` question, in parallel. Each writes only its `## Evidence`, never decides. **Wait for all** (they die with the terminal), then Session End.
**No fog at Step 4?** Small enough to plan directly: create no map, keep the recon as a local file, attach it to `/plan2code-1-plan`, STOP. Charting resolves nothing by hand — stop at Step 8.
## MODE B: Work
Read references/resolve.md
> Fallback: resolve by type — research reads sources, sketch makes something concrete, grill interviews the human, legwork does the manual work.
Assume NO memory of any prior session.
1. `[Work - Step 1: Load]` Read the map whole. No question yet.
2. `[Work - Step 2: Reconcile]` **Always.** `local`: run `reconcile`. The answer wins: a question whose `## Answer` its state contradicts is resolved, a missing `Resolved:` is backfilled from the claim (today if none), a claim with no `## Answer` (a crash) is released, and every marker is rebuilt from the questions. Tell the human each `repairs` entry in one line. Re-fire every question in `refire` (research cut off before its `Research complete:` line; resolve.md → Interrupted research). `--dry-run` first if the map looks badly off. `github`: the reconcile checks in the backend playbook.
3. `[Work - Step 3: Frontier]` Every question open, unclaimed, and unblocked, first in order. `local`: run `frontier` (`frontier`, `blocked` with each blocker, `stranded`, `nextNN`, `counts`).
4. `[Work - Step 4: Choose and claim]` The one the user named, else first on the frontier. Mark it claimed on the question and the map, **saved before any work** (`local`: `claim <NN>`, which refuses a blocked or non-open question and a second open claim). A `research` pick fires unclaimed instead (resolve.md); then claim the next one.
- Stranded on an `out-of-scope` blocker? Check this first: re-frame or rule out, re-run Step 3.
- Frontier empty but questions remain (`allBlocked`)? Report the chain, STOP.
- Nothing open (`nothingOpen`)? The Clearing Gate.
5. `[Work - Step 5: Zoom]` Read the claimed question in full, plus any closed question it references. Obey `## Ground rules`.
6. `[Work - Step 6: Resolve]` Route by type per the resolve playbook. HITL needs the human's own words.
7. `[Work - Step 7: Record]` Write `## Answer`: the decision, what was rejected and why, consequences, a one-line `**Gist:**` last. Sources under `## Evidence`. Mark it resolved, dated today, index the gist + date on the map, bump `**Updated:**` (`local`: `resolve <NN>` does all four; its `unblocked` list feeds the menu).
8. `[Work - Step 8: Graduate]` Fog now sharp? Write those questions (`local`: numbered from `nextNN`, then `reconcile` to index them), delete the graduated bullets. Past the destination? Rule it out of scope, one line in `## Out of scope` (`local`: `rule-out <NN> --reason "<why>"`; handle anything it reports `stranded`). Invalidated? Re-frame or rule out.
9. `[Work - Step 9: Menu]` Run The Clearing Gate. Not cleared: fork-menu (Form C; resolve.md §Cadence) — pick → Step 4, stop → Session End.
## The Clearing Gate
Read references/handoff.md
> Fallback: write `specs/<idea>/PLAN-DRAFT-<YYYYMMDD>.md` from the map, Status `Phase 3 Complete - Resume at Phase 4`, then route to `/plan2code-1-plan`.
The map clears only when ALL hold:
1. Nothing open, claimed, or blocked
2. `## Not yet specified` is EMPTY
3. The destination is reachable with nothing left to decide
4. All four confidence dimensions score ≥ 18/25
Score the four dimensions (handoff.md rubric) and write them to the map's `**Confidence:**` line. `local`: run `gate`; it checks 1, 2 and 4 against the files, plus the handoff preflight and the Requirements, Feasibility and Risk caps; condition 3, the Integration cap and the scores stay yours (its `judgment` list). Any failing: name it, keep working. All passing: follow the handoff playbook, set `**Status:** Cleared`, run `lint`, stop. The PLAN-DRAFT is always local — `/plan2code-1-plan` cannot read a tracker.
## Trail Footer
Read references/trail.md
> Fallback: once the map exists, end every response with a marker path (`●` done · `◉` here · `○` open · `⊘` blocked · `⊝` out of scope) `START`→`⚑`, a named legend, a plain-English confidence note, then ONE closer: asking → Form B, `WAITING ON YOU`, no command; menu → Form C, `NEXT UP` + `OR START FRESH`; session end → Form A, `NEXT STEP` + resume command with `--web` (map issue URL on `github`).
Close EVERY response with the trail once the map exists. You pick the form by what the turn does. `local`: `trail` draws it from the files, so paste its output verbatim. Asking → Form B, `trail --form B --item "<probe>" ... --text` (one `--item` per open probe, up to three). Menu after a decision → Form C, `trail --form C` WITHOUT `--text`: paste its `text`, then build the question tool from its `menu`. Session end → Form A, `trail --form A --text`. `github`: draw it by hand per the backend playbook.
## Session End
Report each question resolved (name + gist), what graduated, what's still open; offer a brief for today. Nothing to commit — `local` is gitignored, `github` already on the tracker. Then the mascot, then the Trail Footer.
**Web console:** at session end or on a `__stop`, post `finish` (Form A's command) BEFORE `stop` (console.md → Pathfinder).
```
⋅
╭───╮
╲│ ★ │╱
│ ◡ │ One more decision down. The fog is thinner!
╰┬─┬╯
```
**When the map cleared**, the mascot says `The way is clear! Time to plan!` and the footer routes to `/plan2code-1-plan` — or `/plan2code-init` FIRST if `## Ground rules` records `AGENTS.md` absent.
## Abort / Recovery
| Issue | Action |
|---|---|
| Session stops mid-question, or map drift | Release the claim, note why (or leave it: the next reconcile releases a claim with no answer). Work Step 2 repairs both; questions win. Never reconcile mid-question: it would release your own live claim. |
| A script cannot run (no `node`) | Say so once; do the step by hand from questions.md (markers, blocking, schema) and the reference that names the step. Footer by hand: the trail's glyph path, then Form A `NEXT STEP · start a new conversation and run:` + the command, Form B `WAITING ON YOU · answer here, in this conversation:` + the items, or Form C `OR START FRESH · new conversation, paste:` + the command. |
| Frontier empty, fog remains | Not sharp yet. Grill it into a question, or clear the map |
| Reference file missing | Use the fallback blockquote under its `Read` line |
| `gh` fails mid-session on a `github` map | Report and STOP. Falling back to local forks the map |
| User skips to planning | Their call. Name what is undecided, route to `/plan2code-1-plan` |
## Learning Capture
If charting surfaced project-specific insights, suggest `/plan2code-init-update` to capture them in `AGENTS.md`.
@@ -0,0 +1,102 @@
# Brief Playbook
> Loaded when the user's argument asks for a brief. Produces one plain-English report file for a date range — the artifact a PM pastes into Slack or reads aloud in a meeting. The only mutation BRIEF mode ever performs is the Work Step 2 reconcile; it never claims, resolves, or answers a question.
> **Backend note.** The report is IDENTICAL either way — same sections, same plain-English rules — and the brief file always lands on **local disk** under `specs/<idea>/pathfinder/briefs/`, even on `**Backend:** github`. What differs is where the inputs come from: on `github` the range filter reads each question issue's close date instead of a `Resolved:` line, and the open / blocked / out-of-scope sections come from the frontier query rather than the checklist — see `github-issues.md`. `pathfinder.mjs` reads local files only, so on `github` do each script step below except `lint` by hand with `gh`, as `github-issues.md` describes.
## When it fires
The argument (or message) asks for a `brief`, `summary`, `recap`, or `minutes` for an idea, with an optional range:
```
/plan2code-0-pathfinder write a brief for today for specs/<idea>/pathfinder
/plan2code-0-pathfinder brief since 2026-08-15 for specs/<idea>/pathfinder
/plan2code-0-pathfinder full brief for specs/<idea>/pathfinder
```
Translate the person's words into the script's range: `today` (the default when unstated) · `this-week` (Monday through today) · `since:<YYYY-MM-DD>` · `<YYYY-MM-DD>..<YYYY-MM-DD>` · `full` (everything since the map was created). Resolve a relative phrase ("since Tuesday") to a date yourself; the script takes dates only and reads today from the clock.
No `map.md` for the idea? There is nothing to brief — say so and route to charting. `**Status:** Cleared`? A brief is still legitimate: cover the full history and point at the PLAN-DRAFT in `## Next step`.
## Procedure
1. **Resolve the idea** exactly as Auto-Discovery does.
2. **Reconcile first**: `node "<S>/pathfinder.mjs" reconcile specs/<idea>/pathfinder`. A brief must never disagree with the map it summarizes, which is why BRIEF mode repairs drift rather than reporting it. Keep its `repairs` for step 5.
3. **Gather**: `node "<S>/pathfinder.mjs" brief-data specs/<idea>/pathfinder --range <range>`. It filters resolved questions by `Resolved:` date and returns every section's source (below), the file to write, and `drift` / `undatedResolved` (both empty after step 2; if not, rerun step 2). Out-of-scope rulings are never date-filtered: standing scope boundaries appear in every brief.
4. **Write** the file at its `file` path (create `briefs/` if absent). The name carries today's date regardless of range; the `**Covers:**` line disambiguates. Same-day re-runs overwrite (`exists: true`): a brief is a report, not a record; the question files remain the record.
5. **Report** in chat: the file path, the range, one line on any reconcile repairs. Then the Trail Footer, Form A (the session is over; the command routes by map status as usual). Run `lint` first: it checks the brief for percent signs and scraper bait.
## The template
Substitute real content; keep the section order. `<in range>` reads naturally per range: "Decided today", "Decided this week", "Decided so far" (full).
```markdown
# Audit log export - decisions brief
**Covers:** 2026-09-02 · **Map:** 6 of 9 decisions made
## Where this is headed
<The destination, verbatim from the map.>
## Decided today
- **Export format** - CSV with a UTF-8 BOM plus a SHA-256 manifest file.
Rejected: JSONL (recipients open these in Excel); XLSX (harder to verify byte-for-byte).
*Recorded as hard to reverse.*
- **Export authorization** - new tenant-scoped compliance role; self-events included but flagged.
Rejected: reusing the admin role (no tenant scoping today).
## Ruled out
- **SIEM push connector** - continuous streaming is a separate effort with its own owner.
## Open - ready to decide next
- **Delivery channel** - can now be decided; the row-count ceiling it was waiting on landed today.
- **Testing posture** - needs Product and Engineering together.
## Open - waiting on something first
- **Redaction rules** - waiting on Legal's answer about exporting raw payloads.
## Still taking shape
- How far back an export may reach - depends on a retention answer nobody has yet.
## How solid is this?
Solid overall, but risk handling is borderline - one limit has no decided behavior past it yet.
## Next step
Run: `/plan2code-0-pathfinder specs/audit-log-export/pathfinder`
---
*Full reasoning behind every decision: `specs/audit-log-export/pathfinder/questions/`*
```
**How each section is built:**
| Section | `brief-data` field | Rendering rule |
|---|---|---|
| Heading + `**Covers:**` | `title`, `covers`, `progress` | `progress` already uses the trail's counting rule (out-of-scope excluded from the total) |
| Where this is headed | `destination` | Verbatim. Never paraphrase it |
| Decided `<in range>` | `decidedHeading`, `decided` (`name`, `gist`, `rejected`, `hardToReverse`) | Question name bold, gist as the first line, each `rejected` entry compressed to ONE line with the why in parentheses. `hardToReverse` renders as *Recorded as hard to reverse.* — never the word "locked". Empty → `note` |
| Ruled out | `ruledOut` | One line each: name + why, link markup dropped |
| Open - ready to decide next | `ready` | Flag each `justUnblockedBy`: "can now be decided; X landed <day>" |
| Open - waiting on something first | `waiting` | Name the blocker in plain words ("waiting on the export format decision"), never `Blocked by: 02` |
| Still taking shape | `fog` | Lightly rephrase each bullet into one plain sentence |
| How solid is this? | `confidence` | Already the plain-English line. Omit the section when it is null |
| Next step | `nextStep` | `command` as given; with a `plan`, point at that PLAN-DRAFT too |
Nothing decided in the range? Keep the file honest: "No decisions were recorded in this period." under Decided, and render the open sections as usual — a meeting that only surfaced questions still gets minutes.
## Rules
- **Zero pathfinder machinery in the file.** No `NN` numbers, no markers, no `Type:`/`State:` tokens, no fog/frontier/grill vocabulary, no `Locked:`. Questions are referred to by name only, no links. The audience may never open a terminal.
- **Scraper safety, same as the PLAN-DRAFT:** no `%`, no raw `NN/25` scores, and no bare `Requirements` / `Feasibility` / `Integration` / `Risk` followed by a number anywhere in the file. The plain-English confidence line is the only rendering permitted.
- **Standing `specs/` rules apply:** no loop tokens, no `- [ ]` checkboxes, no `METRICS_JSON`.
- **Rejections are one line each.** The brief summarizes; the question file argues. The closing pointer line covers readers who want the full reasoning.
- **BRIEF mode does not count as question work.** It does not count toward the session's decision cadence; no claim is ever taken.
- **Do not edit an old brief to "update" it.** Yesterday's brief reflects yesterday's map — write today's instead. Same-day overwrite only.
@@ -0,0 +1,691 @@
# Chart Playbook
> Part of plan2code-0-pathfinder — loaded at the top of MODE A (Chart). Expands the numbered Chart steps.
>
> **Backend note.** Steps 0, 2, and 4 — the gates and the grills — are identical either way, and so is every judgement call below (fog vs question, in scope vs out, the destination test). What differs is where Steps 3 and 5-7 put the bytes: on `**Backend:** github`, `github-issues.md` overrides the `map.md` and question-file templates here, the single-pass rule under *Step 6: Numbering and dependency order*, and the timing of the recon. Read it alongside this file, not instead of it.
Charting produces a map and a set of question files. It resolves nothing by hand. Every judgement below serves one goal: put a sharp question on the map for everything you can phrase now, and leave everything else honestly in the fog.
---
## Step 0: The intent gate
Pathfinder builds an apparatus — a directory, a map, a file per decision. That apparatus earns its keep only when the way to the destination is genuinely foggy. For a small or already-clear ask it is pure overhead, and creating it before the human has agreed to it is the fastest way to make Pathfinder feel heavy and get in the way. So the gate runs **before the first byte hits disk**.
You already have the idea name from Auto-Discovery. Do NOT create the directory yet. Say, in substance:
> "This is Pathfinder. Nothing exists for `<idea>` yet. Pathfinder charts a map of the open decisions when an idea is big or unclear to plan — but that is overhead if this is small or already clear. Three ways to go:
> - **Chart it** — I map the open decisions, we clear them one at a time, then hand a draft to `/plan2code-1-plan`.
> - **Straight to `/plan2code-1-plan`** — the way looks clear enough to plan now.
> - **`/plan2code-quick-task`** — small enough to just do.
>
> My read: `<recommendation with a one-line reason>`. Which?"
Rules for the gate:
- **It is HITL.** You recommend; the human chooses. Never self-select "chart" and start creating files because it is the default path — that is exactly the failure this gate exists to stop.
- **Read the request honestly.** A one-line bugfix, or a change with no open decisions, is not a charting job — recommend an off-ramp and mean it. Reserve "chart" for real fog: several unsettled decisions, unclear scope, or competing designs. A joint Product + Engineering kickoff on a fresh idea is the canonical "chart it" case — a new feature explored by two teams is fog almost by definition.
- **No disk writes.** Naming the idea and talking is free. Creating `specs/<idea>/pathfinder/` is not — it waits for an explicit "chart."
- **On an off-ramp, route and STOP.** Point at `/plan2code-1-plan` or `/plan2code-quick-task`, create nothing, end the session. If `AGENTS.md` is absent, mention `/plan2code-init` first, as with any handoff.
This gate and the Step 4 no-fog off-ramp are the same instinct at two moments: the gate is the human's call before any work; the off-ramp is your call once the breadth-first grill has proven there was no fog after all. Either one ending the session without a map is a success, not a failure.
---
## Step 2: The destination grill
The destination is settled **first** because it fixes scope. Every later judgement — is this a question or fog, is this in scope or past the edge, is this map cleared — is measured against it. A vague destination makes all three unanswerable, and you will spend the rest of the map arguing about boundaries instead of decisions.
A destination is **one or two lines** describing what exists when the map clears. It is not a feature description. It is not a value proposition. It names the artifact and draws the edge.
### The script
Recommend an answer with each probe — the human corrects faster than they compose. Follow the grilling playbook for tone, cadence, and the batching rules; this is the content.
**Six probes, two batches.** They do not all pass the independence test, so they split:
| Batch | Probes | Why they go together |
|---|---|---|
| First | 1 (artifact), 2 (person), 6 (forcing function) | Each stands alone. None reads differently under the others' answers. |
| Second | 3 (sacrificial boundary), 4 (smallest arrival), 5 (arrival signal) | All three presuppose an artifact and an actor. Sending them before batch 1 lands asks the human to draw an edge around something unnamed. |
Two round trips, not six. Three probes each — exactly the cap, so neither batch needs splitting.
Batch 2 bends the independence test on purpose. The arrival signal (5) can shift under the smallest arrival (4), so strictly it should be held back — but holding it costs a third round trip to catch a conflict that is rare and cheap to spot. The trade is to send them together and reconcile at the recap: if the smallest arrival comes back materially smaller than the artifact you were told about, re-check the arrival signal against it before writing the destination. A knowing trade here, not a licence to batch dependent probes elsewhere.
**Both batches go through the structured question tool.** Give each probe genuine candidate options with trade-offs and a recommendation; the built-in `Other` path is where the human composes the actor, a further feature to leave out, the arrival signal, or forcing-function detail in their own words. A clicked option is not written verbatim as the destination — the recap-confirmation tool call turns all six answers into the agreed one-or-two-line destination before disk write.
**Probe 1 — the artifact**
> "When we are done here, what exists that does not exist now: a plan you hand to `/plan2code-1-plan`, a decision recorded before anyone plans, or a change already made in the codebase? My guess: a plan."
*Fishing for:* the shape of the destination. Push back if the answer is "the feature working" — that is past the edge of every pathfinder map. Say so plainly: "That is the build. The map ends at the plan for the build."
**Probe 2 — the person on the other end**
> "Who uses the result, and what do they do with it the day it lands?"
*Fishing for:* the actor and the moment of use. Vague actors ("users", "the business") produce vague scope. Push until you get a role someone could name in an approval — compliance officer, on-call SRE, tenant admin.
**Probe 3 — the sacrificial boundary**
> "Which of these features are we leaving out this time?"
Deliver it as a multi-select (`multi` on the web console) of 3–5 features a reasonable person might expect to be included. Every option names a feature ("Scheduled exports"), never a rule to keep ("Exports stay manual"). Each option's description ends "(Leaving this out = …)", naming what that leaves. Offer a recommended preset of the ones you would leave out; the free-text path lets them name another feature.
*Fishing for:* the first `## Out of scope` bullets — the ticked features become them. This probe does more work than any other. If nothing is ticked, push once: a destination nobody has excluded anything from has not been thought about. The preset is the push.
**Probe 4 — the smallest arrival**
> "What is the smallest version that would still count? If only that existed, would you call it done or would you feel cheated?"
*Fishing for:* the difference between the destination and the wish list. Everything above the smallest arrival is a candidate for out of scope or for a later effort.
**Probe 5 — the arrival signal**
> "How will you know it's done — what do you look at?"
*Fishing for:* a checkable condition. "It feels right" is not one. "Every open decision has an answer and I can hand the draft to planning without re-litigating format" is one.
**Probe 6 — the forcing function**
> "What made this come up now? A deadline, an incident, an audit, a customer?"
*Fishing for:* constraints that will shape half the questions and that nobody volunteers unprompted. A regulatory deadline changes the delivery question, the testing posture question, and the out-of-scope line all at once.
**The probe names above are internal labels, not headings the human reads.** Head each Q block plainly — *What you end up with*, *Who uses it*, *Leave for later*, *Smallest version that counts*, *How you know it's done*, *Why now* — and keep the probe text itself as plain as the quotes above. "The sacrificial boundary" and "the arrival signal" mean something to this playbook and nothing to the person answering. Full rule in the grilling playbook, *Say it in plain English*.
### Worked example — same idea, two destinations
**Idea:** "we should let people export audit logs"
**BAD destination**
> Let users export audit logs so they have their data.
Why it fails, concretely:
| Failure | Consequence downstream |
|---|---|
| No artifact named | Nobody knows whether the map clears at a plan or at shipped code |
| "Users" is not a role | The authorization question cannot even be phrased |
| No edge | Live streaming, SIEM push, and a schema redesign all argue their way in |
| No arrival signal | The Clearing Gate has nothing to check against |
| "their data" is a rationale, not a boundary | Every fog bullet reads as in scope |
**GOOD destination**
> A locked implementation plan for a compliance officer to export a filtered range of audit events from the admin UI and receive them as a single downloadable file. The map ends at the plan, not at shipped code. Continuous streaming to external systems is not on the route.
Three sentences, two lines of substance: artifact (a plan), actor (compliance officer), trigger surface (admin UI), shape of the result (one downloadable file), and an explicit edge (no streaming). Every one of those clauses will be cited later when you decide whether something is a question, fog, or out of scope.
**Write the destination into `map.md` verbatim as agreed.** Do not improve it afterwards. If it needs changing, change it with the human present — a silently redrawn destination invalidates every scope call already made.
---
## Step 3: Codebase recon
Recon is `legwork · AFK` — you do it alone, and you write it down **already resolved**. It exists so that no later session re-reads the same directories, and so that the handoff carries the ground truth `/plan2code-1-plan` Phase 2 (System Context Examination) would otherwise have to rediscover.
`questions/00-codebase-context.md` is always `00`. It always exists. It is created with `State: resolved` and a filled `## Answer` in the same write.
### What to explore
| Area | What to establish | Where to look |
|---|---|---|
| **Directory structure** | The map of the repo at the depth that matters for this destination — not every folder, the ones the work will touch | Top-level listing, then two levels into the relevant subtrees |
| **Key components** | The modules that would be read, changed, or called. Name, path, responsibility | Entry points, route/controller registries, service layers |
| **Patterns and conventions** | How this codebase does the thing you are about to plan: error handling, validation, config, module layout, naming, async style | Two or three recent files in the target area, plus `AGENTS.md` |
| **Integration points** | External systems, queues, storage, auth providers, feature-flag services the work will cross | Config files, environment variable references, client wrappers |
| **Technical debt in the blast radius** | Only debt the destination would collide with. Not a repo-wide audit | Long files in the target area, duplicated helpers, stale TODO markers with no owner |
| **System boundaries** | What this effort owns versus what it merely calls. Where the change stops | Package boundaries, ownership files, API contracts |
Two disciplines keep this file useful:
- **Verify behaviour against actual code, never against a filename.** A file called `auditLogger.ts` may log nothing.
- **Scope the recon to the destination.** A recon of the whole repo is unreadable and stale in a week. If a subtree cannot plausibly be touched by the destination, say so in one line and move on.
Record what you could **not** determine as an explicit gap. Gaps at recon time are often the first real fog bullets.
### The literal file
```markdown
> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.
# Codebase context
Type: legwork · AFK
State: resolved
Blocked by: none
Claimed: 2026-08-03 09:12
Resolved: 2026-08-03
Locked: no
## Question
What does this codebase already provide, constrain, and forbid for an operator-initiated
audit-log export? Establish structure, components, conventions, integrations, debt in the
blast radius, and boundaries — enough that no later session re-reads the same ground and
enough to hand to `/plan2code-1-plan` as its system context.
## Answer
### Directory structure
- `src/api/` — Express routers, one file per resource. `src/api/admin/` is the admin surface.
- `src/services/` — business logic; the only layer allowed to touch `src/db/`.
- `src/db/` — Knex query builders and migrations. `audit_events` lives here.
- `src/jobs/` — BullMQ workers. Existing precedent for long-running work.
- `src/web/admin/` — React admin UI, TanStack Query, colocated route components.
- `test/` — Vitest, mirroring `src/` one-to-one.
Untouched by this destination: `src/billing/`, `src/web/marketing/`.
### Key components
| Component | Path | Responsibility |
|---|---|---|
| `auditEvents.record()` | `src/services/auditEvents.ts` | Sole writer of `audit_events`; called from 31 sites |
| `adminRouter` | `src/api/admin/index.ts` | Mounts admin routes; applies `requireAdmin` |
| `requireAdmin` | `src/api/middleware/auth.ts` | Session check plus role check; no per-tenant scoping today |
| `reportQueue` | `src/jobs/reportQueue.ts` | BullMQ queue used by the existing billing report export |
| `signedUrl()` | `src/services/storage.ts` | S3 pre-signed URL helper, fixed 15-minute expiry |
### Patterns and conventions
- Errors: typed error classes thrown from services, mapped to HTTP by `errorHandler`. Never raw `res.status(500)`.
- Validation: Zod schema per route, exported next to the handler.
- Config: everything through `src/config.ts`; no direct `process.env` reads outside it.
- Async: `async`/`await` throughout. No callback style remains.
- Long-running work: enqueue to BullMQ, return `202` with a job id. Established by billing reports.
- Tests: Vitest, colocated fixtures, no shared mutable state between cases.
### Integration points
- **Postgres 15** via Knex. `audit_events` is ~180M rows, partitioned monthly.
- **Redis** backing BullMQ.
- **S3** for generated artifacts; the billing export already writes there.
- **SES** for transactional mail; templates in `src/mail/templates/`.
- No SIEM, log-shipping, or streaming integration exists today.
### Technical debt in the blast radius
- `audit_events` has an index on `(tenant_id, created_at)` but none on `actor_id`. Any
actor-filtered export will sequential-scan a partition.
- `requireAdmin` does not scope by tenant — a platform admin currently sees all tenants.
Any authorization decision here inherits that gap.
- The billing export writes CSV by hand-rolled string concatenation with no escaping.
Do not copy it.
### System boundaries
Owned by this effort: a read path over `audit_events`, an admin UI surface, an artifact
written to S3, and a delivery notification. Not owned: the write path (`auditEvents.record()`
is untouched), the audit event schema, tenancy semantics in `requireAdmin`.
### Gaps
- Retention policy for `audit_events` is not expressed anywhere in code. Someone outside
engineering owns it.
- No load figures exist for the largest tenant's monthly event count.
**Gist:** Node/Express/Knex/React with an established BullMQ-to-S3 export precedent from
billing; `audit_events` is 180M rows partitioned monthly with no `actor_id` index, and
`requireAdmin` has no per-tenant scoping.
## Evidence
- `src/api/admin/index.ts`, `src/api/middleware/auth.ts`
- `src/services/auditEvents.ts`, `src/services/storage.ts`
- `src/jobs/reportQueue.ts` and the billing export job it drives
- `src/db/migrations/20240914_partition_audit_events.js`
- `AGENTS.md` (conventions section)
```
---
## Step 4: The breadth-first frontier grill
The destination grill went **deep on one thing**. This grill goes **wide on everything**. They are different activities and mixing them is the most common way to produce a bad map.
| | Destination grill (Step 2) | Frontier grill (Step 4) |
|---|---|---|
| Goal | One or two settled lines | An inventory of open decisions |
| Movement | Drill until it is precise | Fan out until you stop finding new areas |
| Follow-ups | Chase every hedge | One clarifier at most, then move on |
| Success | The human commits to a boundary | You can name the areas, sharp and unsharp alike |
| Failure mode | Accepting a slogan | Solving a question instead of finding the next one |
**You are not resolving anything here.** You are taking inventory. The instant an answer feels satisfying, you are probably going deep.
### Recognising that you have gone deep
Watch for these. Any one of them means stop and pull back:
- You have asked three consecutive probes about the same area.
- You are discussing an implementation detail (a column type, a library, a retry count) rather than a decision.
- You are proposing a design instead of asking what has to be decided.
- The human is enjoying it. Depth is more fun than breadth; that is exactly why it steals the session.
- You have written something down that reads like an answer.
### Pulling back
Say it out loud so the human tracks the move, then jump:
> "Good — that is one for the map, not for now. Parking it as *Export format*. Different corner: who is allowed to run an export at all?"
Two mechanics keep the fan-out honest:
1. **Round-robin the areas.** Before you start, list the axes you intend to cross — engineering axes (data, surface, permissions, volume, delivery, failure, operations, testing) AND product axes (user value, success measurement, target user, rollout, analytics, support & docs, legal/compliance). Not every axis applies to every destination: name the ones that do before the first probe, say which you are skipping and why, then take one probe per axis before any second probe on any axis.
2. **Ask for the axis you have not touched.** Near the end: "What have I not asked about that would embarrass us to discover in week three?"
**Breadth-first is the ideal batch.** One probe per axis means the probes are independent by construction — that is what breadth-first *means* — so this grill should run as batches of three, not as a stream of singles. If you catch yourself wanting to batch two probes on the same axis, that is depth wearing a batch's clothes; pull back.
**With Product in the room, the product axes are not optional.** A joint Product + Engineering session that only fans out across engineering axes produces a map Product cannot answer for — and vice versa. Alternate: an engineering batch, then a product batch, so neither team spectates for three turns straight.
### Sample breadth probes
Each opens a different axis. Batch three at a time in list order (skipping axes you ruled inapplicable), with probe 7 riding alongside the "what have I not asked about" closer above; note each answer and move.
**Every batch goes through the structured question tool.** The output is a *sort* into sharp question or fog, so the option descriptions carry the framing and the built-in `Other` path captures elaboration the named choices miss. Never ask these probes as a prose list when the tool exists.
1. **Data** — "What is the smallest and largest thing an operator could reasonably ask for in one export? Give me both ends."
2. **Surface** — "Where does this start: a button in the admin UI, a scheduled thing, an API call someone scripts?"
3. **Permissions** — "Who is allowed to run one, and can they export events about people other than themselves?"
4. **Volume and time** — "If an export takes four minutes, is that fine, bad, or a redesign?"
5. **Delivery and failure** — "The export succeeds but the download link expires before they click it. What should have happened?"
6. **Operations** — "Six months from now someone asks who exported what. Does this feature audit itself?"
7. **Testing** — "What would you need to see pass before you would let this near a customer's compliance data?" *(This one always runs — see the mandatory testing-posture question below.)*
The product axes, same style — one sample probe each:
8. **User value** — "A compliance officer gets this feature Monday morning. What stops being painful for them that day?"
9. **Success measurement** — "Three months after ship, what number or behavior tells you this was worth building?"
10. **Target user** — "Who is this for first — every tenant, one segment, one named customer who asked?"
11. **Rollout** — "Does this land for everyone at once, behind a flag, or piloted with one tenant first?"
12. **Analytics** — "What do you want to know about how people actually use it — and is any of that instrumented today?"
13. **Support & docs** — "A support ticket about this lands in week one. What does the agent need to see to answer it without escalating?"
14. **Legal / compliance** — "Does anything here touch data retention, privacy, or a regulator's definition of a record?"
Record each answer as one line in your working notes with an area label. At the end of the grill you will have two piles: lines you can turn into a sharp question, and lines you cannot. The second pile is the fog.
---
## The fog-vs-question test
This is the single most important judgement in the skill. Get it wrong toward questions and the map fills with unanswerable stubs that block the frontier. Get it wrong toward fog and the map has nothing takeable on it.
> **The test is whether you can STATE the question precisely now — not whether you can ANSWER it now.**
- **Write a question file** when the question is already sharp — *even if it is blocked and nobody can act on it yet*. Blocked questions are real questions; they get `Blocked by:` and a `[!]` row and they wait. Blocked is not the same as unformed.
- **Leave it in `## Not yet specified`** when you cannot yet phrase it that sharply. You can see there is something there; you cannot say what is being asked.
**Do not pre-slice the fog.** A fog patch is deliberately coarser than a question. One patch may graduate into three questions, or one, or none once the frontier reaches it. Splitting fog into question-shaped fragments before it is sharp invents a structure that the answers will contradict, and it costs a later session the work of deleting your guesses. Write the patch as loosely as the view allows.
A useful forcing check: **could a different person, reading only this line, know what a good answer looks like?** If yes, it is a question. If they would have to ask you what you meant, it is fog.
### Worked examples
| Candidate | Verdict | Reasoning |
|---|---|---|
| "CSV or JSONL for the export file?" | **Question** | Two named options, one decision, one sitting. A reader knows what an answer looks like. `grill · HITL`. |
| "Which roles may export events about other users?" | **Question**, blocked | Sharp today, but it depends on the tenant-scoping decision. Write it, set `Blocked by:`, mark the row `[!]`. Blockedness never demotes a sharp question to fog. |
| "Something about how big exports behave" | **Fog** | "Big" has no meaning until the volume ceiling lands. You cannot say whether the question is about pagination, streaming, timeouts, or refusal. One line in `## Not yet specified`. |
| "There is probably something about PII redaction" | **Fog** | The area is visible, the question is not. Once Legal answers, this may graduate into *which fields are redacted*, *who configures it*, and *does redaction apply to the actor or the subject* — or into nothing, if the answer is "export raw." Slicing it now guesses all three. |
| "How should we architect the export pipeline?" | **Neither — split it** | No single answer closes it; it bundles at least four decisions (sync vs queued, storage target, artifact lifetime, notification). Ask what the parts are. The sharp parts become questions, the rest becomes fog. A candidate no single answer closes is not a question. |
**What never belongs in `## Not yet specified`:** anything already decided (it is a resolved question with a gist on its row), anything that already has a question file, and anything past the destination (that is out of scope).
---
## Out of scope versus fog
Fog gathers **only toward the destination**. The destination fixes the scope, so work beyond it is not dim — it is *excluded*. It is not fog, and it must never sit in `## Not yet specified`, where a later session would try to graduate it.
**The distinction is SCOPE, not sharpness.** This is the part people get wrong. An out-of-scope item can be perfectly sharp — "should the export push to Splunk on a schedule?" is a crisp question with a crisp answer. It is still out of scope, because the destination said the map ends at an operator-initiated export. Sharpness decides *fog versus question*. Position relative to the destination decides *in scope versus out*.
| | Fog (`## Not yet specified`) | Out of scope (`## Out of scope`) |
|---|---|---|
| Position | Before the destination | Past the destination |
| Why it is not a question | Cannot be phrased sharply yet | Could be phrased perfectly — it just is not ours |
| Future | Graduates into questions as the frontier advances | Never graduates |
| Reopening | Automatic, as answers land | Only if the destination is redrawn — and then as a fresh effort, not a resumption |
| The act | An admission of ignorance | A scoping decision |
Ruling something out of scope is a **scoping act, not a step on the route**. When a question you already created turns out to sit past the destination — mis-scoped in during charting, or exposed by a later answer — run `rule-out <NN> --reason "<why>"` (resolve.md, Ruling a question out of scope mid-work): it sets `State: out-of-scope`, marks its row `[-]` and leaves one linked line in `## Out of scope` with the reason. Then handle anything it reports `stranded`. It does not get an `## Answer` and it is not a decision the route walked.
Watch for **stranded** questions: a live question whose `Blocked by:` names something now out of scope will never unblock. Re-frame its `## Question` to drop the dependency, or rule it out too. Never leave it sitting.
---
## Step 6: Numbering and dependency order
Upstream wayfinder creates every unit first and wires the blocking edges in a **second pass**, because a server-side tracker assigns ids and nothing can reference a sibling until it has one. On `**Backend:** local` that constraint does not exist — **you choose `NN` yourself**, so charting is a **single pass**: decide the order, then write each file complete, `Blocked by:` filled at the moment of writing.
(On `**Backend:** github` the constraint comes back, and so does the two-pass shape. Rules 1, 6, 7, and 8 below still hold — they are about dependency reasoning, not about ids. Rules 2-5, which are about `NN`, are replaced by sub-issue order; see `github-issues.md`.)
The rules:
1. **Sort by dependency before you write anything.** Sketch the edges on paper first: which decisions must land before which others can even be discussed.
2. **Blockers get lower numbers.** If *Row-count ceiling* blocks *Delivery channel*, the ceiling is `02` and delivery is `05`. This makes `Blocked by: 02` readable at a glance and makes "lowest `NN` first" on the frontier a sane traversal order.
3. **`00` is always the codebase context.** Never anything else.
4. **`NN` is never reused and never renumbered.** Not when a question is ruled out of scope, not when one is deleted, not to close a gap in the sequence. Links and `Blocked by:` lines would rot silently. Gaps in the numbering are normal and harmless.
5. **The next number is max + 1**, and `frontier` reports it as `nextNN`: one past the highest `NN` in the directory listing, on any `Blocked by:` line, or in any `questions/NN-` link in the map or a question file, and never below the map's `**Next NN:**` line. That line is a high-water mark the script keeps (every write command updates it, and it only rises), so even a deleted question that nothing links to any more never has its number handed out again. Never edit it by hand.
6. **Only depend on what genuinely gates the question.** A `Blocked by:` chain that is really a preference for reading order strangles the frontier. Ask: could this question be answered — badly but honestly — without the blocker? If yes, it is not blocked.
7. **Cycles are a phrasing bug.** If A blocks B and B blocks A, the two are one decision. Merge them or re-frame one to drop the edge.
8. **Refer by name in prose.** Bare numbers appear only on `Blocked by:` lines.
A worked ordering for the audit-log export map:
| `NN` | Name | Type | Blocked by | Why here |
|---|---|---|---|---|
| `00` | Codebase context | `legwork · AFK` | none | Always first, always resolved |
| `01` | Export format | `grill · HITL` | none | Nothing gates it; it gates the artifact shape |
| `02` | Row-count ceiling | `research · AFK` | none | A fact about the data, independent of every preference |
| `03` | Export authorization | `grill · HITL` | none | Independent axis; can be argued today |
| `04` | Testing posture | `grill · HITL` | none | Mandatory; independent of everything else |
| `05` | Delivery channel | `grill · HITL` | `02` | Synchronous download versus queued link turns entirely on volume |
| `06` | SIEM push connector | `grill · HITL` | none | Created, then immediately ruled out of scope during charting — `State: out-of-scope`, no `## Answer` |
---
## The `map.md` template
Below is a complete, realistic map for the audit-log export effort **partway through MODE B**, after two questions have resolved — it shows every marker in use. At Chart Step 5 the same file carries `**Status:** Charting`, an empty `## Question Checklist`, and no `[x]` row except `00`. Copy the structure exactly, including the HTML comments — they are written **for the next session**, which has no memory of this one.
`**Status:**` is `Charting` while Step 5 and Step 6 run, becomes `Working` at Step 7, and becomes `Cleared` only at handoff. It is how the orchestrator routes a fresh session, so it must be accurate before you stop.
Set `**Confidence:**` honestly at Step 5 and re-score it every time a question resolves. Charting scores are low by construction — that is the point. The Clearing Gate needs every dimension at 18/25 or better. Never inflate to make the gate pass.
Write the four dimension labels **hyphenated exactly as shown** — `Requirements-clarity`, `Feasibility-technical`, `Integration-points`, `Risk-assessment` — in the `NN/25` form, with no percent symbol anywhere in the file. The metrics collector scrapes planning documents by regex for a bare dimension word followed by whitespace, a colon, or a pipe and then digits; the hyphen breaks that match. A percent sign or a bare `Requirements 18` would be ingested as a completed planning step's confidence score that no planning step ever produced.
Say once, at Step 5: *"This map lives in gitignored `specs/` — local to you, not shared. `git add -f` it to track it."*
```markdown
> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.
# Map: audit-log-export
*Decisions live in `questions/` — one file each. This file is just the index.*
**Status:** Working
**Updated:** 2026-08-03
**Confidence:** Requirements-clarity 18/25 · Feasibility-technical 14/25 · Integration-points 16/25 · Risk-assessment 14/25
**Next NN:** 07
<!-- Status: Charting while the map is being built (Chart Steps 1-6) -> Working once the
checklist is indexed (Chart Step 7) -> Cleared only when the Clearing Gate passes.
A fresh session routes on this line, so it must be correct before the session ends. -->
<!-- Next NN: the number the next new question file takes. Kept by the script (reconcile
adds it at Chart Step 7 and every write command raises it); it never goes down, so a
deleted question's number is never reused. Leave it out at Step 5; never edit it. -->
<!-- Confidence: four dimensions, each scored out of 25, re-scored at every resolution.
The Clearing Gate requires all four at 18/25 or better. Score against evidence. -->
## Destination
A locked implementation plan for a compliance officer to export a filtered range of audit
events from the admin UI and receive them as a single downloadable file. The map ends at the
plan, not at shipped code. Continuous streaming to external systems is not on the route.
<!-- Settled at Chart Step 2 and quoted as agreed. Every question is measured against it:
in scope or past the edge, still needed or now moot. Change it only with the human
present — a silent redraw invalidates every scope call already made. -->
## Ground rules
<!-- Standing constraints for every session on this map. Read before choosing a question,
obeyed while resolving it. Nothing here is re-asked. The first line records AGENTS.md:
"- `AGENTS.md` exists and governs." or, when it is missing, exactly
"- `AGENTS.md` is absent." (the handoff then routes to /plan2code-init first). -->
- `AGENTS.md` exists and governs. Its conventions are not re-litigated by any question here.
- One question _file_ at a time; the fork-menu between decisions; a fresh session recommended after ~3. `research` questions may run as parallel subagents.
- Every user-facing question uses the structured question tool when available; probes batch at most three. Prose Q blocks are fallback only when the tool is unavailable.
- Questions are put to the human in plain English. Technical terms only where the term is the decision.
- HITL questions are answered by the human in their own words. Never self-answered.
- No new runtime dependency is assumed without a `research` question backing it.
- Compliance language is reviewed by Dana before anything user-facing is finalised.
- Sketches are throwaway and live only under `pathfinder/sketch-NN/`.
## Glossary
<!-- Terms this effort uses precisely. Prevents two sessions meaning different things by
the same word — the cheapest correctness win on the whole map. -->
| Term | Meaning here | Avoid |
|---|---|---|
| Audit event | One row in `audit_events`: actor, tenant, action, target, timestamp, payload | log line, activity record |
| Compliance officer | Tenant-scoped role that reviews activity; not a platform admin | admin, auditor |
| Export | One operator-initiated request producing one artifact for one filtered range | download, dump, extract |
| Retention window | How far back `audit_events` is queryable; owned outside engineering | archive period |
| Signed artifact | The generated file plus a checksum a recipient can verify independently | signed file, bundle |
## Question Checklist
<!-- Rebuilt from questions/ every session — the files are ground truth, this is an index.
[ ] open (the frontier) · [/] claimed · [x] resolved · [!] open but blocked
[-] out of scope. Resolved rows carry the one-line gist from the question's Answer,
then the Resolved: date in italics — briefs filter on these dates. -->
- [x] [Codebase context](./questions/00-codebase-context.md) — Node/Express/Knex/React with a BullMQ-to-S3 export precedent; `audit_events` is 180M rows partitioned monthly, no `actor_id` index, `requireAdmin` has no tenant scoping. *(2026-08-03)*
- [x] [Export format](./questions/01-export-format.md) — CSV with a UTF-8 BOM and RFC 4180 quoting, plus a sidecar SHA-256 manifest; JSONL rejected because recipients open these in Excel. *(2026-08-03)*
- [/] [Row-count ceiling](./questions/02-row-count-ceiling.md)
- [ ] [Export authorization](./questions/03-export-authorization.md)
- [ ] [Testing posture](./questions/04-testing-posture.md)
- [!] [Delivery channel](./questions/05-delivery-channel.md) — Blocked by 02
- [-] [SIEM push connector](./questions/06-siem-push-connector.md) — out of scope, see below
## Not yet specified
<!-- The fog: in-scope areas you can see but cannot yet phrase as a question. Graduates into
question files as answers land, and the graduated bullet is deleted from here.
Do NOT pre-slice these into question-sized pieces — one bullet may become three
questions, or none. Nothing already decided, already a question, or out of scope. -->
- How far back an export may reach. There is a retention answer somewhere outside engineering
and nobody has it yet; until then we cannot say whether the question is about a hard limit,
a warning, or a per-tenant setting.
- Redaction of event payloads. Legal may say "export raw", in which case this evaporates —
or it may become several decisions about which fields, who configures them, and whether the
rule follows the actor or the subject. Revisit after Export authorization.
- What happens when an export range straddles a monthly partition that was migrated mid-range.
Cannot phrase this sharply until Export format is applied to a real query plan.
- Whether the export feature audits itself, and if so at what granularity. Suspect this is one
small question but it may turn on the authorization model.
## Out of scope
<!-- Work consciously ruled past the destination. Never graduates; returns only if the
destination is redrawn, and then as a fresh effort. One line each: gist plus why. -->
- [SIEM push connector](./questions/06-siem-push-connector.md) — scheduled push to Splunk or
similar. The destination ends at an operator-initiated export; anything continuous is a
different effort with a different owner.
- Redesign of the `audit_events` schema. The write path is untouched by this destination;
changing it would pull in all 31 call sites of `auditEvents.record()`.
- Adding the missing `actor_id` index. Real, and it will hurt, but it is a database change
with its own review path. Recorded here so the plan can reference it as a dependency
rather than absorb it.
```
---
## The question-file template
Six contiguous `Key: value` lines after the H1. Not YAML. No frontmatter delimiters. No `- [ ]` checkboxes anywhere inside a question file — use plain bullets, including for legwork checklists.
`## Question` is written at charting. `## Answer` is appended only when the question resolves. `## Evidence` holds sources, links, and artifacts, and a `research` subagent writes into it during Chart Step 8 without deciding anything.
### An open question
```markdown
> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.
# Export authorization
Type: grill · HITL
State: open
Blocked by: none
Claimed: none
Resolved: none
Locked: yes
## Question
Who may run an audit-log export, and over whose events?
Three sub-decisions, all of which must land together because any two of them constrain the third:
- Which role gates the export action — the existing `admin` role, a new `compliance` role, or
a per-tenant grant?
- May an exporter include events where they are the actor, or must self-events be excluded to
keep the export usable as evidence?
- `requireAdmin` currently does not scope by tenant, so a platform admin sees every tenant's
events. Does the export inherit that, or does it enforce a tenant scope the rest of the
admin surface does not?
Recommended answer to react to: a new tenant-scoped `compliance` role; self-events included
but flagged in a column; the export enforces tenant scope even though the surrounding admin
surface does not.
Marked `Locked: yes` — the third sub-decision creates a precedent that later admin features
will follow, and reversing it later means re-auditing every export already delivered.
## Evidence
- `src/api/middleware/auth.ts` — `requireAdmin` checks session and role, no tenant predicate.
- Codebase context records the same gap under technical debt.
```
### A resolved question
```markdown
> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.
# Export format
Type: grill · HITL
State: resolved
Blocked by: none
Claimed: 2026-08-03 10:41
Resolved: 2026-08-03
Locked: yes
## Question
What file format does an export produce, and what does a recipient need in order to trust the
file has not been altered?
## Answer
**Decision.** CSV, UTF-8 with a byte-order mark, RFC 4180 quoting, one header row, timestamps
in ISO 8601 UTC. Alongside it a sidecar `.sha256` manifest listing the artifact filename and
its digest.
**Rejected.**
- *JSONL* — better for nested payloads and trivially streamable, but every named recipient
opens these in Excel and would need a conversion step during an audit. The people who
prefer JSONL are not the people receiving the file.
- *XLSX* — solves the Excel encoding problems outright, but adds a generation library and
makes byte-level verification of the artifact meaningfully harder.
- *Detached signature instead of a checksum* — real integrity guarantees, but requires key
management nobody has scoped, and no recipient has asked to verify a signature.
**Consequences.**
- Nested `payload` is flattened to one JSON string column. Anyone needing structure parses
that column.
- The BOM is required or Excel mangles non-ASCII actor names. This must be an explicit test.
- Do not reuse the billing export's CSV writer — it concatenates strings with no escaping.
A quoting-correct writer is now in scope for the plan.
- The checksum makes the artifact self-verifying, which lets Delivery channel consider a
short-lived link without weakening the integrity story.
**Gist:** CSV with a UTF-8 BOM and RFC 4180 quoting, plus a sidecar SHA-256 manifest; JSONL
rejected because recipients open these in Excel.
## Evidence
- RFC 4180, sections 2.5-2.7 (quoting and embedded delimiters).
- `src/jobs/billingExport.ts` — the hand-rolled writer that must not be copied.
- Dana confirmed on 2026-08-03 that external auditors accept a published checksum.
```
---
## Step 4: The no-fog off-ramp
If the breadth-first grill surfaces **no fog** — every area you opened produced either a settled answer or a question you could phrase immediately, and `## Not yet specified` would be empty — then the way to the destination is already visible. The journey is small enough to plan directly and a map would be pure overhead.
Do this, in order:
1. **Keep `questions/00-codebase-context.md`.** This is the important part. It is exactly what `/plan2code-1-plan` Phase 2 (System Context Examination) has to produce anyway, and you have already produced it. Throwing the recon away to "clean up" wastes the most valuable artifact of the session.
2. **Do not create `map.md`.** The off-ramp fires at Step 4 and the map is not written until Step 5, so in the normal flow there is nothing to delete — just stop before writing it. (If you reached Step 4 with a `map.md` already on disk, delete it: a map with empty fog and no open questions will confuse the next session into resuming something that does not exist.)
3. **Tell the user plainly what happened and what to do:**
> "No fog surfaced — the way to the destination is already visible, so this does not need a map. I kept the codebase recon at `specs/audit-log-export/pathfinder/questions/00-codebase-context.md`; attach that file to a `/plan2code-1-plan` session and it covers Phase 2 outright."
4. **Stop.** Do not chart anyway "just in case", do not create questions, do not start planning in this session.
Be honest about the trigger. If two areas are genuinely unformed, that is fog and the map earns its keep. The off-ramp is for the case where you fanned out across every axis and kept landing on solid ground.
---
## Step 6: The mandatory testing-posture question
**Every map includes a `grill · HITL` question on testing posture.** No exceptions, including maps where testing feels obvious.
The reason is mechanical: `/plan2code-1-plan` Phase 1 asks for three things — testing types, whether tests run after each implementation phase, and a coverage target — and `/plan2code-2-document` **string-matches** that answer, either appending a testing block to every phase, creating a dedicated final testing phase, or omitting testing entirely. A map that clears without this answer hands the human a planning session that stalls on its first question. Ask it while there is still someone in the room.
Record the answer in the literals downstream matches on, not in paraphrase. ("Phase" here names `/plan2code-1-plan`'s implementation phases and is a downstream contract string — it is never a pathfinder unit of work.)
It is almost never blocked. Give it whatever number the dependency ordering leaves free, and expect it to sit on the frontier from day one.
```markdown
> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.
# Testing posture
Type: grill · HITL
State: open
Blocked by: none
Claimed: none
Resolved: none
Locked: no
## Question
What testing does this work carry, so `/plan2code-1-plan` Phase 1 can be answered without
stalling? Three parts, all needed:
- **Types** — unit, integration, end-to-end, some combination, or none.
- **Phase testing** — record one of the two literals `/plan2code-2-document` matches:
`Run after each phase` (a testing block closes every implementation phase) or
`Dedicated phase only` (one final testing phase).
- **Coverage target** — record one of Phase 1's three literals: `Critical paths`,
`Moderate (~60-80%)`, or `Comprehensive (>80%)`.
Recommended answer to react to: unit plus integration; `Run after each phase`;
`Critical paths`. Rationale — this touches compliance data, so the correctness of the
CSV writer and the authorization predicate must be pinned by tests, but the admin UI is thin
enough that end-to-end coverage would cost more than it catches.
Two specific cases worth naming in the answer regardless of the general posture, because
Codebase context shows both are easy to get wrong here:
- The UTF-8 BOM survives Excel round-tripping for non-ASCII actor names.
- The tenant-scope predicate actually excludes other tenants' events, asserted against seeded
cross-tenant data rather than a mock.
Note for whoever resolves this: the coverage target is a number the human owns. Do not infer
it from the codebase's current coverage, and do not soften it to whatever the repo already
achieves.
## Evidence
- `AGENTS.md` records Vitest as the runner with fixtures colocated under `test/`.
- Codebase context: no end-to-end harness exists today; adding one is a real cost, not a flag.
```
@@ -0,0 +1,474 @@
# GitHub Issues Backend
> Part of plan2code-0-pathfinder — loaded at the top of EVERY session whose map lives on GitHub Issues. It re-expresses the local-file model in issue terms: where the map lives, where a question lives, how blocking, claiming, and resolving are done, and what stays on local disk regardless.
>
> **Local-file maps never load this file.** If `## Ground rules` says `**Backend:** local`, close it and use `questions.md`.
Everything the skill says about *judgement* is unchanged by the backend: the fog-vs-question test, the destination grill, one question at a time, the decide-menu-checkpoint cadence, HITL is never self-answered, the Clearing Gate rubric. This file changes only *where the bytes go*.
---
## Why a second backend exists
Local files are private scratch — `specs/` is gitignored, so the map is yours alone and nobody else can see it, comment on it, or resolve a question in parallel. That is the right default for a solo effort.
A map on GitHub Issues buys three things local files cannot:
| | Local files | GitHub Issues |
|---|---|---|
| Visibility | One machine, one person | Anyone with repo access, in a UI they already have open |
| Blocking | A `Blocked by:` line only an agent reads | Native issue dependencies — GitHub greys out blocked issues in its own UI |
| Concurrency | One session at a time by construction | Several people can work unblocked questions at once; the assignee is a real lock |
It costs three things too, and the human must know all three before they pick it:
1. **Issues on a public repo are public.** The destination, the rejected alternatives, the codebase recon, the technical debt in the blast radius — all of it is world-readable the moment it is written. Never chart to a public repo's tracker anything that would embarrass the project or leak a customer.
2. **It writes to shared state.** A local map costs nothing to abandon. Twelve stale issues labelled `pathfinder:grill-hitl` on a team's tracker is litter someone has to clean.
3. **It needs `gh`, auth, and issues enabled.** More that can break, in a step whose whole job is to remove friction.
---
## Preflight — before the first write
Run these once, at Chart Step 1, **before** offering GitHub as an option. Any failure means GitHub is not offered at all; say why in one line and continue with local files.
| # | Check | Command | On failure |
|---|---|---|---|
| 1 | `gh` is installed | `gh --version` | Not offered — "no `gh` on this machine" |
| 2 | Authenticated | `gh auth status` | Not offered — "`gh` is not logged in" |
| 3 | Inside a repo with a GitHub remote | `gh repo view --json nameWithOwner,visibility,hasIssuesEnabled` | Not offered — "no GitHub remote here" |
| 4 | Issues are enabled | same call, `hasIssuesEnabled` | Not offered — "issues are disabled on this repo" |
| 5 | Write access | `gh api repos/<owner>/<repo> --jq .permissions.push` | Not offered — read-only access cannot chart |
Record `nameWithOwner` and `visibility` from check 3 — **`visibility` is not optional detail.** If it is `PUBLIC`, the offer must say so in the same breath, e.g. *"GitHub Issues — note `jparkerweb/plan2code` is public, so the whole map is world-readable."*
Once a map exists, preflight shrinks to checks 1 and 2. A session that cannot reach `gh` cannot work a GitHub map: say so and stop, rather than silently starting a local one.
### Labels
Create the label set at **Chart Step 5**, with the map issue — never during preflight, which must stay read-only until the human has actually picked `github`. `--force` makes it idempotent, so it is safe to re-run every session:
```bash
gh label create "pathfinder:map" --color 5319E7 --description "Pathfinder map" --force
gh label create "pathfinder:grill-hitl" --color 1D76DB --description "Decision only the human can make" --force
gh label create "pathfinder:research-afk" --color 0E8A16 --description "Fact-finding, agent alone" --force
gh label create "pathfinder:sketch-hitl" --color FBCA04 --description "Human reacts to something concrete" --force
gh label create "pathfinder:legwork-hitl" --color D93F0B --description "Manual work needing a human" --force
gh label create "pathfinder:legwork-afk" --color D93F0B --description "Manual work the agent can do" --force
gh label create "pathfinder:locked" --color B60205 --description "Hard to reverse; consequences recorded" --force
gh label create "pathfinder:out-of-scope" --color CFD3D7 --description "Ruled past the destination" --force
```
**Type and mode share one label** — `grill-hitl`, not `grill` plus `hitl` — for the same reason the local `Type:` line is one token: two labels can drift apart, and a `research` question that has quietly become HITL is a question nobody is driving.
---
## The equivalence table
This is the whole mapping. Everything below expands a row.
| Local file model | GitHub Issues model |
|---|---|
| `specs/<idea>/pathfinder/map.md` | One issue, labelled `pathfinder:map`, titled `Map: <idea>` |
| `questions/NN-<slug>.md` | A **sub-issue** of the map, titled with the question name |
| `NN` ordering | The map's sub-issue order — the order they were created, which is dependency order |
| `## Question` in the file | The issue body |
| `## Answer` appended | A comment on the issue, opening `## Answer` |
| `## Evidence` | A comment opening `## Evidence` (a research subagent writes its own) |
| `Type:` line | The `pathfinder:<type>-<mode>` label |
| `State: open` | Issue open, **no assignee** |
| `State: claimed` | Issue open, **assigned** |
| `State: resolved` | Issue **closed as completed**, with an `## Answer` comment |
| `State: out-of-scope` | Issue **closed as not planned**, labelled `pathfinder:out-of-scope`, no `## Answer` |
| `Blocked by: 02, 04` | Native issue dependencies (`dependencies/blocked_by`) |
| `Locked: yes` | The `pathfinder:locked` label |
| `Claimed: <timestamp>` | GitHub's own assignment event in the timeline |
| `Resolved: <YYYY-MM-DD>` | The issue's own close timestamp (`closedAt`) — nothing to write, nothing to backfill |
| `## Question Checklist` in `map.md` | **Nothing** — the frontier is a live query, not a written list |
| `## Not yet specified`, `## Out of scope`, `## Ground rules`, `## Destination`, `## Glossary` | The same sections, in the map issue body |
| `sketch-NN/` | Still local disk — see *What stays on local disk* |
| `PLAN-DRAFT-<YYYYMMDD>.md` | Still local disk — see *Handoff* |
### The checklist is deleted, not ported
In local mode `map.md` carries a `## Question Checklist` because a directory of files has no queryable state. GitHub has queryable state, so **the map issue body carries no checklist at all.** Closed questions get one line each under `## Decisions so far`; open questions are not listed anywhere.
This kills the single largest source of drift in the local backend — a checklist that disagrees with the files — and it is why Work Step 2's reconcile pass is much shorter here.
---
## Refer by name
Unchanged, and harder to get right here because GitHub hands you a number for everything. In prose the human reads, write `[Export format](https://github.com/o/r/issues/42)` — never `#42`, never "issue 42", never a bare number. A wall of `#42, #43, #44` is illegible; names read at a glance.
Bare `#<n>` appears in exactly two places: inside a fallback `Blocked by:` body line when native dependencies are unavailable, and inside a `gh` command.
---
## Chart Steps 3-4 — hold the recon, protect the off-ramp
In `local` mode Step 3 writes `questions/00-codebase-context.md` the moment the recon is done, because a file in gitignored scratch costs nothing if the session then takes the Step 4 off-ramp. **On a shared tracker it costs something**: a stray issue nobody asked for, on a repo other people are watching.
So on `github`, Step 3 does the recon and **holds it in the session**. It becomes an issue at Step 6, alongside the other questions.
If the Step 4 breadth-first grill surfaces **no fog**, the off-ramp fires before anything has been created:
1. Write the recon to `specs/<idea>/pathfinder/questions/00-codebase-context.md` — a **local file**, exactly as the local backend would. It is what `/plan2code-1-plan` Phase 2 needs, and it is too valuable to throw away.
2. Create **nothing** on the tracker. No map issue, no question issues, no labels.
3. Tell the user plainly and STOP.
The tracker only ever sees an effort that earned a map.
---
## Creating the map (Chart Step 5)
Title is `Map: <idea>` — the kebab-case idea name, verbatim, so `gh issue list --label pathfinder:map` reads as an index of efforts.
```bash
gh issue create --label "pathfinder:map" --title "Map: audit-log-export" --body-file - <<'EOF'
> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.
**Status:** Charting
**Updated:** 2026-08-08
**Confidence:** Requirements-clarity 8/25 · Feasibility-technical 6/25 · Integration-points 6/25 · Risk-assessment 5/25
<!-- Status: Charting (Chart Steps 5-6) -> Working (Chart Step 7) -> Cleared at the gate.
A fresh session routes on this line, so it must be correct before the session ends. -->
## Destination
A locked implementation plan for a compliance officer to export a filtered range of audit
events from the admin UI and receive them as a single downloadable file. The map ends at the
plan, not at shipped code. Continuous streaming to external systems is not on the route.
## Ground rules
- **Backend:** github — this issue is the map; questions are its sub-issues.
- `AGENTS.md` exists and governs. Its conventions are not re-litigated by any question here.
- One question _issue_ at a time; the fork-menu between decisions; a fresh session recommended after ~3. `research` questions may run as parallel subagents.
- Every user-facing question uses the structured question tool when available; probes batch at most three. Prose Q blocks are fallback only when the tool is unavailable.
- HITL questions are answered by the human in their own words. Never self-answered.
- Sketches are throwaway and live on local disk only, under `specs/audit-log-export/pathfinder/sketch-<issue>/`.
## Glossary
| Term | Meaning here | Avoid |
|---|---|---|
| Audit event | One row in `audit_events`: actor, tenant, action, target, timestamp, payload | log line |
## Decisions so far
<!-- The index — one line per CLOSED question: enough to judge relevance, then open the
issue for the detail. Open questions are NOT listed; they are open sub-issues. -->
## Not yet specified
<!-- The fog: in-scope areas you can see but cannot yet phrase as a question. Graduates into
sub-issues as answers land, and the graduated bullet is deleted from here. -->
## Out of scope
<!-- Work consciously ruled past the destination. Never graduates. One line each: gist plus why. -->
EOF
```
Two things that must be exact:
- **`**Status:**` is still a literal line in the body.** It is how a fresh session routes, exactly as in local mode. `Charting` → `Working` → `Cleared`.
- **`**Backend:** github` is the first `## Ground rules` bullet.** It is how a fresh session knows to load this file at all. Without it, a session that opens the map issue has no way to know which playbook it is in.
Confidence keeps the hyphenated `Requirements-clarity 8/25` form for the same reason it does in local mode — the metrics collector scrapes bare dimension words followed by digits, and would ingest a planning confidence nobody scored.
**Say once, at Step 5:** *"The map lives on `<owner>/<repo>`'s issue tracker — `<PUBLIC or PRIVATE>`, so `<world-readable / visible to anyone with repo access>`. Everything charted here is visible there."*
---
## Creating the questions (Chart Step 6) — two passes, not one
The local backend charts in a **single pass** because you choose `NN` yourself and can write `Blocked by: 02` into a file before `02` exists. **On GitHub that is impossible** — an issue has no id until the server assigns one, and a dependency edge needs the blocker's id. So charting here reverts to upstream's shape:
**Pass 1 — create every question issue, in dependency order.** Blockers first. The creation order becomes the sub-issue order, which becomes the reading order for the frontier and the trail, so it is doing the job `NN` does locally. Capture each new issue's number *and* database id as you go.
```bash
# Create, capturing the URL; the number is its last path segment.
gh issue create --label "pathfinder:grill-hitl" --title "Export format" --body-file - <<'EOF'
> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.
## Question
What file format does an export produce, and what does a recipient need in order to trust
the file has not been altered?
Recommended answer to react to: CSV with a UTF-8 BOM plus a sidecar SHA-256 manifest.
EOF
# The database id — needed for BOTH wiring steps below. Not the #number, not the node_id.
gh api repos/<owner>/<repo>/issues/<number> --jq .id
```
**Pass 2 — wire the structure.** Two edges per question, both keyed on **database ids**:
```bash
# a) Attach as a sub-issue of the map. sub_issue_id is the CHILD's database id.
gh api --method POST repos/<owner>/<repo>/issues/<map-number>/sub_issues \
-F sub_issue_id=<child-db-id>
# b) Add each blocking edge. issue_id is the BLOCKER's database id.
gh api --method POST repos/<owner>/<repo>/issues/<blocked-number>/dependencies/blocked_by \
-F issue_id=<blocker-db-id>
```
**The database id is the single most common failure in this backend.** `gh api repos/o/r/issues/42 --jq .id` returns something like `2716143027`. The `42` is the *number*; `I_kwDO...` is the *node id*. The node id is rejected outright. The *number* is worse: a small integer like `42` is itself a perfectly valid database id — of some unrelated issue created years ago — so the call can succeed and silently attach the wrong thing. Fetch `.id` for every issue you are about to reference, and never hand-assemble one.
Charting still writes `00-codebase-context`'s equivalent — the Step 3 recon you held. Create it **first**, labelled `pathfinder:legwork-afk`, post the recon as an `## Answer` comment, and close it as completed in the same pass. It is resolved on arrival, exactly as in local mode, and it is what the handoff's `## System Context` is built from.
### If the endpoints are unavailable
Sub-issues and dependencies are recent GitHub features. On an instance that rejects either endpoint, fall back in the body — and say plainly, once, that the frontier will not render in GitHub's UI:
| Missing | Fallback |
|---|---|
| Sub-issues | Put `Part of #<map>` on the first line of each question body, and a task list of the questions in the map body |
| Dependencies | Put `Blocked by: #12, #14` on its own line at the top of the question body |
Prefer the native mechanisms every time they work. The whole reason to pay GitHub's costs is that the human sees the frontier in the UI without opening the map.
---
## The frontier query (Work Step 3)
The frontier is every question that is **open, unblocked, and unassigned**. Lowest position in sub-issue order wins — the same traversal `lowest NN first` gives locally.
```bash
# 1. The map's children, in order, with the state you need to filter on.
gh api repos/<owner>/<repo>/issues/<map-number>/sub_issues \
--jq '.[] | {number, title, state, assignee: .assignee.login,
blocked: .issue_dependencies_summary.blocked_by,
labels: [.labels[].name]}'
```
Then, in order:
1. Drop anything `state: closed` — that is resolved or out of scope.
2. Drop anything with an `assignee` — claimed by another session.
3. Drop anything still blocked.
4. The first survivor is the next question.
For step 3, `issue_dependencies_summary.blocked_by` counts **open** blockers, which is exactly the live gate — a blocker that closes drops the count without anyone editing anything. **Treat it as a fast pre-filter, not the authority.** If it comes back `null` or absent from the list response, or you need to *name* the blockers for the trail footer or a fully-blocked report, ask the endpoint that owns the answer:
```bash
gh api repos/<owner>/<repo>/issues/<n>/dependencies/blocked_by --jq '.[] | {number, title, state, reason: .state_reason}'
```
A question is unblocked when every blocker listed there is closed. That call is also the only way to see the next trap:
**A blocker closed as `not planned` is out of scope and will never resolve.** Its dependent is not merely blocked, it is *stranded* — the same trap as locally. Re-frame the dependent's body to drop the dependency, cut the edge, or rule it out too. Never leave it sitting: the summary count cannot tell you the difference, so this check is on you.
```bash
# Cut a dependency edge. The blocker's database id goes in the PATH here, not the body.
gh api --method DELETE \
repos/<owner>/<repo>/issues/<blocked-number>/dependencies/blocked_by/<blocker-db-id>
```
---
## Claim (Work Step 4)
```bash
gh issue edit <n> --add-assignee "@me"
```
**The session's first write, before any work.** The assignee *is* the claim — GitHub timestamps it for you, so there is no `Claimed:` line to maintain. An open, unassigned question is unclaimed; that is the whole protocol.
Unlike the local backend, other people may genuinely be working this map at the same time. Re-read the issue immediately after assigning; if someone else's login is on it, you lost the race — release yours and take the next frontier item.
---
## Resolve (Work Step 7)
Three writes, in this order. The order matters: the answer must exist before the issue closes, or a crash between them leaves a closed question with no decision in it.
```bash
# 1. The answer, as a comment. Same anatomy as a local ## Answer:
# decision, rejected alternatives with reasons, consequences, one-line **Gist:**.
gh issue comment <n> --body-file - <<'EOF'
## Answer
**Decision.** CSV, UTF-8 with a byte-order mark, RFC 4180 quoting, one header row.
Alongside it a sidecar `.sha256` manifest.
**Rejected.**
- *JSONL* — trivially streamable, but every named recipient opens these in Excel.
- *XLSX* — fixes Excel encoding, but adds a library and makes byte-level verification harder.
**Consequences.**
- Do not reuse the billing export's CSV writer; it concatenates strings with no escaping.
- Makes [Delivery channel](https://github.com/o/r/issues/45) sharper — the artifact is
self-verifying, so a short-lived link no longer weakens the integrity story.
**Gist:** CSV with a UTF-8 BOM and RFC 4180 quoting, plus a sidecar SHA-256 manifest;
JSONL rejected because recipients open these in Excel.
EOF
# 2. Close as completed.
gh issue close <n> --reason completed
# 3. Append the gist to the map's Decisions so far (read body, edit, write back).
gh issue view <map-number> --json body --jq .body > /tmp/map.md
# ...append: - [Export format](<issue-url>) — <gist> *(YYYY-MM-DD)*
gh issue edit <map-number> --body-file /tmp/map.md
```
Then bump `**Updated:**` and re-score `**Confidence:**` in the same map edit.
**The close date IS the `Resolved:` date.** There is no line to set and none to backfill — `closedAt` is written by GitHub the moment step 2 runs, and a brief filters on it. Stamp the same date in italics on the `## Decisions so far` line so the map reads as a dated record on its own.
**Never edit the question body to hold the answer.** The body is the question as asked; the comment is the answer. Editing the body rewrites history and destroys the record of what was actually put to the human — which is half of why the answer is defensible three weeks later.
### Editing the map body safely
Every map mutation is read-modify-write on a body other sessions may be editing concurrently. Read it fresh immediately before the edit, apply your change to *that* text, and write it straight back. Never edit from a copy you read at the top of the session — you will silently revert whatever landed in between.
---
## Ruling a question out of scope (Work Step 8)
```bash
gh issue edit <n> --add-label "pathfinder:out-of-scope"
gh issue close <n> --reason "not planned"
```
Then one line under the map's `## Out of scope`, giving the name as a link plus the reason. **No `## Answer` comment** — there is no decision here, only a scope boundary. Add one comment saying why it is out, so the closed issue explains itself.
`not planned` versus `completed` is the load-bearing distinction: it is how a later session tells a decision that was made from a question that was ruled off the route, and it is what GitHub's UI shows at a glance. Getting it backwards puts a scope boundary into the Provenance table of the plan.
---
## Reconcile (Work Step 2)
Much shorter here — the tracker holds the state, so there is no checklist to rebuild. Four checks:
| Check | Symptom | Repair |
|---|---|---|
| Crashed mid-answer | Open, assigned, and an `## Answer` comment already exists | The comment wins. Close as completed, add the gist to Decisions so far, say so. |
| Stale claim | Open, assigned, no `## Answer`, and the assignee is you from a dead session | Unassign, say so, put it back on the frontier. **If it is someone else's login, leave it** — that is a live session, not a crash. |
| Interrupted research | An open `research` question with no `## Evidence` comment ending in `**Research complete:**` | Re-fire its subagent (`resolve.md` → Interrupted research), say so in one line. |
| Index drift | A closed, completed question with no line under `## Decisions so far` | Read its `## Answer` comment, append the gist. |
Then re-read `## Not yet specified` in full — that part is identical to local mode, and the bullet left behind after its question exists is just as corrosive here.
Local mode's `Resolved:` backfill has no counterpart here: `closedAt` cannot be absent on a closed issue. If a `## Decisions so far` line is missing its date, take the date from `closedAt` during the index-drift repair above.
---
## What stays on local disk
Three things never move to the tracker, whatever the backend:
| Artifact | Where | Why |
|---|---|---|
| Runnable sketches | `specs/<idea>/pathfinder/sketch-<issue-number>/` | Throwaway code has no business in an issue, and the quarantine rule (never in the project's own source tree) is unchanged. Link the path from the issue and note the reader needs the repo checked out. |
| `PLAN-DRAFT-<YYYYMMDD>.md` | `specs/<idea>/` | `/plan2code-1-plan` reads a **file**. This is a hard downstream contract — see Handoff. |
| Anything secret | Nowhere | Credentials, tokens, customer data. A `legwork` checklist says *where* a credential lives, never what it is — and on a public tracker that rule stops being a convention and starts being an incident. |
Sketch directories are named for the issue number rather than a local `NN`, so `sketch-42` belongs to the question at `#42`. Same rules otherwise: throwaway, one command to run, never merged.
Research subagents work the same way with one substitution: the brief carries the **issue URL** instead of a file path, and the instruction is to post findings as a comment opening `## Evidence` via `gh issue comment` — and to decide nothing. The comment ends with the `**Research complete:** <YYYY-MM-DD>` line, and the firing session waits for every subagent before Session End, exactly as on local disk. `## Answer` and the close are still written by the session that fired it.
---
## The trail footer
Identical in shape; the inputs come from the query instead of the checklist.
- **Heading** — `🧭 <idea> · <Status> · <closed>/<total> cleared`, where `<total>` excludes anything labelled `pathfinder:out-of-scope`.
- **Glyph order** — sub-issue order, the same order Chart Step 6 created them in.
- **Glyphs** — `●` closed as completed · `◉` open and assigned to you · `○` open, unassigned, unblocked · `⊘` open with `blocked_by > 0` · `⊝` closed as not planned.
- **Named legend** — names, never `#numbers`. `(blocked:<name>)` names the blocker rather than numbering it, since there is no stable `NN` to point at.
- **Confidence** — the plain-English line, from the map body's `**Confidence:**`.
Form A's resume command changes, because there is no local path to resume from:
```
NEXT STEP · start a new conversation and run:
`/plan2code-0-pathfinder https://github.com/<owner>/<repo>/issues/<map-number>`
```
Form B is unchanged — a turn that asks the human something still says `WAITING ON YOU`, still names the outstanding probes, and still emits no resume command.
Form C — the fork-menu after a recorded decision — is unchanged in shape, with two substitutions: the takeable options come from the frontier query rather than the checklist rows, and the `OR START FRESH` command carries the map issue URL. The `RECOMMENDED` marker and the ~3-decision checkpoint work exactly as `trail.md` describes.
With no script to draw them, the labels are exact: Form A `NEXT STEP · start a new conversation and run:`, Form B `WAITING ON YOU · answer here, in this conversation:`, Form C `OR START FRESH · new conversation, paste:`. The Form C question tool uses the header `Next step` and the question "Continue here, or start fresh? Everything is saved." (add "I recommend starting fresh, to keep me sharp." at the ~3-decision checkpoint). Its options are up to three takeable question names plus `Start fresh`, one marked `RECOMMENDED`.
---
## BRIEF mode
The report is identical either way, and the file still lands on **local disk** at `specs/<idea>/pathfinder/briefs/brief-<YYYYMMDD>.md` — a brief is a plain-English artifact for people, not tracker state, and writing it as an issue comment would bury it. Three substitutions against `brief.md`:
| `brief.md` says | Here |
|---|---|
| Filter resolved questions by `Resolved:` in range | Filter the map's sub-issues closed as completed by `closedAt` in range |
| Reconcile backfills absent `Resolved:` dates | Nothing to backfill; run the four reconcile checks above instead |
| Open / blocked / out-of-scope sections read the checklist | They read the frontier query — unassigned-unblocked, `blocked_by > 0`, and `pathfinder:out-of-scope` respectively |
`## Not yet specified`, `## Destination`, and `**Confidence:**` come from the map issue body, same as always. The no-machinery rule tightens rather than relaxes: no `#numbers`, no label names, no `gh` commands in the file. The reader may never have opened the tracker either.
---
## Handoff (The Clearing Gate)
The gate's four dimensions, the 18/25 bar, the hard caps, and the honesty rules are unchanged. Only the preflight and the plumbing differ.
**Preflight, GitHub form:**
| # | Check |
|---|---|
| 1 | Reconcile pass run (above) |
| 2 | Zero open sub-issues — `gh api .../sub_issues --jq '[.[] \| select(.state=="open")] \| length'` returns `0` |
| 3 | `## Not yet specified` in the map body is empty |
| 4 | Every completed question has an `## Answer` comment carrying a `**Gist:**` |
| 5 | `ls specs/<idea>/` shows no existing `PLAN-DRAFT-*.md` |
| 6 | No `## Answer` defers a choice to "whoever implements this" |
**The draft is written to local disk**, at `specs/<idea>/PLAN-DRAFT-<YYYYMMDD>.md`, with the byte-exact status line `**Status:** Phase 3 Complete - Resume at Phase 4`. This is not a preference. `/plan2code-1-plan` discovers its input with `ls specs/`; it has no notion of an issue tracker, and a draft that exists only as an issue is a draft the rest of plan2code cannot see. Create `specs/<idea>/` if charting never needed it.
Three substitutions inside the template:
| Local | GitHub |
|---|---|
| `**Planning record:** specs/<idea>/pathfinder/map.md` | `**Planning record:** <map issue URL>` |
| `[Export format](./pathfinder/questions/01-export-format.md)` | `[Export format](https://github.com/o/r/issues/42)` |
| `## Out of scope` copied line for line, only the link prefix changing | Copied line for line, links already absolute — **nothing changes at all** |
Everything else — the mapping table, Section 5 left empty, the two load-bearing `**Next:**` bullets, no scrapable confidence numbers — is unchanged.
**Freeze the map:**
1. Set `**Status:** Cleared` in the map issue body.
2. Bump `**Updated:**`.
3. Add `**Plan:** specs/<idea>/PLAN-DRAFT-<YYYYMMDD>.md` under the status line.
4. `gh issue close <map-number> --reason completed`.
5. **Close nothing else, delete nothing, edit no answers.** Every question issue stays exactly as it is — it is the rationale record behind the plan.
Closing the map is the one addition over local mode, and it earns its place: `gh issue list --label pathfinder:map --state open` then reads as *the efforts still being charted*, which is the question a person scanning the tracker actually has.
A later session that opens a map issue reading `Status: Cleared` must not resume it. Point at the PLAN-DRAFT and `/plan2code-1-plan`, and stop. A redrawn destination is a fresh effort with a fresh map issue.
---
## Failure modes
| Symptom | Cause | Fix |
|---|---|---|
| `Not Found` from a `sub_issues` or `dependencies` POST | An issue *number* was passed where a database *id* is required | `gh api repos/o/r/issues/<n> --jq .id`, retry with that |
| The wrong issue got attached | A number from another repo happened to be a valid id | Detach, re-fetch `.id` from the right repo, re-attach |
| Frontier is empty but open questions remain | Every one is blocked, or every one is assigned | Name the chain and stop — or, if the assignees are stale claims from your own dead sessions, reconcile first |
| A blocked question never unblocks | Its blocker was closed as `not planned` | Stranded. Re-frame to drop the dependency and cut the edge, or rule it out too |
| Two sessions resolved the same question | The claim was written after the work, not before | Claim is the *first* write. Merge the two answers into one comment, keep one close |
| A map edit lost someone's line | The body was edited from a stale copy read earlier in the session | Re-read the body immediately before every map write |
| `/plan2code-1-plan` finds nothing to resume | The PLAN-DRAFT was posted as an issue instead of written to `specs/<idea>/` | Write the file. The draft is always local |
| Secrets in the tracker | A `legwork` answer pasted a credential | Rotate the credential first, then delete the comment. Editing it is not enough — GitHub keeps the edit history |
@@ -0,0 +1,388 @@
# Grilling Playbook
> Part of plan2code-0-pathfinder — loaded once per session, used by Chart and Work alike.
Grilling is how a `grill · HITL` question resolves, how Chart Step 2 names the destination, and how Chart Step 4 maps the frontier. It is also the fallback for any question whose type gives you no better route. The output of a grill is a decision in the human's own words — never a decision you made on their behalf.
**Probe ≠ question file.** A *probe* is one turn of the interrogation; a *question file* is one `questions/NN-<slug>.md` on the map. Batching applies to probes only. **One question file at a time still holds** — a batch of three probes resolves ONE file; a session moves to the next file only through the fork-menu between decisions (resolve.md §Cadence), never by interleaving two grills.
**Grilling a room.** When several people answer — a Product + Engineering session — the mechanics are unchanged, plus three rules: address probes to the room, not a person, unless one person owns the call; the recap must capture the decision in the room's words, and whose call broke any tie is worth a clause in the `## Answer`; and a contested pick is NOT a decision — when Product and Engineering disagree, record both positions under `## Evidence`, leave the question `open`, and move on (or offer a `sketch` — reacting to something concrete resolves more standoffs than argument). Never average two positions into an answer nobody actually gave.
## The interview protocol
**Grilling is batched.** Put up to **three independent probes** to the human per turn, then wait. Never more than three, and never two probes where one's wording depends on the other's answer.
The old rule here was one probe per turn. It was safe and it was unusably slow: a charted map carries a dozen open questions, and a decision that costs one round trip per probe is a decision the human abandons half-finished. Batching is the default now; the discipline moved from *ask one* to *prove they are independent, then send three*.
| Rule | Why |
|---|---|
| Up to 3 probes per turn, never more | Past three the human skims, and a skimmed answer is worse than none. Three is a ceiling, not a quota — send two if only two are independent. |
| Only batch mutually independent probes | The independence test below. A probe whose wording or recommendation shifts based on another probe's answer waits for the next turn. |
| Use the structured question tool for every question when available | The built-in UI is the primary experience: selectable answers, visible skips, and `Other` for composition. Prose provides context, never the answer control. |
| Keep detail in descriptions, context, and `Other` | Batching buys round trips; it must never buy them by shrinking a decision to fit a picker. |
| Wait for the whole batch before sending the next | Their answers reshape what comes next. Pre-writing turn 2 wastes it. |
| Recommend an answer with every probe | A bare question makes the human do all the work. A recommendation gives them something to push against, which is faster and sharper. |
| Write it in plain English; keep the technical word only where that word IS the decision | A probe the human has to decode is a probe they answer approximately. See *Say it in plain English*. |
| Track the batch; re-ask what came back unanswered | This is what buys the batch. Dropped probes going unnoticed was the entire case for asking one at a time. |
| Walk one branch of the decision tree at a time | Batch across the branch's width, never down its depth. Do not ask about export scheduling before you know whether exports exist. |
| Do not act until they confirm shared understanding | Recap, get the confirmation, then write. No `## Answer` before the confirmation. |
| Never write implementation code during a grill | Grilling produces decisions. If you feel the pull to build, you have reached the edge of the map — say so and hand off. |
Shape of a single probe, batched or not:
```
Question — one decision, stated so it can be answered in a sentence.
Why it — one line: what it blocks, what breaks if it goes the other way.
matters
Recommend — your pick, with the reason. One line.
Options — the genuine alternatives, if there are more than two.
```
Worked probe, from a grill on `[Export format](./questions/04-export-format.md)`:
> **Q:** When a custodian exports a conversation that includes a 40 MB video attachment, does the export bundle the file or link to it?
> **Why it matters:** Bundling sets the size ceiling on an export job and decides whether exports can stream; linking makes the export useless once retention expires the blob.
> **Recommendation:** Bundle, with a per-job size cap of 2 GB and an automatic split into part files above that — reviewers open exports offline in tools that cannot follow links.
> **Alternatives:** Link-only (smaller, breaks offline); hybrid by MIME type (two code paths, two failure modes).
Batching does not shrink a probe. Three probes means three of these, each with its own why-it-matters and its own recommendation. Three bare questions in a list is not a batch, it is a form to fill in.
### Say it in plain English
Every probe gets read once, by a busy human, in a terminal. Write it the way you would say it out loud to a colleague who knows the product but has never read this skill. Plain is not vague — the two failures are opposite and both cost you the decision: woolly wording gets a woolly answer, dense wording gets a guessed one.
**Default to everyday words.** Short sentences, concrete nouns. "What should happen when the export is too big to email?" beats "what are the failure semantics of the artifact delivery path under a size-limit violation?" Same decision, one of them answerable on the first read.
**Keep the technical term where that term IS the decision.** A format name, a real file path, a column name, a version number, a limit, a `## Glossary` term the map already settled — those are load-bearing, and softening them makes the probe unanswerable. `.eml`-in-a-ZIP versus NDJSON *is* the choice; "normal email files versus one big machine-readable stream" is the gloss you put beside it, never the replacement for it.
> The test: would swapping the term for a plain phrase lose information the human needs in order to choose? Lose information — keep the term and gloss it. Lose nothing — cut it.
**Gloss an unavoidable term once, inline, then use it freely:** "…stored under Object Lock (S3's write-once mode — once it is set, even we cannot delete early)." Once per session, not once per probe. Re-explaining a term to the person who owns the system is its own insult.
**Never put Pathfinder's machinery in front of the human.** They are deciding something about their product; the vocabulary below is internal bookkeeping and buys them nothing:
| Do not say | Say |
|---|---|
| "this is a `grill · HITL`" | "this one is yours to call" |
| "the frontier holds two takeable questions" | "two things we can decide right now" |
| "graduating this out of the fog" | "this is sharp enough to write down as a real question now" |
| "the sacrificial boundary" | "which of these features are we leaving out this time?" |
| "shall I set `Locked: yes`?" | "worth recording why we picked this, so nobody re-opens it in six months?" |
| "Q3 is blocked by 02" | "the export format question has to land before this one" |
| "the UI cannot handle this question" | nothing — put context before the tool and preserve `Other` |
**Refer to questions by name, never by number** — "[Export format](./questions/04-export-format.md)", not "04". The number means something to the file system and to nobody else.
**No metaphor where a fact fits.** Maps, fog, and trails belong in the footer and the mascot. Inside a probe they cost a translation step: "three things here are still undecided" beats "the fog is thick in this quarter of the map."
**On a what-to-drop question, name the thing dropped.** Whenever a question asks what gets dropped, left out or cut, every option names the thing being dropped, never a rule to keep. Otherwise ticking an option reads as a double negative, and the person has to work out whether a tick means yes or no. Bad: "☐ Exports stay manual" · Good: "☐ Scheduled exports (Leaving this out = someone runs each export by hand)"
The same discipline covers everything else the human reads — the recap turn in *Landing the grill*, the option labels and descriptions in the structured tool, the sketch probes, and the HITL checklists in the resolution playbook. Plain in the question, precise in the term that carries the decision.
### The independence test
A probe may join the current batch only if **all three** hold:
| Test | Fails when |
|---|---|
| Its wording would not change under any answer to another probe in the batch | "How do we name the part files?" reads differently if the format turns out to be a single stream |
| Its recommendation would not change either | You would recommend a 2 GB cap under ZIP parts and no cap under NDJSON |
| It does not presuppose another probe's answer | "How often do scheduled exports run?" assumes scheduled exports exist |
In doubt, hold it back. A held probe costs one extra round trip. A dependent probe sent early costs a wrong answer recorded as a decision, and you will not find out until the plan contradicts itself.
Independent probes are usually the ones that came from **different areas** — data, interface, security, operations, testing. Dependent probes are usually consecutive steps down one thread.
### Delivering a batch: structured UI first
**Use the environment's structured question tool for every user-facing question when it is available.** This includes destination probes, frontier probes, claimed-question grills, recap confirmations, lock offers, sketch reactions, intent gates, and fork-menus. The built-in UI is faster to scan, preserves visible skips, and always provides an `Other` path for composed answers. Numbered prose Q blocks exist only as the fallback when the tool is unavailable.
One question object per probe, up to three per call. Never split a batch between a tool popup and prose — the prose half gets lost when the human answers in the UI.
**Composed answers still use the tool.** Do not invent a false-complete option set merely to make a picker. Offer the genuine candidates you know; the automatic `Other` option is where the human composes a boundary, rationale, hybrid, old-incident detail, or alternative you could not know. When no genuine candidate exists, offer useful control choices such as `Use recommendation` and `Defer for now`; the human writes the real answer through `Other`. A click is still followed by the normal recap-confirmation turn, which captures reasoning in the human's words.
**Artifacts do not force prose questions.** Show the state table, fake request/response, ASCII UI, or worked example in the response, then immediately invoke the structured tool for the pointed reaction probes. The artifact is prose; the questions are UI controls. Never ask "thoughts?" outside the tool.
**Never reshape the decision to fit the tool.** Descriptions carry trade-offs; `Other` carries composition. Split bundled decisions into separate probes and rename colliding short headers. Do not cut a real alternative or hide complexity. If the UI cannot represent the question faithfully even with descriptions + `Other`, state the full context immediately before the tool and keep the actual answer control in the tool.
The turn closes with the waiting footer (Form B in `trail.md`): it names outstanding probes and carries no resume command. The recap confirmation is also asked through the tool and remains Form B.
#### The structured tool
The primary channel. One question object per probe, up to three in a single call:
- **Header** — the decision in two or three plain words (`Export format`, `Size cap`). Not a type, not a marker, not a number.
- **Question** — the probe, with its why-it-matters. This is prose and it is not rationed; the same sentences you would have written in a Q block go here.
- **Options** — the genuine alternatives, each described by its trade-off, with the recommended one named as such in its description. Two to four; the free-text escape hatch covers the rest. Label plainly, then let the description carry the precise term: `One file per message` labelling the `.eml`-in-a-ZIP option, with `.eml` named in the description.
A short *label* is not a short *decision*. The label is a handle — `Fixed tick count` — and the description carries the trade-off that makes it choosable. A label bundling several independent answers is a probe that wants splitting.
**A click is a decision, not a sentence.** The HITL rule wants an `## Answer` traceable to something the human actually said, and a selected option label is thin evidence on its own. What makes tool-delivered answers legitimate is the recap turn in *Landing the grill* — you play the choices back in the tool's question text and they confirm or correct in their own words. Never skip the recap on the grounds that the tool already captured the answer; the tool captured the *pick*, and the recap captures the *agreement*.
**If a reply comes back thinner than the decision** — a bare click on something you now realise carries weight — do not paper over it. Fold the why into the recap turn as one more probe before writing the `## Answer`.
#### Numbered Q blocks
Fallback ONLY when the structured question tool does not exist or its call fails. Never choose prose merely because the answer is detailed, composed, architectural, `Locked: yes`, or attached to an artifact — descriptions + `Other` + context immediately before the tool cover those cases.
**Copy this shape exactly.** The blank lines are load-bearing, not decoration:
````markdown
Three independent decisions are open. Answer in any order, skip any you want to punt — "Q2: a, c; Q3: the hybrid" is a perfectly good reply.
---
**Q1 — Export format**
When a custodian exports a year of a channel, what do they get back?
*Why it matters:* fixes the size ceiling, decides whether exports can stream, and determines whether the review vendors can ingest without a conversion step.
*Recommendation:* **(a)** — both vendors named in Codebase context read `.eml` natively.
- **a)** ZIP of one `.eml` per message, plus `manifest.csv`
- **b)** A single NDJSON stream — compact and streamable, but nobody downstream parses it
- **c)** PST — what Legal asked for by name, but single-writer with a ~50 GB ceiling
---
**Q2 — Leave for later**
Which of these features are we leaving out this time? Tick any number.
*Why it matters:* the ones you tick become the first things written down as out of scope, and every later "is that in or out?" call is measured against them. A destination nobody has excluded anything from has not been thought about.
*Recommendation:* **(a)** and **(c)** — neither is needed for the first review hand-off.
- **a)** Scheduled exports (Leaving this out = someone runs each export by hand)
- **b)** Exports of deleted messages (Leaving this out = anything already purged stays out of the file)
- **c)** Export to a cloud bucket (Leaving this out = the file downloads to the custodian's machine only)
- **d)** Redaction inside the export (Leaving this out = the review vendor redacts after ingest)
````
Number them, keep the numbers stable across turns and sessions, and say out loud that partial answers are welcome — the invitation is what makes the skip visible instead of silent.
#### Formatting rules for a Q block
A batch is only worth sending if the human can read it. These are mechanical, and getting them wrong turns three careful probes into one unreadable paragraph:
| Rule | Why |
|---|---|
| **A blank line between every element** — the `**Qn — Name**` line, the question, *Why it matters*, *Recommendation*, and the option list | Markdown joins consecutive non-blank lines into a single paragraph. Without blank lines the entire batch renders as a wall of text and the human skims it, which is the failure the three-probe cap exists to prevent. |
| **Never hard-wrap a sentence across source lines** | The wrap is invisible to the renderer, so it buys nothing and costs you the paragraph break. Write each sentence as one logical line however long it is; the terminal wraps it. |
| **Options are a bullet list, one option per bullet** — `- **a)** …` | Indented continuation lines are the specific thing that collapses: under four spaces the indent is stripped, at four or more it becomes a code block. A bullet list survives every renderer and keeps the options scannable. |
| **`---` between probes** | Three probes run together is one block of text. The rule gives the eye a stop and makes "answer Q2 and Q3" easy to aim at. |
| **The question itself gets its own line, not a run-on with the heading** | `**Q1 — Export format.** When a custodian…` buries the decision inside a paragraph. Name it, break, then ask it. |
| **Never use spaces to convey structure** | Whatever hierarchy you indent by hand disappears on render. Structure comes from blank lines, bullets, and bold — nothing else. |
For the recap in *Landing the grill*, summarize each decision in the structured tool's question text and ask the human to confirm or correct it; use `Other` for corrections.
### When answers come back partial
Assume they will. The human answers two and drops one, and the dropped one is often the hardest and most valuable.
1. **Diff what came back against what you sent.** Skipped, answered with "Other: skip", or silently omitted all count as unanswered.
2. **Lead the next tool call with the unanswered probes**, using their original headers and full wording. Not "you missed Export scope" — the whole probe again, with its recommendation, because they have lost the context by now. Unanswered probes come *before* any new probe, and they count against the cap of three.
3. **Skipped twice, stop pushing.** Record it under `## Evidence` as an open probe with your recommendation verbatim, then either narrow it into something answerable or spin it out — a fresh question file if you can phrase it sharply, a `## Not yet specified` line if you cannot.
4. **Never promote your own recommendation into `## Answer`.** A probe the human declined twice is unanswered, not decided. Writing it up as decided is self-answering a HITL question, which breaks the skill.
Never let a dropped probe fall off the end of the session unrecorded.
## Facts you look up, decisions you ask
This is the single rule that keeps a grill from feeling like an interrogation. If a **fact** can be found by exploring the environment — filesystem, codebase, tools, docs, config, git history, the map's own resolved questions — go find it. The **decisions** are the human's; put each one to them and wait.
| You look it up | You ask |
|---|---|
| Which Postgres version the app runs against | Whether the export index is allowed to add a new table |
| Whether `ExportJob` already has a `status` column | What states that column is allowed to hold |
| How the current retention sweep is scheduled | Whether exports must survive a retention sweep |
| Whether the repo uses Vitest or Jest | Whether these paths get unit tests, integration tests, or neither |
| What the S3 bucket lifecycle rule is today | Whether we are allowed to change it |
| What `AGENTS.md` says about naming conventions | Anything `AGENTS.md` does not already answer |
Look first, then ask. A question that opens with "I checked `src/export/job.ts` — it already has a `status` enum with `queued | running | failed`. Does a partial success need a fourth state?" is worth three of "how should export status work?"
**Never ask what `AGENTS.md`, the map's `## Ground rules`, or a resolved question already answers.** Re-asking a settled decision reopens it by accident and costs you the human's trust for the rest of the session.
**When the lookup is expensive**, that is not a grill — it is a `research · AFK` or `legwork` question. Say so, note it, and keep grilling the decisions you can still put to the human.
## The HITL rule, stated hard
An agent that answers its own grill has broken the skill.
A `grill · HITL` question resolves **only** through live exchange with the human. Not from the codebase, not from a plausible default, not from "the obvious industry standard," not from what you would have picked. The whole value of the question is that a human with context you do not have chose one branch over another.
Signs you are about to break it:
- You wrote a recommendation and then wrote the `## Answer` without a reply in between.
- You wrote "assuming the user would want X" anywhere.
- You resolved a question in a session where the human said nothing but "go".
- The `## Answer` contains no sentence traceable to something the human actually said.
**When the human goes unreachable mid-grill:**
1. The claim **stays**. `State: claimed` and `Claimed:` are left exactly as they are.
2. Append to `## Evidence` — never `## Answer` — the exchange so far: the probes already answered, **every probe of the last batch still outstanding**, and your recommendation for each, verbatim.
3. End the session. Report the question by name and say it is mid-grill and waiting on the human. This one IS a session end, so the Trail Footer takes **Form A** — the pathed resume command, not `WAITING ON YOU`.
4. Invent nothing. No provisional answer, no "pending confirmation" answer, no default recorded as a decision.
The next session picks the claim back up and re-sends the outstanding batch. If the session was truly abandoned rather than paused, Work Step 2 reconciliation resets it to `open` on its own — that is its job, not yours.
## The four disciplines
Run all four continuously during any grill. They are not stages; they fire whenever the trigger appears in what the human just said.
### Challenge against the glossary
When a term conflicts with the map's `## Glossary`, call it out immediately, mid-sentence if necessary.
> "The glossary defines **Export** as a completed archive file delivered to a custodian. You just used it for the background job that builds one. Which do you mean — or do we need a second term?"
### Sharpen fuzzy or overloaded language
When a term is vague or carries two meanings, propose a precise canonical term and get a ruling.
> "You keep saying 'account.' Sometimes you mean the organization paying us, sometimes the individual login. Those are a **Customer** and a **User** and they have different retention rules. Which one owns the export quota?"
### Stress-test relationships with concrete invented scenarios
Do not ask abstractly whether a relationship holds. Invent a specific scenario that probes the edge and force a precise boundary.
> "A custodian leaves the company on the 3rd. Their retention policy expires their messages on the 5th. Legal opens a hold on the 4th. On the 6th, does the export still contain those messages?"
Invent the numbers, names, and dates. Vague scenarios get vague answers.
### Cross-reference claims against the actual code
When the human states how something works, check whether the code agrees, and surface contradictions instead of quietly picking a side.
> "You said retention deletes rows. `RetentionSweep.run()` sets `deleted_at` and leaves the row in place — a soft delete. Which is the behavior we are designing against?"
A contradiction is a finding, not an embarrassment. Surface it in the same turn you found it.
## The glossary
Domain modeling would write a `CONTEXT.md`. Pathfinder does **not** — plan2code owns `AGENTS.md`, and a competing root glossary file would collide with it. The map's `## Glossary` section is the one place resolved terms live.
Entry format — one row in the map's `## Glossary` table: the term, a one-or-two-sentence definition, and the rejected synonyms in the Avoid column:
```
| Term | Meaning here | Avoid |
|---|---|---|
| Export | A completed, immutable archive file delivered to a custodian. Always the artifact, never the process that produces it. | download, extract, dump |
| Export Job | The background unit of work that produces an Export. Has states; an Export does not. | export run |
| Custodian | The person whose communications an Export contains. Not necessarily the person who requested it. | user, owner, subject |
```
Rules:
| Rule | Detail |
|---|---|
| Update inline, never defer | The moment a term is resolved, write it into `## Glossary` and save. A term you meant to add at the end of the session is a term you lost. |
| Be opinionated | When several words compete, pick one and put the rest in the Avoid column. A glossary that lists synonyms as equals has decided nothing. |
| Keep definitions tight | One or two sentences. Define what it **is**, not what it does. |
| Only project-specific terms | "Retention Policy" belongs. "Timeout", "retry", "DTO" do not, however heavily the project uses them. |
| It is a glossary and nothing else | No implementation details, no open questions, no scratch notes, no decisions. Decisions live in question files. |
| Group under sub-bullets only when clusters emerge | A flat list is fine for one cohesive area. |
Every term you resolve is a term no later session re-litigates. That is the whole return on the discipline.
## The Lock test
Domain modeling would offer an ADR here. Pathfinder sets `Locked: yes` on the question instead — the decision record and the question are the same file.
Offer `Locked: yes` **only** when all three hold:
| Test | Meaning | Fails when |
|---|---|---|
| Hard to reverse | Changing your mind later costs real time or migration | You could flip it in an afternoon |
| Surprising without context | A future reader will ask "why did they do it this way?" | It is the obvious choice anyone would make |
| A real trade-off | There were genuine alternatives and one was picked for specific reasons | There was only ever one option |
If any one is missing, skip it. An easy-to-reverse decision will just get reversed. An unsurprising one raises no questions to answer. One with no alternative records nothing beyond "we did the obvious thing."
**What earns a lock:** architectural shape ("the export index is a materialized view, not a table"); integration patterns between components ("retention and export communicate by event, never by direct call"); technology choices carrying lock-in (database, message bus, auth provider — not every library, just the ones that would take a quarter to swap); boundary and scope decisions, including the explicit no-s; deliberate deviations from the obvious path ("hand-written SQL here, not the ORM, because the ORM cannot express the retention join"); constraints invisible in the code ("no cross-region replication — the data residency contract forbids it"); and non-obvious rejections ("we considered and rejected GraphQL, for reasons someone will otherwise re-propose in six months").
**What `Locked: yes` obliges the `## Answer` to contain:**
1. The decision itself, in one or two sentences.
2. **Every alternative genuinely considered, and why each was rejected.** This is the part that makes a lock worth having. "We rejected X" with no reason is not a lock.
3. The consequences a later reader would not guess.
4. The one-line `**Gist:**` that Work Step 7 requires, same as any answer.
Offer it, do not impose it: *"This one looks hard to reverse and the reasoning will not be obvious in six months. Lock it?"* The human decides.
**Where locks go at handoff:** the handoff playbook lifts every `Locked: yes` question into the PLAN-DRAFT's **Architecture** section, and their rejected alternatives and standing constraints into **Assumptions**. Unlocked answers still inform the draft, but locks are the ones that survive verbatim into planning. Grill them harder for that reason.
## The one grill every map must resolve, and the lens applied to all of them
### 1. Testing posture — `grill · HITL`
Chart Step 6 requires this question. It exists because `/plan2code-1-plan` Phase 1 asks for exactly three things and stalls without them: testing types, whether tests run after each phase, and the coverage target. A map that clears without answering them hands the human a plan session that immediately re-asks.
The first three probes below pass the independence test against each other — none reads differently under another's answer — so **send all three as one structured-tool batch**. Every probe has named alternatives and fixed literals; `Other` remains available.
| Probe | Recommend by default |
|---|---|
| Which types are in scope — unit, integration, E2E, or none? | Unit plus integration; E2E only where a real browser or real broker is the only honest test |
| Does the suite run after each implementation phase, or once at the end? | `Run after each phase` — work that cannot be verified when it lands cannot be signed off |
| Coverage target: critical paths, moderate (~60-80%), or comprehensive (>80%)? | Critical paths, named explicitly, rather than a percentage nobody defends |
| What is deliberately not tested, and why? | **Fourth probe, and it does not ride in the batch.** It has no option set — the explicit no-s have to be composed — and it reads differently once the types are settled. Fold it into the recap turn, where you are already waiting on them. |
| What is already there — runner, fixtures, CI wiring? | **Not a probe.** Look it up before you send the batch, and cite it in the probes above |
Record the answer in the exact literals `/plan2code-2-document` string-matches: the types; `Run after each phase` or `Dedicated phase only`; and `Critical paths`, `Moderate (~60-80%)`, or `Comprehensive (>80%)`. Not prose — a paraphrase matches no branch downstream.
### 2. Test seams and verifiability — a lens, not a question file
For each major decision on the map, ask the same question: **how will anyone know it works?** A decision nobody can verify is a decision that silently rots.
This is not a question of its own and never gets a file or an `NN`. Run it inside whatever grill is claimed; anything it surfaces that needs deciding separately becomes a new question at Work Step 8.
These six all interrogate one decision from different sides, so they batch cleanly — but they ride along inside the claimed grill rather than owning a turn. Fold the two or three that bite into the batch you were already sending; never spend a whole turn on all six.
| Probe | What it flushes out |
|---|---|
| What observable behavior changes if this decision is implemented correctly? | Decisions with no observable effect — usually a sign the question was about implementation, not design |
| What is the cheapest thing that fails when it breaks? | The seam. If the answer is "a customer complains," there is no seam yet |
| Can this be tested without a live third-party account, a real S3 bucket, a wall-clock sleep? | Untestable-by-construction designs, while they are still cheap to change |
| Where does the boundary go so a test can stand at it — an interface, a queue, an HTTP edge, a pure function? | The seam the plan will need to name |
| What does the failure look like in production — log line, metric, alert, dead-letter queue? | Verifiability after ship, not just in CI |
| If we get this wrong, how long before we find out? | Decisions that need a canary or a feature flag rather than a test |
If a decision survives all six with no answer, it is not ready to leave the map. Either re-frame the question, or add the seam as a constraint in the `## Answer` so the plan inherits it.
## Anti-patterns
| Failure mode | What it looks like | Fix |
|---|---|---|
| Batching **dependent** probes | "What format, what do we name the part files, and how big is a part?" | Only the first is independent. Send it; hold the other two — they are unanswerable until format lands. |
| Drip-feeding one probe at a time | Twelve open questions on the map, one probe per response, the human gives up on session four | Batch up to three independent probes. On a charted map the human's round trips are the scarce resource, not your token budget. |
| Losing a probe the human skipped | Sent three, got two back, moved on and never mentioned the third | Diff the batch. Lead the next turn with what came back empty, restated in full. |
| A batch of naked questions | Three one-liners with no recommendations and no why-it-matters | Every probe in a batch carries its own recommendation and its own stake. Otherwise you have offloaded the thinking, not the round trips. |
| Asking in prose when the tool exists | Numbered Q blocks or a loose "which one?" despite AskUserQuestion being available | Put every answer control in the structured UI. Use descriptions for trade-offs and `Other` for composition. Prose may provide context or an artifact immediately before the tool, never the question itself. |
| Flattening a gnarly probe into labels | An architectural decision reduced to three option labels with no useful descriptions | Keep the tool, not the flattening: put the real trade-off in each description, preserve `Other`, and recap the reasoning before recording. |
| Treating an artifact as an excuse for prose | An ASCII UI followed by "thoughts?" | Show the artifact, then invoke the structured tool with pointed reaction probes naming the rows/elements. |
| Retreating to prose over a fixable batch | Two probes collide on the short header, or one label bundles three answers | Rename the headers; split the bundled probe. Reshape the batch, never the decision. |
| Sending four probes because the tool accepts four | A fourth probe added to a clean batch of three because there was room in the call | The cap is three regardless of what the environment allows. The ceiling is the human's attention, not the tool's schema. |
| Grilling in Pathfinder's own vocabulary | "The frontier has one takeable `grill · HITL` — shall we graduate 02 out of the fog and lock it?" | Plain English. Markers, types, `NN` numbers, and fog are your bookkeeping; the human is deciding about their product. |
| Plain-washing the load-bearing term | "Do you want the friendly file or the compact one?" where the real choice is `.eml`-in-a-ZIP versus NDJSON | Plain wording, precise nouns. Name the formats and gloss them; a decision made on a euphemism cannot be written into `## Answer`. |
| Asking what the codebase already answers | "Do you use Postgres or MySQL?" | Look. Every avoidable question spends trust you need for the hard ones. |
| Accepting a vague answer and moving on | "Handle it sensibly" → recorded as the decision | Push once more, concretely: "Sensibly meaning we drop the attachment, or fail the whole job?" A vague answer is not an answer. |
| Leading the human to your preferred answer | "You'd want Postgres here, right?" | Recommend openly, then present the real alternatives with their real merits. A recommendation invites a fight; a leading question suppresses one. |
| Grilling past the decision into implementation | "Should the retry helper take a callback or return a promise?" | That is the plan's job, or the implementer's. Stop at the decision. The pull to keep going is the edge of the map. |
| Drifting off the claimed question | Claimed `[Export format](./questions/04-export-format.md)`, forty minutes later deep in auth | Name the drift out loud, capture the new thread as a fresh question or as a line in `## Not yet specified`, and return. One question _file_ at a time — switching happens at the menu, not mid-grill. |
| Self-answering a HITL question | An `## Answer` with no words the human said | Delete it. Reopen the question. See the HITL rule. |
| Recording the decision but not the rejections | "We chose event-driven." | Rejections are half the record — and mandatory when `Locked: yes`. Ask what else was on the table before you close. |
| Grilling a fact | "How long does the retention sweep take?" | If it is measurable, measure it — or make it a `research · AFK` question. Do not make the human guess at their own system. |
| Letting the glossary go stale | Three terms resolved, none written down | Write each one the moment it lands. Deferring loses them. |
| Closing without confirmation | Answer written straight after the last reply | Recap the whole chain of decisions, get the explicit confirm, then write. |
## Landing the grill
When the branch is walked out. **Steps 1-3 are ONE turn, not three** — recap, confirmation request, and lock offer go out together, because a lock offer sent after a separate confirmation costs a round trip to ask a yes/no the human could have answered alongside the recap.
1. Recap the decisions in order, in the human's own terms, using glossary vocabulary. Include anything a structured-tool reply left implicit, so the confirmation covers the reasoning and not just the picks.
2. Ask for confirmation through the structured question tool (`Confirm` / `Correct via Other`). Do not skip this — the recap is where the human catches the one thing you misheard, and where a clicked option becomes words they said.
3. Apply the Lock test as another question in the same tool call. Offer, do not impose.
4. Write `## Answer` per Work Step 7: the decision, what was rejected and why, consequences, and a one-line `**Gist:**`. Evidence, links, and transcript fragments go under `## Evidence`.
5. Anything the grill surfaced that belongs to a different question goes to the map — a fresh question if you can phrase it sharply, `## Not yet specified` if you cannot, `## Out of scope` if it sits past the destination.
@@ -0,0 +1,423 @@
# Handoff Playbook — Clearing the Map into a PLAN-DRAFT
Loaded at The Clearing Gate. Turns a cleared map into `specs/<idea>/PLAN-DRAFT-<YYYYMMDD>.md` that `/plan2code-1-plan` resumes from at Phase 4, then freezes `pathfinder/` as the rationale record.
**Backend note.** The scoring rubric, the hard caps, the honesty rules, the template, and the mapping table are the same either way — and the draft is written to local disk either way, because `/plan2code-1-plan` reads a file, not a tracker. On `**Backend:** github`, `github-issues.md` replaces only the preflight table, the question links (issue URLs, already absolute), and the freeze steps. `pathfinder.mjs` reads local files only, so on `github` do each script step below except `lint` by hand with `gh`, as `github-issues.md` describes.
Nothing here is creative. The gate is scored, the mapping is fixed, the template is literal. Follow it exactly or the resuming plan session silently loses work.
## Preflight — before scoring anything
Run `reconcile`, then `node "<S>/pathfinder.mjs" gate specs/<idea>/pathfinder`. Its `checks` cover the mechanical preflight, read from the files:
| `checks` id | Means | If it fails |
|---|---|---|
| `markers-match-files` | The map's markers are the files' (reconcile ran) | Run `reconcile`. Markers are derived, never authored. |
| `no-open-claimed-blocked` | Zero `[ ]`, `[/]`, `[!]` | Not cleared. Return to the frontier. |
| `fog-empty` | `## Not yet specified` is empty | Not cleared. Graduate the fog into questions, or admit it is out of scope. |
| `every-resolved-has-gist` | Every resolved file has a real `## Answer` with a `**Gist:**` | The file wins over the marker. Repair, then re-check. |
| `hard-caps` | None of the Requirements, Feasibility or Risk caps below holds (the Integration cap is yours to check) | That dimension cannot exceed 17: fix the record first. |
| `scores-at-least-18` | The map's `**Confidence:**` line has all four at 18 or better | Not cleared. Keep working the weak dimension. |
`existingPlanDraft` lists a `PLAN-DRAFT-*.md` already in `specs/<idea>/`: read it and update it in place, never add a second dated draft. `planDraftPath` is the file to write either way.
One preflight item stays yours: **the destination is reachable with nothing left to decide**: no `## Answer` defers a choice to "whoever implements this". If one does, the map is not cleared: name the open decision and graduate it into a question.
Score the four dimensions (below), write them to the map's `**Confidence:**` line, and rerun `gate` until `mechanicalPass` is true.
## The clearing-gate scoring rubric
Four dimensions, 0-25 each, scored against **written evidence in `questions/`** — never against your recollection of the conversation. The gate needs **every dimension at 18/25 or better**. There is no averaging: 25/25/25/14 fails.
### Band scale (applies to all four)
| Band | Meaning |
|---|---|
| 23-25 | Decided, written down, and consequences recorded. A developer could act on it without asking a question. |
| 18-22 | Decided and written down. Residual detail remains, but it is *specification* detail that Phase 5/6 settles — not a decision anyone still has to make. |
| 12-17 | A real decision is still open, or an answer exists with no evidence behind it. **Gate fails.** |
| 0-11 | The area was never charted. **Gate fails**, and the map was cleared prematurely. |
The 18-boundary is the honest line between *"needs designing"* and *"needs deciding"*. Pathfinder owns deciding. If someone still has to decide, you are not done.
### What each dimension scores
| Dimension | Scores | Raises it | Lowers it |
|---|---|---|---|
| **Requirements Clarity** | Are the requirements unambiguous? | Every resolved `grill` answer states the decision AND what was rejected; the testing-posture answer names types, cadence, and coverage | Answers phrased as preferences ("probably NDJSON") instead of decisions; a requirement that only exists in the map gist and not in a question file |
| **Technical Feasibility** | Do we know HOW to build each component? | Resolved `sketch` questions with a real artifact under `sketch-NN/`; `research` answers citing primary sources under `## Evidence`; `00-codebase-context.md` naming the actual files that change | A mechanism nobody has exercised in this codebase; a research answer whose `## Evidence` is empty or cites only a blog post |
| **Integration Points** | Are all external dependencies identified? | Every system named in `## Destination` has a resolved question touching it; auth, quota, and failure behavior named per integration | An integration mentioned in an answer but never questioned; a config store that was read but never written to during a sketch |
| **Risk Assessment** | Are blockers documented with mitigations? | Answers that record consequences; `Locked: yes` answers that say what breaks if reversed; ceilings with a stated behavior at the ceiling | A recorded limit with no decided behavior past it; a `Locked: yes` answer with no consequences section |
### Hard caps
A cap overrides your judgment. While a cap condition holds, the dimension **cannot** exceed 17, so the gate cannot pass. `gate` reports the Requirements, Feasibility and Risk caps from the files (`caps`); the Integration cap needs you to match the destination's systems to questions.
| Cap | Condition |
|---|---|
| Requirements ≤ 17 | The testing-posture question is not `resolved`, or its answer omits any of types / cadence / coverage |
| Feasibility ≤ 17 | Any `research` question is `resolved` with an empty `## Evidence` |
| Integration ≤ 17 | A system named in `## Destination` has no resolved question touching it |
| Risk ≤ 17 | Any `Locked: yes` answer records no consequences |
### Honesty rules
- Score the **written record**, not the conversation. If the human agreed to something in a session and nobody wrote it into a `## Answer`, it does not exist and it does not earn points.
- A filled `## Answer` is not automatically 25. An answer that decides but records no consequences tops out around 20.
- Never round up to clear the gate. A 17 that "feels like an 18" is the exact case the gate exists to catch.
- Never move a decision to `## Out of scope` to raise a score. Out-of-scope is a scoping act with a reason; scope-cutting to pass a gate is score inflation with extra steps.
- If two dimensions are borderline, write the one-line justification for each score into the Session End report. Justifications that cannot be written are scores that cannot be defended.
### Worked example — idea `audit-log-s3-export`
Destination: *"A spec a developer can implement: nightly export of tenant audit logs to customer-owned S3 buckets, with a signed manifest per run."*
Resolved questions: [Codebase context](./questions/00-codebase-context.md), [Export format](./questions/01-export-format.md), [Scheduling model](./questions/02-scheduling-model.md), [Destination auth](./questions/03-destination-auth.md), [Retention and replay](./questions/04-retention-and-replay.md), [Testing posture](./questions/05-testing-posture.md), [Throughput ceiling](./questions/06-throughput-ceiling.md). Ruled out: [Failure notification](./questions/07-failure-notification.md).
**First scoring pass:**
| Dimension | Score | Justification against evidence |
|---|---|---|
| Requirements Clarity | 22/25 | Four `grill` answers state decisions and rejections (NDJSON chosen, CSV rejected for nested actor payloads). Testing posture settled: integration + unit, run after each phase, moderate coverage. Minus 3: the manifest's exact field list is unspecified — a Phase 5 spec detail, not an open decision. |
| Technical Feasibility | 21/25 | `sketch-02` uploaded a 400 MB multipart object to a real bucket end to end. `00-codebase-context.md` names `AuditExportJob` and the existing Hangfire registration as the extension points. Minus 4: no component in this codebase has ever assumed a cross-account IAM role; the pattern is documented in AWS docs cited under `## Evidence` but unexercised here. |
| Integration Points | 23/25 | Three integrations, each with a resolved question: S3 (Destination auth), Hangfire (Scheduling model), tenant config store (Codebase context). Auth and quota named per integration. Minus 2: the tenant config store's write path was read but never exercised by a sketch. |
| Risk Assessment | **16/25** | Consequences recorded on Export format and Destination auth. But Throughput ceiling establishes 200 MB per tenant per day at p95 and **records no decided behavior above it** — truncate, spill to the next run, or fail the run is still undecided. |
**Gate result: FAIL** on Risk Assessment (16 < 18), and the orchestrator third clearing condition — the destination reachable with nothing left to decide — fails with it: a decision is genuinely still open. Do not write a PLAN-DRAFT. Name the failure, graduate `questions/08-overflow-behavior.md` (`grill · HITL`, `Blocked by: none`) from the gap, and end the session on the frontier.
**Second scoring pass, one session later**, with [Overflow behavior](./questions/08-overflow-behavior.md) resolved (spill to the next run, alarm at three consecutive spills):
| Dimension | Score |
|---|---|
| Requirements Clarity | 22/25 |
| Technical Feasibility | 21/25 |
| Integration Points | 23/25 |
| Risk Assessment | 20/25 |
All four at 18 or better. Gate passes. Proceed to the mechanism.
## The mechanism, stated plainly
Pathfinder writes `specs/<idea>/PLAN-DRAFT-<YYYYMMDD>.md` with the header line:
`**Status:** Phase 3 Complete - Resume at Phase 4`
`/plan2code-1-plan` **already recognises that exact string.** Its "Check for Existing Progress" block, which runs before Phase 1, reads:
`- Status "Phase 3 Complete - Resume at Phase 4": Resume at Phase 4`
That is the whole handoff. The string is the contract, and it needs **zero changes to the planning skill** — pathfinder is impersonating the Large-project Context Checkpoint that 1-plan's own Phase 3 performs, which writes the same status for the same reason.
Consequences of that being a literal string match:
- Copy it byte for byte. Plain ASCII hyphen-minus surrounded by single spaces. An en dash, a colon, or "Phase 3 complete" in lower case breaks the match and 1-plan starts over at Phase 1 — throwing away every decision the map holds.
- It goes on its own `**Status:**` line in the header block, not buried in prose.
- The file must be named `PLAN-DRAFT-<YYYYMMDD>.md` and live directly in `specs/<idea>/`. `PLAN-DRAFT-*.md` is a reserved name inside `pathfinder/` — never write it there.
- Use the date in `gate`'s `planDraftPath` (it reads the clock). Do not guess it. On `github` (no `gate`), take today's date from the shell and check `ls specs/<idea>/` for an existing draft.
**Discovery is shell-only.** `specs/` is gitignored, so Glob silently returns nothing and every downstream skill would report "no PLAN-DRAFT found". Use `ls specs/` and `ls specs/<idea>/` (Bash) or `Get-ChildItem specs/` (PowerShell) — the same rule 1-plan and `/plan2code-2-document` follow when they look for the file you are about to write.
**Do not append `## Planning Metrics` or any metrics comment.** Pathfinder is not a metered step. 1-plan's Phase 7 owns that block and will add it when it finishes the plan. Do not emit any of the loop's completion tokens listed in the skill's Rules anywhere under `specs/`.
## Map to PLAN-DRAFT mapping
Everything in the draft traces to something on the map. Nothing is invented at handoff time — if a section has no source, that is a gate failure you missed, not a paragraph to write from imagination.
| Source on the cleared map | Becomes |
|---|---|
| `## Destination` | Section 1 Executive Summary (2-3 sentences, present tense) and Section 7 Success Criteria (the destination restated as checkable outcomes) |
| Resolved `grill` answers describing behavior | Section 2.1 Functional Requirements, one `FR-N` per decided behavior |
| Resolved `grill` answers describing performance, security, scale, operability | Section 2.2 Non-Functional Requirements, one `NFR-N` each |
| `## Out of scope` | Section 2.3 Out of Scope — copied line for line, wording and order intact. The ONLY permitted change is the link prefix: `./questions/` becomes `./pathfinder/questions/`, because the draft sits one level above the map. Do not re-word, re-order, or summarise; a re-worded scope boundary is a re-litigated one. |
| The testing-posture question's answer | Section 2.4 Testing Strategy table (Types / Phase Testing / Coverage) |
| Resolved `research` answers and their `## Evidence` | Section 3 Tech Stack — the cited source becomes the Justification cell |
| `questions/00-codebase-context.md` | The `## System Context` section, and the components table in 4.3 |
| Answers with `Locked: yes` | Section 4 Architecture (4.1 Pattern rationale, 4.4 Data Model, 4.5 API Design) **and** Section 9 Assumptions — a locked decision is an assumption downstream work is allowed to rely on |
| Consequences recorded across all `## Answer` sections | Section 6 Risks and Mitigations — the consequence is the Risk, the decision that bounds it is the Mitigation |
| `## Ground rules` | The `AGENTS.md` line in `## System Context`; conventions the plan must not violate |
| The shape of the map (question count, integrations touched, components named) | The `## Scope Assessment` section |
| Resolved `sketch` questions and their artifacts | Section 3 Justification cells and Section 6 Mitigation cells ("proven by `pathfinder/sketch-02/`") |
| Every resolved question, by name | `## Pathfinder Provenance` |
| — | **Section 5 Implementation Phases stays empty.** 1-plan Phase 6 breaks the work into phases. Pathfinder decides; it does not slice. Leave the placeholder note in place and do not put implementation checkboxes there. |
Questions ruled `out-of-scope` never appear in Provenance and never become requirements. Their one line in `## Out of scope` is their only trace — that is the point of the marker.
## The PLAN-DRAFT template
Write this literally, substituting real content. Keep the section numbering exactly as shown — 1-plan and `/plan2code-2-document` both address sections by number.
````markdown
> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.
# Audit Log Export - Implementation Plan
**Created:** 2026-08-03
**Status:** Phase 3 Complete - Resume at Phase 4
**Charted by:** `/plan2code-0-pathfinder` over 9 sessions
**Planning record:** `specs/audit-log-s3-export/pathfinder/map.md` (no PLAN-CONVERSATION - this plan was charted, not conversed)
**Confidence (pathfinder):** Requirements-clarity 22/25 · Feasibility-technical 21/25 · Integration-points 23/25 · Risk-assessment 20/25
---
## 1. Executive Summary
Tenants can have their audit logs exported nightly to an S3 bucket they own, with a
signed manifest per run so they can prove completeness. Export runs on the existing
Hangfire schedule, writes NDJSON, and assumes a customer-provided cross-account IAM
role with an external ID. Runs that exceed the per-tenant daily ceiling spill into the
next run rather than truncating.
## 2. Requirements
### 2.1 Functional Requirements
- [ ] **FR-1:** Export each tenant's prior-day audit events as newline-delimited JSON, one object per event, UTF-8, no BOM
- [ ] **FR-2:** Write a per-run manifest listing object keys, event counts, byte counts, and a SHA-256 per object
- [ ] **FR-3:** Sign the manifest with the platform export key; publish the public key at a stable URL
- [ ] **FR-4:** Assume the tenant-configured IAM role with the tenant's external ID; never use platform-owned credentials against a customer bucket
- [ ] **FR-5:** Allow an operator to replay any run within a 7-day window without duplicating manifest sequence numbers
- [ ] **FR-6:** Spill events above the per-tenant daily ceiling into the next scheduled run, oldest first
- [ ] **FR-7:** Raise an alarm after three consecutive spilling runs for the same tenant
### 2.2 Non-Functional Requirements
- [ ] **NFR-1:** Sustain 200 MB per tenant per day at p95 without extending the nightly window past 04:00 UTC
- [ ] **NFR-2:** Never log tenant event bodies, bucket names, or assumed-role ARNs above debug level
- [ ] **NFR-3:** A failed run must leave no partial objects visible in the customer bucket
- [ ] **NFR-4:** Export must add no schema changes to the audit event write path
### 2.3 Out of Scope
<!-- copied line for line from pathfinder/map.md ## Out of scope; only the link prefix changes -->
- [Failure notification](./pathfinder/questions/07-failure-notification.md) — email/webhook delivery of run failures belongs to the platform alerting effort, not this export. The alarm in FR-7 is raised, not delivered.
- **On-demand export from the tenant UI** — the destination is the scheduled export. A user-triggered export is a separate effort with its own map.
- **Log formats other than NDJSON** — Parquet was raised and ruled past the destination.
### 2.4 Testing Strategy
| Aspect | Decision |
|---|---|
| Types | Unit + Integration |
| Phase Testing | Run after each phase |
| Coverage | Moderate (~60-80%) |
## System Context
**Project type:** Existing codebase — .NET 8 service, `src/Platform.Audit/`
| Aspect | Finding | Source |
|---|---|---|
| Entry points to change | `AuditExportJob`, registered in `HangfireStartup.ConfigureRecurringJobs()` | [Codebase context](./pathfinder/questions/00-codebase-context.md) |
| Existing patterns to follow | Jobs resolve tenant scope via `ITenantScopeFactory`; no job reads config directly | [Codebase context](./pathfinder/questions/00-codebase-context.md) |
| Integration surfaces | S3 (customer-owned), Hangfire scheduler, `TenantConfigStore` | [Destination auth](./pathfinder/questions/03-destination-auth.md), [Scheduling model](./pathfinder/questions/02-scheduling-model.md) |
| Technical debt in the path | `AuditQuery` materialises full result sets; streaming reader needed before FR-1 | [Codebase context](./pathfinder/questions/00-codebase-context.md) |
| System boundaries | Read-only against the audit store; writes only to customer buckets and the run-log table | [Retention and replay](./pathfinder/questions/04-retention-and-replay.md) |
| Conventions in force | `AGENTS.md` present and read; its logging and DI conventions govern | `pathfinder/map.md` ## Ground rules |
## Scope Assessment
**Assessment: Medium** — 11 requirements across 3 integrations, 5 components touched. No Large threshold is met.
| Indicator | Value |
|---|---|
| Requirements decided | 11 (7 FR + 4 NFR) |
| Components | 5 (`AuditExportJob`, `NdjsonWriter`, `ManifestSigner`, `S3RoleAssumer`, `ExportRunLog`) |
| Integrations | 3 (S3, Hangfire, `TenantConfigStore`) |
| Decisions charted | 8 resolved, 1 ruled out of scope |
## 3. Tech Stack
<!-- Phase 4 completes this table. Rows below are decided; do not re-open them. -->
| Category | Technology | Version | Justification |
|---|---|---|---|
| Serialization | `System.Text.Json` NDJSON writer | .NET 8 | [Export format](./pathfinder/questions/01-export-format.md) — no new dependency; CSV rejected for nested actor payloads |
| Object storage | `AWSSDK.S3` multipart upload | 3.7.x | [Throughput ceiling](./pathfinder/questions/06-throughput-ceiling.md) — proven in `pathfinder/sketch-02/` against a real bucket at 400 MB |
| Cross-account auth | STS `AssumeRole` + external ID | — | [Destination auth](./pathfinder/questions/03-destination-auth.md) — AWS confused-deputy guidance cited in that file's `## Evidence` |
| Scheduling | Existing Hangfire recurring job | in-repo | [Scheduling model](./pathfinder/questions/02-scheduling-model.md) — a new scheduler was rejected |
## 4. Architecture
### 4.1 Pattern
Pipeline inside the existing job host: query → stream → chunk → upload → manifest → sign.
Chosen because the audit store is the only source and the export is strictly one-way.
[Locked] A separate export microservice was rejected — see [Scheduling model](./pathfinder/questions/02-scheduling-model.md).
### 4.2 System Context Diagram
<!-- Phase 5 refines. Boundaries above are settled. -->
### 4.3 Components
| Component | Responsibility | Inputs | Outputs | Depends on |
|---|---|---|---|---|
| `AuditExportJob` | Orchestrates one tenant-run | Tenant id, run date | Run result | `TenantConfigStore` |
| `NdjsonWriter` | Streams events to chunked NDJSON | Event stream | Byte stream, counts | — |
| `S3RoleAssumer` | Assumes the tenant role, returns a scoped client | Role ARN, external ID | `IAmazonS3` | STS |
| `ManifestSigner` | Builds and signs the run manifest | Object metadata | Signed manifest | Platform export key |
| `ExportRunLog` | Records runs for replay and spill detection | Run result | Run rows | Platform DB |
### 4.4 Data Model
Manifest sequence numbers are per tenant, monotonic, and reused on replay.
[Locked] See [Retention and replay](./pathfinder/questions/04-retention-and-replay.md).
### 4.5 API Design
<!-- Phase 5 fills. No public API surface was decided during pathfinding. -->
## 5. Implementation Phases
<!-- Intentionally empty. Phase 6 of /plan2code-1-plan breaks the requirements
above into implementation phases. Pathfinder decides; it does not slice. -->
## 6. Risks and Mitigations
| Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|
| Cross-account role assumption is unexercised in this codebase | Medium | High | Spike `S3RoleAssumer` against a second AWS account before any other component |
| A tenant exceeds the daily ceiling indefinitely | Medium | Medium | Spill oldest-first plus a three-run alarm — [Overflow behavior](./pathfinder/questions/08-overflow-behavior.md) |
| Partial objects visible after a failed run | Low | High | Upload to a run-scoped prefix, publish the manifest last — the manifest is the commit point |
| `AuditQuery` materialises full result sets | High | High | Streaming reader is a prerequisite, not an optimisation |
| Customer revokes the role mid-run | Low | Medium | Fail the run whole; replay window covers recovery |
## 7. Success Criteria
- [ ] A tenant with a configured role receives NDJSON and a signed manifest for the prior day, nightly
- [ ] The published public key verifies the manifest signature
- [ ] A 250 MB tenant-day completes without extending the window past 04:00 UTC
- [ ] A replay inside 7 days reproduces the run without a new sequence number
- [ ] A failed run leaves nothing visible in the customer bucket
## 8. Open Questions
<!-- 1-plan's template removes this section when empty; pathfinder keeps it with "None" so a
resuming session can see the map cleared clean, rather than that the section was forgotten. -->
None. The map cleared with zero open questions.
## 9. Assumptions
- Tenants can create an IAM role in their own account — [Destination auth](./pathfinder/questions/03-destination-auth.md) [Locked]
- The nightly Hangfire window remains available and is not contended by other jobs — [Scheduling model](./pathfinder/questions/02-scheduling-model.md) [Locked]
- Manifest sequence reuse on replay is acceptable to tenant compliance teams — [Retention and replay](./pathfinder/questions/04-retention-and-replay.md) [Locked]
- Audit events are immutable once written, so a replay reproduces byte-identical output
## Pathfinder Provenance
Charted over 9 sessions. Each requirement above traces to a decision below; open the
question for what was rejected, why, and what it costs.
| Question | Gist |
|---|---|
| [Codebase context](./pathfinder/questions/00-codebase-context.md) | `AuditExportJob` and `HangfireStartup` are the extension points; `AuditQuery` needs a streaming reader first |
| [Export format](./pathfinder/questions/01-export-format.md) | NDJSON with a per-run signed manifest; CSV rejected for nested actor payloads |
| [Scheduling model](./pathfinder/questions/02-scheduling-model.md) | Reuse the existing Hangfire recurring job; a dedicated export service was rejected |
| [Destination auth](./pathfinder/questions/03-destination-auth.md) | Customer-owned bucket via assumed role plus external ID; no platform-held customer credentials |
| [Retention and replay](./pathfinder/questions/04-retention-and-replay.md) | 7-day replay window, sequence numbers reused on replay |
| [Testing posture](./pathfinder/questions/05-testing-posture.md) | Unit + integration, run after each phase, moderate coverage |
| [Throughput ceiling](./pathfinder/questions/06-throughput-ceiling.md) | 200 MB per tenant per day at p95; multipart upload proven in `pathfinder/sketch-02/` |
| [Overflow behavior](./pathfinder/questions/08-overflow-behavior.md) | Spill oldest-first into the next run; alarm after three consecutive spills |
Ruled out of scope: [Failure notification](./pathfinder/questions/07-failure-notification.md) — recorded in 2.3.
---
**Next:** Resume with `/plan2code-1-plan` at Phase 4 (Tech Stack).
- **Phase 7 verification:** sections 1, 2, System Context and Scope Assessment are already settled — their source of truth is `specs/audit-log-s3-export/pathfinder/map.md`, not this conversation. Verify sections 3-7 only.
- **Phase 7:** replace THIS file in place. Do not create a second PLAN-DRAFT in this folder.
````
### Two things in that template that are not optional
**No scrapable confidence numbers anywhere in the file.** Write the confidence as `Requirements-clarity 22/25 · Feasibility-technical 21/25 · Integration-points 23/25 · Risk-assessment 20/25` — hyphenated dimension labels, sub-scores over 25, no total, no percent sign.
The reason is exact. When a PLAN-DRAFT carries no `METRICS_JSON` comment, the metrics collector falls back to scraping it by regex: an overall-confidence pattern that requires a literal `%`, and four breakdown patterns that match a bare `Requirements` / `Feasibility` / `Integration` / `Risk` followed directly by whitespace, a colon, or a pipe and then digits. **The breakdown patterns do not require a percent sign.** A pathfinder-written draft always lacks that comment until `/plan2code-1-plan` Phase 7 appends one, so both the percent sign *and* the bare dimension words have to be kept off the page — otherwise the pipeline records a planning-step confidence that no planning step ever produced. The hyphen in `Requirements-clarity` breaks the match; a table row reading `| Requirements | 11 |` does not, which is why the Scope Assessment row is labelled `Requirements decided`.
**The `**Next:**` footer must ship with both bullets.** Pathfinder cannot edit the planning skill, so those two instructions travel inside the artifact:
- *Without the verification bullet*, 1-plan's Phase 7 does exactly what it is told to do — "re-read conversation as source of truth" — finds a fresh conversation that starts at Phase 4 and contains no requirements discussion at all, concludes sections 1 and 2 are unsupported, and silently drops the requirements that N pathfinder sessions produced. The bullet redirects the source of truth for the settled sections to `map.md`.
- *Without the replace-in-place bullet*, 1-plan's Phase 7 creates `PLAN-DRAFT-<its own date>.md` alongside yours. `/plan2code-2-document` then finds two drafts in the folder, hits its "Multiple found: List all, ask which to document" branch, and asks the user to disambiguate between a pathfinder draft and a plan draft that partially supersedes it.
Never drop the footer to make the file tidier. It is load-bearing.
After writing the draft, run `lint`: it scans a pathfinder-written PLAN-DRAFT with the metrics collector's own patterns and names any bait that reached the page.
## System Context and Scope Assessment — why they buy you Phase 4
These two named, unnumbered sections are what make "Resume at Phase 4" legitimate rather than a shortcut. They stand in for the phases pathfinder already did the work of:
| Draft section | Satisfies | Because pathfinder already |
|---|---|---|
| Sections 1 and 2 (including 2.4 Testing Strategy) | 1-plan **Phase 1: Requirements Analysis** | Grilled every functional and non-functional decision, and always charted a testing-posture question — that question exists specifically so Phase 1's testing prompt is already answered |
| `## System Context` | 1-plan **Phase 2: System Context Examination** | Wrote `questions/00-codebase-context.md` at Chart Step 3: directory structure, key components verified against actual code, patterns and conventions, integration points, technical debt, boundaries — Phase 2's own checklist, item for item |
| `## Scope Assessment` | 1-plan **Phase 3: Scope Assessment** | Produced the counts Phase 3 measures — requirements, components, integrations — as a byproduct of charting. Map the totals onto Phase 3's Small / Medium / Large table and state the verdict |
Populate `## System Context` from `00-codebase-context.md` and nothing else. It is the one question guaranteed to exist on every map, it was resolved on the spot with the codebase open, and it is a `legwork · AFK` answer — factual, not preferential. Cite it in the Source column so a skeptical reader can check the finding against the file.
Populate `## Scope Assessment` from the shape of the map. Count resolved questions that produced requirements (not `00-codebase-context.md`, not out-of-scope ones), count distinct components named across the answers, count distinct external systems. Apply Phase 3's thresholds honestly: Large if **any** threshold is met. Score the counts, never the session count — a map can take nine sessions to clear and still be Medium, and the `Phase 3 Complete - Resume at Phase 4` status string works regardless of the verdict, so there is nothing to gain by inflating it. Pathfinder cannot count implementation phases (Section 5 is deliberately left empty), so assess on requirements, components, and integrations only.
If the charting session found `AGENTS.md` absent, say so in the `Conventions in force` row rather than leaving it blank. The plan session needs to know the conventions were never available, not guess that they were checked.
## Freeze the map
Once the PLAN-DRAFT is written and saved:
1. Set `**Status:** Cleared` in `map.md`.
2. Bump `**Updated:**` to today.
3. Add a plan pointer line under the status: `**Plan:** ../PLAN-DRAFT-20260803.md`.
4. Leave **everything** under `pathfinder/` exactly where it is — `map.md`, every file in `questions/`, every `sketch-NN/` directory.
**Never delete `pathfinder/`.** It is the rationale record behind the plan: what was decided, what was rejected, why, and what it costs. It sits in the same class as `PLAN-CONVERSATION-*.md` — the transcript a plan is defensible against — and `/plan2code-4-finalize` archives it alongside `PLAN-DRAFT.md` and `PLAN-CONVERSATION-*.md` into `specs--completed/<idea>/`. Deleting it turns every locked decision in the plan into an unexplained constraint six months from now.
Do not tidy it either. Do not collapse resolved questions into the map, do not prune `## Evidence`, do not remove sketch directories because the code is throwaway. The sketch is the proof behind a feasibility score.
**A later session that finds `Status: Cleared` must not resume work on it.** The map is finished; there is nothing left to decide inside it. Point at the PLAN-DRAFT and `/plan2code-1-plan`, and stop. If the destination has been redrawn — the scope grew, an out-of-scope item came back, the goal changed — that is a **fresh effort with a fresh map**, not a resumption: a new kebab-case idea name, a new `specs/<new-idea>/pathfinder/`, charting from Step 1. The frontier stops at the destination, so a new destination gets a new frontier. Reopening a cleared map silently invalidates the PLAN-DRAFT that was built from it, and nothing downstream would notice.
## What to tell the user
Session End for the cleared case reports six things, in this order:
| # | Report |
|---|---|
| 1 | The destination, restated — what the map was finding its way to, now reached |
| 2 | How many decisions were made, and the headline ones **by name** |
| 3 | What was ruled out of scope, and why (one line each) |
| 4 | The four confidence scores, with a one-line justification for any below 21 |
| 5 | The PLAN-DRAFT path |
| 6 | The next command — carried by the Trail Footer, which at `Cleared` routes to `/plan2code-1-plan` |
Example:
> **Destination reached:** a spec a developer can implement for nightly audit-log export to customer-owned S3 buckets with a signed manifest.
>
> **8 decisions made** across 9 sessions. The load-bearing ones: [Export format](./questions/01-export-format.md) settled on NDJSON with a signed manifest; [Destination auth](./questions/03-destination-auth.md) settled on assumed roles with an external ID so we never hold customer credentials; [Overflow behavior](./questions/08-overflow-behavior.md) settled on spilling rather than truncating.
>
> **Ruled out of scope:** failure-notification delivery (belongs to the platform alerting effort), on-demand export from the tenant UI (its own effort), Parquet output (past the destination).
>
> **Confidence:** Requirements-clarity 22/25 · Feasibility-technical 21/25 · Integration-points 23/25 · Risk-assessment 20/25. Feasibility is held at 21 because cross-account role assumption is unexercised in this codebase — it is a documented pattern, not a proven one here, and it is the first thing to spike.
>
> **Written:** `specs/audit-log-s3-export/PLAN-DRAFT-20260803.md`. It resumes planning at Phase 4 — sections 1, 2, System Context and Scope Assessment are already settled. The reasoning behind every one of them stays in `specs/audit-log-s3-export/pathfinder/`; do not delete it.
Then the closing block from the skill's Session End — the mascot with the message *The way is clear! Time to plan!* followed by the Trail Footer, whose trail shows every stop walked to the `⚑` destination and whose command is `/plan2code-1-plan`.
**One branch.** If `## Ground rules` records `AGENTS.md` as **absent**, recommend `/plan2code-init` FIRST, and offer `questions/00-codebase-context.md` as its input:
> Before planning: this project has no `AGENTS.md`, and `/plan2code-1-plan` blocks on that. Run `/plan2code-init` first and attach `specs/audit-log-s3-export/pathfinder/questions/00-codebase-context.md` — the recon pass already established the structure, conventions, and integration points it asks for. Then `/plan2code-1-plan`.
Nothing to commit — `specs/` is gitignored. Say so once, then stop.
## Failure modes
| Symptom | Cause | Fix |
|---|---|---|
| 1-plan starts at Phase 1 and re-asks for requirements | The status string does not match byte for byte | Compare against the quoted line above; watch for en dashes and casing |
| 1-plan cannot find the draft at all | Glob was used to discover `specs/` | Shell only: `ls specs/` |
| `/plan2code-2-document` asks which of two drafts to use | The replace-in-place bullet was dropped from the footer | Merge the two drafts into the pathfinder-dated one, delete the other, restore the footer |
| The finished plan is missing requirements the map decided | The verification bullet was dropped from the footer | Re-derive 2.1 and 2.2 from the resolved answers, restore the footer |
| Metrics report a planning confidence nobody scored | A percent sign, or a bare dimension word followed by a number, reached the file | Hyphenate the dimension labels and drop the percent sign |
| A locked decision in the plan has no visible reason | `pathfinder/` was deleted or pruned | Unrecoverable. This is why the freeze step exists |
| Gate passes but the first implementation session immediately hits an undecided question | A dimension was rounded up | The gate was the check. Score the written record, not the feeling |
@@ -0,0 +1,47 @@
# Questions & Map Format
> Part of plan2code-0-pathfinder — the on-disk format both modes share: directory layout, `NN` numbering, the question-file schema, `Type:` vocabulary, and the marker / blocking rules. The main file keeps only the marker legend and a layout gist; the authority is here.
>
> **This file describes the `local` backend only.** If `## Ground rules` says `**Backend:** github`, the equivalence table in `github-issues.md` replaces every rule below — there are no files, no `NN`, no schema lines, and no checklist.
## Layout
```
specs/<idea>/
├── pathfinder/
│ ├── map.md <- the index
│ ├── questions/NN-<slug>.md <- 00-codebase-context.md always exists
│ ├── briefs/brief-<YYYYMMDD>.md <- plain-English reports, written by BRIEF mode only
│ └── sketch-NN/ <- optional runnable sketch, throwaway
└── PLAN-DRAFT-<YYYYMMDD>.md <- written ONLY when the map clears
```
`questions/` is ground truth; `map.md` is a rebuildable index that gists and links. A filled `## Answer` beats any `State:` line. Detail lives in exactly one place — the question file.
## Numbering
`NN` is zero-padded from `00`, assigned in dependency order (blockers lower), **never reused or renumbered** — links and `Blocked by:` would rot silently. Next = max + 1 (`frontier`'s `nextNN`), never below the map's `**Next NN:**` high-water line, which the script keeps so a deleted question's number stays retired (chart.md, Step 6 rule 5). A one-digit file name (`1-export.md`) still reads as `01`, but `lint` asks for the rename. `00` is always `00-codebase-context.md`, never anything else. Gaps in the sequence are normal and harmless.
## The six schema lines
Each question file carries six contiguous `Key: value` lines after its H1 — NOT YAML frontmatter, no `---` delimiters:
| Line | Values |
|---|---|
| `Type:` | `grill · HITL` \| `research · AFK` \| `sketch · HITL` \| `legwork · HITL` \| `legwork · AFK` — one token, so type and mode cannot drift |
| `State:` | `open` \| `claimed` \| `resolved` \| `out-of-scope` |
| `Blocked by:` | `none` \| `02, 04` |
| `Claimed:` | `none` \| `<YYYY-MM-DD HH:mm>` — a work-in-progress lock, not a completion record |
| `Resolved:` | `none` \| `<YYYY-MM-DD>` — set in the same write as `State: resolved` (Work Step 7). The date the decision landed; briefs filter on it |
| `Locked:` | `yes` only when hard to reverse AND surprising without context AND a real trade-off |
**Backfill (older maps):** a file with a filled `## Answer` but `Resolved: none` (or no `Resolved:` line at all) gets the line backfilled at reconcile — from the `Claimed:` date, or today if `Claimed: none`. Say so once in the session report.
**Type meanings.** **grill** (default) — a decision only the human can make. **research** — a fact outside this directory gates it. **sketch** — the human needs something concrete to react to. **legwork** — manual work that must happen before a decision is possible.
## Markers and blocking
**Map markers**, rebuilt from the files every session: `[ ]` open — **these rows ARE the frontier** · `[/]` claimed · `[x]` resolved · `[!]` open but blocked · `[-]` out of scope. Resolved rows carry the gist plus the `Resolved:` date in italics: `— <gist> *(YYYY-MM-DD)*`.
`scripts/pathfinder.mjs` implements exactly these rules (`reconcile` rebuilds the markers, `frontier` computes blocking, `lint` checks the schema), so this file is the reference for reading its output and for writing a question file. Apply the rules by hand only when the script cannot run.
**Unblocked** ⟺ every `NN` in `Blocked by:` is `resolved`. **Stranded:** a blocker gone `out-of-scope` never resolves — the question is not merely blocked. Re-frame its `## Question` to drop the dependency, or rule it out too. Never leave it sitting.
@@ -0,0 +1,566 @@
# Resolution Playbook
> Loaded at the top of MODE B. Work Step 6 routes here by `Type:`; Work Step 8 uses the fog procedure at the end.
>
> **Backend note.** Every resolution technique here is backend-independent — the type table, the primary-source rule, the sketch tiers, the checklist discipline, the `## Answer` anatomy. On `**Backend:** github`, `Type:` is a `pathfinder:<type>-<mode>` label, `## Answer` and `## Evidence` are comments rather than file sections, and a research subagent gets an issue URL instead of a path; see `github-issues.md`. Sketches stay on local disk regardless. `pathfinder.mjs` reads local files only, so on `github` do each script step below except `lint` by hand with `gh`, as `github-issues.md` describes.
Every question resolves into the SAME shape — a filled `## Answer` plus whatever `## Evidence` backs it. The type only decides how you get there.
| `Type:` | Who drives | Parallel? | What "resolved" means |
|---|---|---|---|
| `research · AFK` | Subagent, alone | **Yes** — many at once | Facts found and cited; no decision made |
| `sketch · HITL` | Agent builds, human reacts | No | The human reacted and chose |
| `legwork · AFK` | Agent, alone | No | The work is done; resulting facts recorded |
| `legwork · HITL` | Human does, agent waits | No | The human confirmed it is done |
| `grill · HITL` | Human decides, agent interrogates | No — but probes batch, up to 3 per turn | The human said it in their own words |
**One question _file_ at a time** holds for every row except `research` — several may resolve in one session, but strictly in sequence, per the Cadence below. `research` is also the one type that is **never claimed**: its subagents fire unclaimed (the file stays `State: open`), so `claim` and its one-claim rule govern grill, sketch and legwork only, and a grill can be claimed while research runs. A research question is finished with `## Answer` plus `resolve` once its subagent is back, or re-fired from `reconcile`'s `refire` if it was cut off. HITL rows are never self-answered — an agent that writes its own `## Answer` on a `grill` has broken the skill.
Every file you create along the way — question files, sketch READMEs — opens with the banner:
`> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.`
---
## Cadence: decide → menu → checkpoint
The Work loop no longer ends after one decision. It ends when the human stops it, or when the checkpoint recommends stopping — whichever comes first. This holds for a solo developer and for a Product + Engineering room alike; a meeting is just a session with more people answering, and needs no special invocation.
```
claim ONE → resolve → recap → confirm → record → mark map [x]
→ MENU (trail.md Form C): takeable frontier + resume command, human picks or stops
→ claim the next … repeat …
```
**The menu turn.** After every recorded decision, use the structured question tool per trail.md §Form C: offer up to three takeable questions (from the map's checklist rows — name + gist, no file re-reads) plus `Start fresh`; `Other` covers another frontier name. Flag anything just unblocked, mark exactly one description `RECOMMENDED`, and print the resume command under `OR START FRESH` before the tool. Lowest `NN` stays the *recommendation* default, never the selection — the human picks.
**The checkpoint is the menu with the recommendation flipped.** The old one-per-session rule had two jobs: crash-safety and context hygiene — a fresh window per decision keeps the model sharp. Sequential claims preserve the first. The checkpoint preserves the second: after **~3 questions resolved this session** (soft cap), or earlier after a heavy, contested, or `Locked: yes` decision that would benefit from settling overnight, mark the tool's `Start fresh` option RECOMMENDED and say why in the question text — "to keep me sharp." The human can override in either direction; a recommendation is something to push against, not a gate.
The restart costs one paste and loses nothing: every claim, answer, and map update hits disk before anything else happens, and the next session assumes no memory anyway. The fresh session reconciles, renders the trail, and presents the menu again.
**In-loop hygiene** — what keeps each cycle lean between checkpoints:
- Read only what the cycle needs: the claimed question file plus resolved neighbors' *gists* (Work Step 5). Never re-read the whole `questions/` directory between cycles — reconcile ran once at session start, and this session is the only writer.
- The menu renders from the map's checklist rows, never from re-reading question files.
- Strictly one claim open at any moment. Claim → record → release, then the menu. Never two. (Research subagents run unclaimed beside it; see the top of this file.)
**Group sessions.** When Product and Engineering (or any room) work the map together, the grilling playbook's group-session rules apply: probes address the room, the recap captures the decision in the room's words, and a contested pick is never recorded as decided — both positions go under `## Evidence`, the question stays `open`, and the menu moves on (or offer to convert it to `sketch`; reacting to something concrete resolves more standoffs than argument). At Session End, offer `write a brief for today` — the brief doubles as the meeting minutes.
**Stopping is always legitimate.** Ending after one decision is fine. The skill no longer forces deliberation time between decisions; it *advises* it, via the checkpoint, where a decision was heavy enough to deserve settling overnight.
---
## `research · AFK`
The only type an agent resolves alone, and the only type that may run several at once. Charting fires them in a batch at Chart Step 8; MODE B fires any that appear later the same way, unclaimed: never run `claim` on a research question. When one tops the frontier at Work Step 4, fire it, then claim the first non-research question.
### Spin up a subagent
One subagent per research question. Do not read the docs yourself in the main session — the point is that the main session's context stays clean for the decision work.
The subagent's brief must carry, verbatim:
1. The absolute path of the question file it owns.
2. The `## Question` text.
3. The instruction to write into `## Evidence` of THAT file and nothing else — `## Answer` and `State:` are written by the session that fired it, at Work Step 7.
4. The primary-source rule below.
5. The no-deciding rule below.
6. The instruction to end `## Evidence` with `**Research complete:** <YYYY-MM-DD>` as its last line, written only once every claim is in. Without that line, the next session cannot tell finished research from research that was cut off.
### Wait for every subagent
Subagents run inside this session's process, even when the harness runs them in the background. Close the terminal, or end the session, and they die mid-write, leaving `## Evidence` half-filled. So:
- **Do not reach Session End, or post a web console `finish`, while any research subagent is still running.** Wait for each one to report back. The main session does not read the sources while it waits; it only collects completions.
- While waiting in the web console, post `{"agent":{"status":"working","activity":"Researching 3 questions (1 of 3 back)"}}` and update the count as each returns, so the page never says "done" while work is still in flight.
- Never tell the human the research "will finish on its own" or is "running in the background, ready for next time". It is not: it lives in this terminal.
- Once all are back, Session End reports each research question with a one-line summary of what was found (not decided).
### Interrupted research
At Work Step 2, `reconcile` lists in `refire` every research question that is not resolved and whose `## Evidence` lacks the `**Research complete:**` line (with its path and `## Question` text for the brief): it was never finished, by a closed terminal, a crash, a rate limit. Re-fire its subagent with the same brief, plus one instruction: keep what is already in `## Evidence`, verify it, fill the gaps, then write the line. Say so in one line to the human. A map written before this line existed shows the same symptom; re-firing it costs a little time and nothing else.
### Primary sources only
Investigate against **primary sources** — official documentation, the library's own source code, the RFC or spec text, the first-party API reference, the vendor's own pricing page, the actual response from a live endpoint. Never a secondary write-up of them. A blog post, a Stack Overflow answer, or a model's recollection is a *lead*, not a source: follow every claim back to the source that owns it, and cite that.
| Claim about | Source that owns it |
|---|---|
| A library's behavior | That library's source or its own docs for the installed version |
| An HTTP API's shape | The vendor's API reference, or a real captured request/response |
| A file format | The format specification |
| A limit or quota | The vendor's own limits page, dated |
| This project's behavior | The code in this repo, by path |
If no primary source can be found, say so explicitly in `## Evidence` and mark the claim UNVERIFIED. An honest gap is worth more than a confident secondhand sentence — the gap becomes a `legwork` question (go run it and see) or a `grill` (the human decides under uncertainty).
**Version-pin everything.** "Node supports it" is not a finding. "Node 20.11 supports it; the project's `.nvmrc` pins 18.19, which does not" is.
### It does not decide
A research question's `## Answer` states the FACTS FOUND and what they imply. It never picks. The moment the findings enable a choice, that choice becomes a **separate `grill`** question, written at Work Step 8.
This split is the whole reason research is safe to run AFK. An agent that reads three storage options and then declares a winner has quietly made an architectural decision the human never saw.
### The dropped git branch
Upstream captures each prototype and research run on a throwaway git branch. **Pathfinder deliberately drops that.** `specs/` is already untracked scratch space, and plan2code never touches git during pathfinder. The question file IS the durable capture. Do not create branches, do not commit, do not mention branches in the `## Answer`.
### Worked example
`questions/04-object-lock-retention.md` after Work Step 7 recorded its subagent findings:
```markdown
> Pathfinder planning note - decisions, not implementation work. Archive with the spec; do not delete.
# Object Lock retention semantics for the archive bucket
Type: research · AFK
State: resolved
Blocked by: none
Claimed: 2026-08-03 14:02
Resolved: 2026-08-03
Locked: no
## Question
If archived messages land in S3 under Object Lock, what retention modes exist,
which of them can be shortened or lifted after the fact, and what does each
require of the bucket at creation time? We need the facts before deciding
whether a 7-year hold can coexist with a customer's right-to-delete request.
## Answer
**Facts found**
1. Object Lock supports two retention modes: GOVERNANCE and COMPLIANCE.
2. Under GOVERNANCE, a principal holding `s3:BypassGovernanceRetention` can
shorten or remove a retention period on an object version.
3. Under COMPLIANCE, no principal — including the account root — can shorten
or remove it until the retain-until date passes.
4. Object Lock can only be enabled at bucket creation, and enabling it forces
versioning on permanently.
5. A legal hold is independent of retention mode, has no expiry, and is
toggled by any principal with `s3:PutObjectLegalHold`.
**What this implies**
- A right-to-delete request is satisfiable under GOVERNANCE and is NOT
satisfiable under COMPLIANCE before the retain-until date. These two
obligations cannot both be met on the same object version.
- Retrofitting Object Lock onto the existing `archive-prod` bucket is not
possible; it would require a new bucket and a full copy.
**Decision this enables (not made here)**
Which mode the archive uses, and how right-to-delete is honored under it,
is now a sharp question — see Work Step 8, spun off as a grill.
**Gist:** COMPLIANCE mode is irreversible even for root; GOVERNANCE is
bypassable — and Object Lock cannot be added to the existing bucket.
## Evidence
- Two modes, and the governance-bypass permission —
AWS S3 User Guide, "Object Lock overview", section "Retention modes".
https://docs.aws.amazon.com/AmazonS3/latest/userguide/object-lock-overview.html
(retrieved 2026-08-03)
- COMPLIANCE cannot be shortened by any user including the root user —
same page, "Compliance mode" paragraph, verbatim: "no user can overwrite or
delete the object version during the retention period."
- Enable-at-creation-only, and forced versioning —
AWS S3 User Guide, "Enabling Object Lock", first note block.
https://docs.aws.amazon.com/AmazonS3/latest/userguide/object-lock-configure.html
- Legal hold independence and no expiry —
same guide, "Legal holds" section.
- Existing bucket has Object Lock disabled — verified against this account:
`aws s3api get-object-lock-configuration --bucket archive-prod`
returns `ObjectLockConfigurationNotFoundError`. (run 2026-08-03)
- UNVERIFIED: whether our compliance counsel treats GOVERNANCE as sufficient
for the SEC 17a-4 attestation. No primary source exists for this — it is a
human judgment, not a fact. Belongs in a grill.
**Research complete:** 2026-08-03
```
Note what the example does: every claim carries its own citation, a live command counts as a primary source, the one thing that cannot be sourced is flagged rather than smoothed over, and the closing `**Research complete:**` line marks the subagent's work as whole.
---
## `sketch · HITL`
Raise the fidelity of the discussion by making something cheap and concrete for the human to react to. "How should this behave?" and "what should this look like?" produce vague answers in the abstract and sharp ones in front of an artifact.
Two tiers. **Start at Tier 1 every time.**
### Tier 1 — paper sketch (the default)
No executable code. You write a concrete thing into the question's `## Evidence`, show it to the human, and they react. Most sketch questions never need more than this.
Shapes that work:
| Shape | Use when the question is |
|---|---|
| Outline | "What are the steps, and in what order?" |
| State table | "What states exist and which transitions are legal?" |
| Worked example with real numbers | "Does this rule produce sane results?" |
| Fake request/response pair | "What should this API actually look like?" |
| ASCII UI | "What goes on this screen and what is primary?" |
| Decision table | "Under which conditions do we do which thing?" |
Rules for a paper sketch: use REAL-looking content, never placeholders. Real customer names, real message counts, real timestamps, real error strings. A table full of `foo` and `item 1` gets nodded at; a table with `retention_expired` in it gets argued with, and the argument is the point.
Present it, then ask the pointed questions the sketch actually opens — up to three, batched, in the same turn as the artifact. A sketch is the one place where several probes come free: the human has the whole picture in front of them, so a second and third question cost them almost nothing, and a state table with three questionable rows should not take three sessions.
Pointed is the discipline that survives batching. Never "thoughts?" — every probe names its row or element: "row 5 says a flagged message still archives when retention expires. Right, or does the hold pin it in place?" Show the artifact inline, then immediately invoke the structured question tool for up to three pointed reaction probes. The artifact stays in prose; every answer control stays in the UI, with `Other` available for hybrids and corrections.
Follow the structured-tool rules in `grilling.md`: short distinct headers, the pointed question plus why-it-matters in the question field, genuine alternatives with trade-offs in descriptions, and one recommendation per probe.
Putting up a sketch is a **Form B turn** in `trail.md`: close it with `WAITING ON YOU` naming the probes, never a resume command. The human is meant to react to the artifact in this conversation, and a footer telling them to start a new one throws the sketch away.
#### Paper sketch example A — state table
For `questions/06-review-lifecycle.md`:
```markdown
| From | Event | To | Legal? |
|--------------|----------------------|--------------|-------------|
| ingested | policy match | flagged | yes |
| ingested | no match, 24h passes | archived | yes |
| flagged | reviewer clears | archived | yes |
| flagged | reviewer escalates | escalated | yes |
| flagged | retention expires | archived | ← QUESTION |
| escalated | case closed | archived | yes |
| escalated | retention expires | escalated | stays put |
| archived | legal hold applied | held | yes |
| held | hold released | archived | yes |
| held | retention expires | held | hold wins |
Open on this sketch: an item sitting in `flagged` when its retention window
expires. Row 5 currently drops it to `archived` unreviewed. The alternative is
that expiry cannot fire while a human review is outstanding — retention pauses.
```
#### Paper sketch example B — fake request/response pair
For `questions/09-export-job-api.md`:
```markdown
POST /v1/exports
{
"channel_ids": ["ms-teams-legal", "ms-teams-trading"],
"from": "2025-01-01T00:00:00Z",
"to": "2025-12-31T23:59:59Z",
"format": "eml",
"include_attachments": true
}
202 Accepted
{
"export_id": "exp_9fK2mQ",
"state": "queued",
"estimated_messages": 418377,
"estimated_bytes": 12884901888,
"poll_url": "/v1/exports/exp_9fK2mQ",
"expires_at": "2026-08-10T14:00:00Z"
}
GET /v1/exports/exp_9fK2mQ
200 OK
{
"export_id": "exp_9fK2mQ",
"state": "partial_failure",
"messages_written": 418202,
"messages_failed": 175,
"failure_manifest_url": "https://.../exp_9fK2mQ-failures.csv",
"download_urls": ["https://.../exp_9fK2mQ-part-001.zip", "..."]
}
Open on this sketch: `partial_failure` hands back a download plus a manifest of
what is missing. The alternative is all-or-nothing — 175 failures void the whole
12 GB export. Which does a compliance officer actually want at 4pm on a Friday?
```
#### Paper sketch example C — ASCII UI
For `questions/11-reviewer-queue-layout.md`:
```markdown
+----------------------------------------------------------------+
| Review Queue [ Mine 42 ] [ Team 318 ] [ Overdue 7 ] |
+---------------------------+------------------------------------+
| ! 2d K. Ondrusek | From: Kamil Ondrusek |
| "…move the block…" | To: trading-desk (14 members) |
| trading-desk | 2026-08-01 09:14 MS Teams |
|---------------------------| |
| 1d A. Whitfield | Policy hit: BLOCK-TRADE-LANGUAGE |
| "confirming size" | Confidence: 0.91 |
| trading-desk | |
|---------------------------| > can you move the block before |
| 4h R. Iyer | the close? size is 40k |
| "attached the deck" | |
| legal-general | [ Clear ] [ Escalate ] [ Hold ] |
+---------------------------+------------------------------------+
Open on this sketch: the policy hit and its confidence sit in the detail pane,
so the list gives no reason to pick one item over another beyond age. Should the
list rank by confidence instead of age, and show the rule name per row?
```
### Tier 2 — runnable sketch (available, and gated)
Every other plan2code step forbids writing code during planning. **Pathfinder is the one exception**, because some questions genuinely cannot be settled on paper: "does this state model actually hold once you push it through the ugly cases?", "what should this feel like?" A paper state table always looks fine. Driving it by hand for ninety seconds is where it falls over.
**All three gates must open before you write a line of code:**
1. The paper sketch was tried and did not settle it. Not skipped — tried. Say what the paper sketch failed to resolve.
2. There is an obvious way to run it in this project — an existing runtime and task runner. Do not add a package manager, language, or framework for a sketch.
3. The user says go. Ask explicitly: *"Paper didn't settle row 5. I can build a throwaway terminal app under `specs/audit-export/pathfinder/sketch-01/` that lets you drive the state machine by hand — about 60 lines, one command, deleted after. Go?"*
Any gate that stays shut: stay on paper, or convert the question to a `grill`.
#### Rules for a runnable sketch
1. **Throwaway from day one, and clearly marked.** Its README's first line is the banner, and its second says what question it exists to answer.
2. **It lives ONLY at `specs/<idea>/pathfinder/sketch-NN/`** — never in the project's own source tree, never beside the module it is sketching for. This is where pathfinder deliberately departs from upstream: upstream co-locates prototypes with the real code; pathfinder quarantines them, because `specs/` is gitignored scratch and the project tree is not.
3. **One command to run.** Print the exact command to the user. They must not have to remember a path or a flag. If the project has a task runner, use the runner's own idiom, but keep the entry point inside `sketch-NN/`.
4. **No persistence.** State lives in memory. Persistence is what the sketch is checking, not something it leans on. If the question is specifically about storage, use a local file named so its disposability is obvious.
5. **Skip all polish.** No tests, no error handling beyond what makes it run, no abstractions, no "we might want X later."
6. **Surface the full relevant state after every action** (logic) **or on every variant switch** (UI). The user must see the whole picture change, not a delta.
7. **Never merged.** The sketch is not lifted into the project. Only the validated decision it produced survives, in the `## Answer`. Reference the sketch directory from `## Evidence` so a later reader can re-run it, but the code is scaffolding, not output.
#### Pick the branch: logic or UI
| Question shape | Branch | Artifact |
|---|---|---|
| "Does this state model / data model / rule hold?" | **Logic** | One tiny interactive terminal app |
| "What should this look like?" | **UI** | Several radically different variations, switchable |
Getting this wrong wastes the entire sketch. If the question is genuinely ambiguous and the user is unreachable, default by what the question touches — a backend module or a rules engine points to logic, a page or component points to UI — and **state the assumption in the first lines of the sketch's README**, so the human can reject the framing before reading the code.
**Logic branch.** Build the smallest interactive terminal app that pushes the machine through the cases that are hard to reason about on paper. Keep the logic itself pure — a reducer, a state machine, or a small set of pure functions over a plain data type — with the terminal shell as a thin wrapper that imports it and never the reverse. Each frame: clear the screen, print the whole current state one field per line, then print the key legend, e.g. `[f] flag [c] clear [e] escalate [h] hold [t] advance clock 1d [q] quit`. Re-render the entire frame after every keystroke; never append to scrollback. The whole frame fits on one screen. The interesting moment is the user saying "wait, that shouldn't have been possible" — that is a bug in the *idea*, which is the entire point. Add actions on request; sketches evolve.
**UI branch.** Generate **several radically different variations side by side, switchable** — not one polished take. Default to 3, cap at 5. They must disagree about structure: different layout, different information hierarchy, different primary affordance. Three tweaked card grids is wallpaper, not a sketch. If two drafts come out similar, redo one with an explicit constraint against the shape they share. Switch by a URL search param plus a small floating bar (previous / current variant name / next), following whatever routing convention the project already uses — but with the files under `sketch-NN/`. Wire variants to stubbed data, never to real mutations; the question is what it should look like, not whether the backend works. The most valuable feedback is usually "I want the header from B with the list from C" — that hybrid IS the answer, and it goes in the `## Answer`.
#### After a runnable sketch
Record the verdict and the question it settled in `## Answer`. Record the sketch path, the run command, and what the user actually said while driving it in `## Evidence`. Leave the directory in place — it is gitignored scratch, it costs nothing, and the next session may want to re-run it. Never copy any of it into the project.
---
## `legwork · HITL` or `legwork · AFK`
The one type that DOES rather than decides. There is nothing here to research, sketch, or grill — a decision is simply blocked until some manual work happens.
Typical: provisioning access to a system, signing up for a service so its API can actually be judged, moving a data sample somewhere it can be looked at, requesting a sandbox tenant, reading the codebase (`questions/00-codebase-context.md` is always this type).
**It earns its place only by unblocking a decision, never by delivering the destination.**
### AFK or HITL
| Mode | When | How it resolves |
|---|---|---|
| `legwork · AFK` | The agent can do it with the tools it has — read the code, run a query, count rows, inspect a config | Do it, record the facts, resolve |
| `legwork · HITL` | It needs a human's hands, credentials, card, or signature | Hand over a precise numbered checklist and WAIT |
Drive it alone wherever you can. Do not hand a human a checklist for work you could have done yourself.
### The HITL checklist
Numbered, specific, and verifiable — every line names the exact place to click, the exact value to use, and what the human should see when it worked. Plain English throughout: exact names and values where they carry the work, no Pathfinder vocabulary anywhere (see *Say it in plain English* in the grilling playbook). No `- [ ]` checkboxes; question files never carry them.
```markdown
1. Go to https://console.vendor.example/settings/api and sign in with the
shared ops account (credentials in 1Password, item "Vendor Ops").
2. Create an API key named `pathfinder-eval-2026-08`. Scope it to
read-only — untick "Write" and "Admin".
3. Copy the key into 1Password as a NEW item named "Vendor Eval Key".
Do not paste it into this chat or into any file under specs/.
4. On the same page, note the "Rate limit" value shown for the key
and tell me the number.
5. Under Settings > Data, note whether "Historical backfill" is listed
as included or as a paid add-on, and tell me which.
Tell me when 1-5 are done, plus the two values from steps 4 and 5.
```
Immediately after the checklist, invoke the structured question tool: `Done` · `Blocked` · `Need help` · `Defer`; the question text names every value the human must include, and `Other` carries those values/details. Then stop and wait. Do not guess, proceed, or mark it resolved on the assumption it went fine.
Handing over a checklist is a **Form B turn** in `trail.md` — close with `WAITING ON YOU` naming the checklist and the values you asked for, and no resume command. The human may be gone for hours, but the session is still theirs to come back to; only park it as a session end (Form A) once you are actually stopping.
### What its `## Answer` records
Two parts: **what was done**, and **the resulting facts later questions depend on**. Credentials locations (never the credentials), new URLs, row counts, version numbers, quota limits, table shapes, file paths.
```markdown
## Answer
**Done.** Read-only API key provisioned against the shared ops account and
stored in 1Password as "Vendor Eval Key". No key material is stored under specs/.
**Facts other questions depend on**
- Key location: 1Password item "Vendor Eval Key" (ops vault).
- Rate limit: 600 requests/minute per key, burst 1000.
- Historical backfill beyond 90 days is a paid add-on, not included in the
eval tier — so any evaluation against real 2024 traffic needs a purchase.
- Base URL for the eval tenant: https://eval-3f2.vendor.example/api/v2
(differs from the production host in their docs).
**Consequences** — the 90-day eval ceiling means the volume question cannot be
answered against real historical data on this tier; it has to be extrapolated
or the add-on has to be bought. That is a fresh decision, not one to make here.
**Gist:** Read-only eval key in 1Password; 600 rpm; history capped at 90 days
without a paid add-on.
```
### The guard
If the legwork turns out to BE the deliverable rather than an unblocker — you are migrating the data, not sampling it; you are building the integration, not evaluating it — **it is out of scope for pathfinder.** Stop. Say so plainly:
> "This has stopped being legwork that unblocks a decision and become the work itself. Pathfinder plans; it doesn't build. I'm ruling this out of scope and it belongs in planning."
Rule the question out with `rule-out <NN> --reason "<what it turned into>"` (it marks the row `[-]` and adds the `## Out of scope` line), and hand off to `/plan2code-1-plan` for that piece. Never let pathfinder quietly become the implementation.
---
## `grill · HITL`
The default type: a decision only the human can make. Route to **the grilling playbook** — it owns the interrogation technique, structured-question-tool-first delivery, batching rules (up to three independent probes per turn), prose fallback when the tool is unavailable, and the recommend-then-ask pattern.
Three things this playbook adds on top:
- **Zoom before you grill.** Work Step 5 already had you read the claimed question plus anything it references. Bring the resolved neighbors' gists into the first message so the human is not re-litigating settled ground.
- **Batch probes, not question files.** One question file at a time is unchanged. A batch of three probes resolves ONE `questions/NN-*.md`; the next file starts only after this one's menu turn (§Cadence).
- **The human's own words.** A `grill` resolves only through live exchange. Never write the `## Answer` from what you inferred they would probably say — and a probe they skipped twice is unanswered, not decided.
A good grill `## Answer` contains four things:
1. **The decision** — stated flatly, in the human's terms, not hedged.
2. **What was rejected and why** — the alternatives that were live during the conversation, each with the reason it lost. This is the part that stops the decision from being reopened in three weeks.
3. **The consequences for other questions** — which open questions this constrains, which fog patches it just made sharp, which resolved answers it complicates. Name them; never number them in prose.
4. **A one-line bold Gist.**
---
## Writing the `## Answer`
Same anatomy for every type. Append it at Work Step 7; never edit `## Question` to match the answer. Then run `node "<S>/pathfinder.mjs" resolve specs/<idea>/pathfinder <NN>`: it sets `State: resolved` **and** `Resolved:` to today (from the clock; the date is what lets a brief report "decided on <day>"), and writes the map row from your `**Gist:**`. It refuses an Answer with no Gist line.
| Part | Required | Content |
|---|---|---|
| The decision (or, for `research`, the facts found) | Always | What was settled, stated flatly |
| Rejected alternatives, with reasons | When alternatives existed | Each option that lost, and why |
| Consequences | Always | Effects on other questions, named not numbered |
| `**Gist:**` | Always | One line, last |
| `## Evidence` | When there is any | Sources, sketch paths, transcript quotes, commands run |
Worked example — `questions/03-export-format.md`:
```markdown
## Answer
**Decision.** Exports are written as one `.eml` file per message inside a ZIP,
with a top-level `manifest.csv` giving message id, channel, participants,
timestamp, SHA-256, and relative path. One ZIP per 2 GB, numbered `part-001`.
**Rejected**
- **Single NDJSON file.** Compact and trivially streamable, but the review
vendors named in Codebase context both ingest `.eml` natively and neither
parses NDJSON. Rejected because it moves the conversion cost onto the
customer's e-discovery team.
- **PST.** What the legal team asked for by name, but PST is a single-writer
format with a practical 50 GB ceiling and no first-party writer outside
Outlook. Rejected on the ceiling alone — the 2025 trading-desk export is
~12 GB and growing 40% year over year, so the ceiling is 3 years out.
- **One ZIP, no parts.** Rejected because S3 presigned downloads over 5 GB
fail on several corporate proxies the support team has already seen.
**Consequences**
- Makes [Export job API](./questions/09-export-job-api.md) sharper: the
response must return an ARRAY of download URLs, not one.
- Constrains [Integrity attestation](./questions/12-integrity-attestation.md) —
a per-message SHA-256 already exists in the manifest, so attestation can hang
off the manifest rather than needing a separate hash pass.
- Kills the "streaming export" fog bullet: parts and streaming are exclusive.
Removed from `## Not yet specified`.
**Gist:** One `.eml` per message in 2 GB ZIP parts, with a manifest.csv
carrying per-message SHA-256.
```
### The gist
The gist is what gets copied into the map row. It is not a summary of the answer — it is the one line that lets a future session decide, at a glance, whether to open the file.
- One line. Fits in a table row without wrapping twice.
- Says what was DECIDED, not what was discussed. "Chose ZIP parts" is weak; "One `.eml` per message in 2 GB ZIP parts" is judgeable.
- Carries the number or name that matters, if there is one.
- Never the full answer. If it needs a semicolon and a subordinate clause, cut it.
The map row it produces — gist, then the `Resolved:` date in italics:
```markdown
- [x] [Export format](./questions/03-export-format.md) — one `.eml` per message in 2 GB ZIP parts, manifest.csv carries SHA-256 *(2026-08-03)*
```
---
## Graduating the fog (Work Step 8)
`## Not yet specified` is the fog: in-scope questions you can *see* coming but could not phrase sharply when you wrote the map. Resolving a question clears the fog immediately ahead of it. Step 8 is where you collect what just became visible.
**The test is whether you can state the question precisely NOW — not whether you can answer it.** A question you cannot act on for weeks still gets a file, with `Blocked by:` filled in. A question you could answer this minute but cannot phrase without hand-waving stays fog.
### Procedure
1. **Re-read `## Not yet specified` in full.** Every bullet, every session, after every resolution. Not the ones you remember — you are assuming no memory of prior sessions, and the bullet that graduates is usually the one you forgot was there.
2. **Ask of each bullet: did the answer just make this sharp?** Can you now write a `## Question` paragraph a stranger could act on, without "we'll need to figure out" anywhere in it?
3. **If yes, write the question file NOW** — in this same session, before Work Step 9. Its `NN` is `frontier`'s `nextNN` (max existing + 1, from the directory, never reused). Fill all six schema lines, with `Blocked by:` wired in the same pass. `reconcile` then adds its `[ ]` or `[!]` row to `## Question Checklist`.
4. **DELETE the bullet from `## Not yet specified`.** Immediately, in the same edit as writing the file.
5. **If no, leave the bullet alone** — untouched, not reworded into something that merely sounds sharper.
6. **Re-check the counts.** One fog patch may graduate into three questions, or into none. Both are normal. A patch that graduates into three was written at the right coarseness; a patch that graduates into exactly one every time was probably a question all along.
### The bullet you left behind
**A fog bullet still sitting in `## Not yet specified` after its question file exists is the single most common drift in this skill.** It is quiet — nothing errors, the map still renders — and it is corrosive:
- The Clearing Gate requires `## Not yet specified` to be EMPTY. A stale bullet blocks the gate forever, so the map never clears even when every question is resolved.
- A later session reads the bullet, does not recognize the question file as the same thing under a different phrasing, and writes a duplicate. Now two files hold half a decision each.
- It violates the rule that detail lives in exactly one place.
Detection is cheap: after writing any graduated question file, re-read `## Not yet specified` top to bottom and confirm the bullet is gone. If a bullet reads like a question you have already written a file for, delete the bullet — the file always wins.
### Ruling a question out of scope mid-work
The destination fixes the scope. When a resolution reveals that a question — the one you just claimed, or another on the map — sits past the destination:
1. Decide the one-line reason. Do NOT write a `## Answer`; there is no decision, only a scope boundary.
2. Run `node "<S>/pathfinder.mjs" rule-out specs/<idea>/pathfinder <NN> --reason "<why>"`. It sets `State: out-of-scope`, adds the reason under `## Question`, marks the row `[-]` and adds the linked line to `## Out of scope`.
3. Handle its `stranded` list: questions whose `Blocked by:` names it. A blocker that is `out-of-scope` will never be `resolved`, so the dependent is permanently blocked. Re-frame its `## Question` to drop the dependency, or rule it out too. Never leave it sitting.
Out-of-scope work never graduates back. The frontier stops at the destination. It returns only if the destination is redrawn, and then as a fresh effort with a fresh map.
```markdown
## Out of scope
- [Slack connector](./questions/08-slack-connector.md) — the destination names
MS Teams only; Slack is a separate effort with its own compliance posture.
```
### Re-framing or deleting a question the answer invalidated
An answer can also break questions that already exist. Three cases:
| What happened | Do this |
|---|---|
| The question still matters but is asked wrong | **Re-frame.** Rewrite `## Question` in place. Keep `NN`, keep the file, keep the links. Add one line noting which answer forced the re-frame. |
| The question no longer exists — the answer subsumed it | **Delete the file, then run `reconcile`** (it drops the map row). Add one line to the answering question's `## Answer` consequences saying what it absorbed. Do not renumber anything. |
| The question is now two questions | **Re-frame the original to the narrower half; write a new file at max+1 for the other.** Wire `Blocked by:` between them if one gates the other. |
Never leave a question standing that you know is wrong on the theory that a later session will notice. It will not — it assumes no memory, and a well-formed `## Question` reads as intentional.
Two hard constraints on all three cases: **`NN` is never reused and never renumbered** — links and `Blocked by:` lines would rot silently. And run `reconcile` straight after any re-frame or deletion, so the index never disagrees with the files (it drops the row of a deleted file and adds one for a new file).
@@ -0,0 +1,86 @@
# Trail Footer
> Loaded at the top of the skill. Read once; it applies to EVERY response in both modes. Defines the map visual and the pathed resume command that close each turn.
>
> **Backend note.** The glyphs, the two forms, and the discipline are identical either way. On `**Backend:** github` the inputs come from the sub-issue query rather than the checklist, and Form A's command is the map issue URL — see `github-issues.md`. `pathfinder.mjs` reads local files only, so on `github` do each script step below except `lint` by hand with `gh`, as `github-issues.md` describes.
The trail is how a human with no memory of the last session sees, at a glance, how far the map has come and what is left — and copies the exact command to resume without hunting for a path that lives in gitignored `specs/`. It closes every response once a map exists.
It is **presentation only**, rebuilt fresh from the files every response and never written to disk. The script draws it; you choose its form.
```
node "<S>/pathfinder.mjs" trail specs/<idea>/pathfinder --form A|B|C [--item "<probe>" ...] --text
```
`--text` prints the footer itself: paste it verbatim as the last thing in the response. Use it for Forms A and B. For Form C leave `--text` off: the JSON carries the same `text` plus the `menu` the question tool needs. Run `reconcile` first in a session (Work Step 2) so the files it reads are settled.
---
## When it renders, and which form
The **turn type** picks the form, never the map's status. This is the one decision the script cannot make for you.
| Situation | Footer |
|---|---|
| `map.md` exists and the turn **ends the session** — Session End, a cleared map, a fully blocked frontier, a parked grill | **Form A**: the trail, then the resume command |
| `map.md` exists and the turn **asks the human something** — a probe batch, a sketch put up for reaction, a `legwork · HITL` checklist, the Chart Step 2 destination grill, the Chart Step 4 frontier grill | **Form B**: the trail, then the waiting notice. Never a resume command |
| `map.md` exists and the turn **offers the menu after a recorded decision** — the continue-or-stop fork of the Work cadence | **Form C**: the trail, the `OR START FRESH` command, then the question tool |
| Intent Gate (Step 0), the no-fog off-ramp, or any route-and-stop before Step 5 | **None** — no map on disk yet, and no path to resume |
One idea per session, so there is only ever one trail.
### What the script draws (so you can read it, not build it)
```
🧭 audit-log-export · Working · 2/6 cleared
START ●━●━◉··○··○··⊘··⊝ ····⚑
● done · ◉ here · ○ open · ⊘ blocked · ⊝ out of scope · ⚑ destination · ~4 fog
1 Codebase context ✔ · 2 Export format ✔ · 3 Row-count ceiling ◀ here · 4 Export authorization
5 Testing posture · 6 Delivery channel (blocked:3) · 7 SIEM push (out of scope)
Confidence: solid, but Risk is borderline.
WAITING ON YOU · answer here, in this conversation:
Q2 Ceiling behaviour past the cap (re-ask) · Q3 Who sees the truncation warning
```
- The heading counts resolved questions over every question except out-of-scope ones.
- One glyph per question in `NN` order: `●` resolved, `◉` claimed, `○` open, `⊘` blocked, `⊝` out of scope. A solid `━` leads into a walked stop (`●`, `◉`); `··` leads into one still ahead. `····⚑` stands for the fog; with no fog the path ends `━⚑` once every stop is walked (or ruled out) and `··⚑` while any is still ahead. A cleared map ends `━⚑ arrived` and drops both legend lines.
- The named legend numbers stops by position, and `(blocked:3)` points at stop 3.
- The confidence line is the map's four internal scores reduced to plain English: `solid` (all ≥ 20), `solid, but <Dimension> is borderline` (all ≥ 18, some at 18-19), `not yet — <Dimension(s)> still need work` (any < 18). The raw numbers never reach the human. No `**Confidence:**` line on the map, no confidence line in the trail.
---
## Form A — the turn ends the session
The command follows the map's status: `/plan2code-0-pathfinder specs/<idea>/pathfinder --web` while `Charting` / `Working`; `/plan2code-1-plan --web` once `Cleared`; and `/plan2code-init` first when `## Ground rules` records `AGENTS.md` absent and it still is. The path is the exact argument Auto-Discovery resolves an idea from, so the human pastes it back with zero edits.
## Form B — the turn asks the human something
Pass every outstanding item as `--item`, at its stable number (`--item "Q1 Duration model"`), so a partial reply is cheap and a dropped probe is visible to both of you. The script refuses more than three: the three-probe cap.
**Emit no resume command on a Form B turn.** There is nothing to resume — the session is alive and holding a claim. A resume command here reads as *we're done*, and the human either walks away mid-decision or burns the next turn asking what you meant. This is the single most common way the footer misfires, and it costs the batch the round trips batching was introduced to save.
One exception: `grilling.md`'s unreachable-human procedure. Parking a mid-grill question and stopping IS a session end — use Form A, and say in the body that the question is parked mid-grill with its batch outstanding.
## Form C — the menu after a recorded decision
Used ONLY on the Work cadence's fork-menu turn (resolve.md §Cadence), immediately after `resolve` recorded an `## Answer`. The script refuses it while a question is claimed or once the map is `Cleared`. Pass `--resolved-this-session <N>`, and `--heavy` when the decision just recorded was heavy, contested, or `Locked: yes`.
It prints the stop branch, labeled `OR START FRESH · new conversation, paste:`, and returns `menu`: the header, the question text, up to three takeable questions plus `Start fresh`, and which option to mark `RECOMMENDED` (a question early in the session; `Start fresh` after ~3 resolved or a heavy decision, with the reason in the question text, "to keep me sharp").
Then invoke the structured question tool with that menu. Put each question's gist or newly-unblocked note (from `resolve`'s `unblocked`) in its description. The automatic `Other` path lets the human name a frontier question not shown (`menu.more`).
Rules for Form C:
- **The command label is mandatory and exact.** It makes the command the stop branch, not a farewell. Never render a bare command on a menu turn.
- **The choice itself always uses the structured tool.** Never print `NEXT UP` as a prose menu when the tool exists.
- **Form C never replaces Form B.** A turn that asks probes, shows a sketch, or hands over a checklist is Form B, no command. Form C fires only between decisions, when nothing is claimed.
- If the human walks away mid-menu, nothing is stranded: no claim is open, and the command to come back is already on screen.
---
## Discipline
- **Never print a bare resume command on a turn that asks a question.** Before you choose the form, ask whether the response above it ends with something for them to answer: mid-question (probes, sketch, checklist) → Form B, no exceptions but the parked-grill one; between decisions at the fork-menu → Form C.
- **Draw it, never type it.** A hand-built trail drifts from the files; the script reads them every time. If the trail disagrees with the checklist above it, run `reconcile`, but only at session start or while nothing is claimed: `reconcile` resets a claim with no answer to `open`, so mid-question it would drop your own live claim. With a claim open, finish that question instead; `resolve` rebuilds the checklist, and the next trail is drawn from it. On `github` (the script cannot read the tracker), build it by hand from the sub-issue query, following "What the script draws" exactly.
- **One trail.** Never render two ideas' trails, and never invent a stop the map does not list.
@@ -0,0 +1,484 @@
# Build Sessions on the Web Console
For `/plan2code-3-implement`, `/plan2code-3-implement-review` and
`/plan2code-quick-task`. Read `console.md` first: everything there still
applies (the wait loop, one post per send, `finish` before `stop`, the house
rules). This file adds what a build needs, because a build is mostly you
working, and the page has to be told so.
**The console changes where you ask, not what you build.** Every rule of the
workflow stands: the same specs, the same checkbox writes to disk after every
task, the same sign-off wording, the same "no git commands". The files under
`specs/` and the code are the record. The page is a window onto them.
---
## Opening
Open first, before reading the spec — the page shows a starting screen while
you work out the phase:
```
node "<CONSOLE>" open --workflow implement
```
| Workflow you are running | `--workflow` | `--spec` |
| --- | --- | --- |
| `/plan2code-3-implement` | `implement` | `specs/<feature>` |
| `/plan2code-3-implement-review` | `implement-review` | `specs/<feature>` |
| `/plan2code-quick-task` | `quick-task` | leave it out |
`--spec` is what the page builds a resume command from if you ever cannot
answer a stop yourself. Pass it to `open` when you already know the spec;
otherwise post `"specDir": "specs/<feature>"` and a real `title` ("<feature>:
Phase 3") in your first update, once the phase is chosen, or with the phase
question as its first item if there is one to ask. `--file` works on `open`
too, but opening bare and posting the task list as your first update gets
them the page while you are still composing.
**Launched from the dashboard?** Then the session is already open and the
person is watching it — `open --resume <sid> --no-open --workflow <yours>`
(plus `--spec` once known) instead of a fresh `open`, exactly as console.md →
Launches says. Everything below is unchanged; the only difference is whose
session it was a minute ago.
The first payload for a phase you are about to build:
```jsonc
{
"headline": { "stage": "Phase 3: API layer", "cleared": 0, "total": 9 },
"docs": [
{ "id": "phase", "title": "Phase 3 tasks", "version": 1, "blocks": [
{ "id": "tasks", "state": "settled", "md": "- [ ] **Task 3.1:** ...\n- [ ] **Task 3.2:** ..." }
]},
{ "id": "phasefile", "title": "phase-3.md", "version": 1, "blocks": [
{ "id": "body", "state": "settled", "md": "...the whole phase-3.md file, verbatim..." }
]}
],
"agent": { "status": "working", "activity": "Checking the prerequisites", "quietMinutes": 15 }
}
```
`headline.cleared` / `total` are tasks done out of tasks in the phase: the
progress bar is the first thing someone watching a build looks at. The session
meter counts the build's tasks from `cleared` as it rises, so keep it current,
and start every phase (a second one in the same run too) at `"cleared": 0`.
`phasefile` is the whole `phase-X.md`, verbatim from disk, as its own tab —
the person can read the prerequisites and task specs while you build. Re-post
it whenever the file changes, which is every task: the checkbox write that
ticks the list above ticks the file too, and the Phase Completion Summary
lands in it at the end. It renders as markdown, so the file's checkboxes show
as ticks exactly as they stand on disk.
---
## While you build
Post after every task, in the same breath as writing its checkbox to disk.
Write the patch file into the session's own folder (the `session` path `open`
printed), never into the project or into `~/.plan2code/console/` directly:
```jsonc
{
"headline": { "cleared": 4 },
"docs": [
{ "id": "phase", "version": 5, "blocks": [ { "id": "tasks", "state": "settled", "md": "...the whole list, 4 ticked..." } ] },
{ "id": "phasefile", "version": 5, "blocks": [ { "id": "body", "state": "settled", "md": "...phase-3.md as it now stands on disk..." } ] }
],
"agent": { "activity": "Task 5 of 9: the rate limiter" }
}
```
- **`activity`** is one plain sentence saying what you are doing now. The page
shows it beside Planny and in place of "questions coming". No internal
vocabulary, no loop completion markers. Number tasks by their position in
the phase, of the phase's task count (`Task 5 of 9`), not by the file's
`5.5`-style labels: the phase is already in the headline. The page rewrites
the old form as a safety net.
- **`quietMinutes`** is how long you might go without posting (1 to 60). The
page calls a working agent stuck after two minutes of silence unless you say
otherwise, and a build task can honestly take twenty. Set it to the longest
gap you expect, usually 10 to 30. Every post is also a heartbeat. **Clear it
(`"quietMinutes": null`) whenever you go back to waiting**: it only counts
while you are `working`, but leaving it set invites a stale value later.
- `blocks` inside a doc are replaced whole, so post the whole task list each
time. It is small. The same goes for `phasefile`: re-read `phase-X.md` and
post it fresh, because the checkbox write just changed it.
- If the task list grows mid-build, raise `headline.total` in the same post;
the page never shows more done than the total, but the bar reads truer when
you keep it current.
- Do not call `wait` while building. There is nothing to collect, and the page
holds anything the person sends until you next look.
### Quick questions between tasks
The person can ask a Quick question on the Ask tab at any point in a build
(console.md → Quick questions). After each task's progress post, check for
one before starting the next task:
```
node "<CONSOLE>" chat --session <sid>
```
| Exit | What you do |
| --- | --- |
| `0` | Answer everything waiting, under console.md → Quick questions, in one post, then start the next task. A `workspace` list in the output is a change to the session's folders (console.md → Workspace): take it into account from the next task on; it needs no reply. |
| `10` | Nothing waiting. Carry on. |
| `20` | The server is gone: `open --resume <sid>` as usual, then carry on. |
This is an instant check, not a wait: it never sleeps and never touches card
sends. It does not replace `wait` at a gate, and no build calls `wait`
mid-build for cards. An approved chat edit is made here, between tasks, never
in the middle of one.
A blocker (`[!]`), a `SPEC NOTE` and a verified or assumed prerequisite are
not questions. Put them in the phase doc as they happen, and in the report.
### When the workflow asks something
Publish it, set `"agent": { "status": "waiting", "quietMinutes": null }` in
the SAME post, and go into the wait loop. When the answer lands, set `working`
again and carry on.
A question stops the build. Never leave yourself `working` with a question
open, and never "carry on while you decide": the page reads `working` as "not
your turn", and the person is left looking at a question they cannot answer.
If there is other work the answer does not touch, finish it before you ask,
then ask and wait.
| Moment | `kind` | Notes |
| --- | --- | --- |
| Several phases can be picked | `choice` | One option per phase, `detail` = status and task count, `token` = the phase number. Put the "run another agent on a different phase" tip in `body`. |
| A phase is already in progress | `confirm` | "Resume Phase 3?" `yes` / `no`. |
| `SPEC CONFLICT` | `choice` or `text` | The conflict in `body`, the readings as options where there are clear ones. Never guess on architecture. |
| Test failures | `choice` | `1` Fix now, `2` Document and proceed, `3` Investigate first. |
| A small phase, continue with the next? | `confirm` | |
| A spec gap only they can close | `text` | |
| Something learned for AGENTS.md | `confirm` | The proposed lines in `consequences`. |
Titles stay plain: "Which phase next", "Tests are failing", "Keep going?".
---
## Sign-off
Put the Completion Report on the page as a doc, then ask for approval as a
`review` item pointing at it. The review offer rides on this card, as a verdict
ahead of the approval buttons — not on the finish after it, where it could
only ever be offered once nothing could come of it:
```jsonc
{
"docs": [{ "id": "report", "title": "Completion report", "version": 1, "blocks": [
{ "id": "summary", "state": "settled", "md": "## Summary\n\n..." },
{ "id": "files", "state": "settled", "md": "## Files\n\n..." },
{ "id": "verify", "state": "settled", "md": "## How to check it\n\n..." }
]}],
"items": [{
"id": "signoff", "kind": "review", "title": "Approve this phase", "doc": "report",
"body": "Phase 3 is built. Read the report, try the checks, then approve it, review the code first, or say what to change.",
"verdicts": [
{ "id": "review", "label": "Review the code first" },
{ "id": "approve", "label": "Approve this phase" },
{ "id": "changes", "label": "I want changes" }
],
"token": { "review": "review", "approve": "approved", "changes": "needs changes" },
"nothingYet": "Nothing is marked complete until you approve.",
"gate": { "id": "signoff", "authoritative": true },
"fallbackText": "Reply \"approved\" to mark this phase complete, \"review\" to run the code review first, or describe any issues."
}],
"agent": { "status": "waiting", "quietMinutes": null }
}
```
`reply` carries `Approve this phase: approved`, which is the terminal's
`approved`. If the note that came with an approval asks for a change, treat it
as "describe any issues": do the change first and ask again. `changes` means
exactly what it says: address it, bump the report's version, and reopen the
item (`"status": "reopened"`) with a reply in its `thread`. `review` is the
person asking for the code review before they approve anything — see "The
review, mid-session" below.
`gate` marks the one approval that counts. The server refuses a second open
authoritative gate, which is what keeps a composed workflow from asking for
sign-off twice.
---
## After approval: the finish
Do everything the workflow does on approval (the checkboxes, the summary, the
upcoming phases, the commit command) and print it in the terminal as usual.
Then post the finish, wait for the dashboard button as console.md → Finishing
says, and `stop`:
```jsonc
{
"items": [{ "id": "signoff", "status": "answered", "answer": { "verdict": "approve" } }],
"run": { "event": "implement-phase", "id": "phase-3", "tasks": 9 },
"finish": {
"headline": "Phase 3 is done",
"body": "9 of 9 tasks, all on disk.\n\n| Phase | Tasks | Goal |\n| --- | --- | --- |\n| Phase 4: ... | 6 | ... |\n\nCommit it with:\n\n```\ngit add -A && git commit -m \"...\" -m \"AI Assisted\"\n```",
"command": "/plan2code-3-implement specs/lunch-vote/overview.md",
"doc": "report",
"dashboard": true
},
"agent": { "status": "waiting" }
}
```
`finish.command` is the next pipeline step, the same one the terminal names:
the next `/plan2code-3-implement`, or `/plan2code-4-finalize` after the last
phase. An `implement` finish never carries `review`: the offer was on the
sign-off card, and approving past it was the answer. `implement-review` ends
the same way; that phase was reviewed already. Only a finished `quick-task`
still offers the review on its finish — see below.
`run` gives the approved phase its final score on the session meter (console.md →
Posting an update → Session meter), in the same post that records the approval;
it takes over what the build already counted from `cleared`, never adds to it:
`implement-phase` for Implement, `implement-review-phase` for Implement +
Review, id `phase-N`, and `tasks`: how many of the phase's tasks were
completed (`[x]`, not blocked `[!]`). The phase scores by its size, so a
twelve-task phase weighs four times a three-task one; leave `tasks` out and
every phase weighs the same. Quick task and a standalone Review are counted when
their session opens, so they post no `run` for that; a review run on the page
after a build reports its own (below).
---
## The review, mid-session
The same focused review runs from either of two doors:
- **`implement`:** the `review` verdict on the sign-off card, before anything
is approved. A note sent with it is theirs — read it as focus for the
review. Settle the card with it in the same post that goes back to work:
```jsonc
{
"items": [{ "id": "signoff", "status": "answered", "answer": { "verdict": "review" } }],
"headline": { "stage": "Reviewing Phase 3", "cleared": 0, "total": 0 },
"run": { "event": "review", "id": "review" },
"agent": { "status": "working", "activity": "Reviewing the 7 files this phase changed", "quietMinutes": 15 }
}
```
- **`quick-task`:** the **Review it now** button on the finished screen, which
`finish.review` puts there. It arrives through the wait loop as an action
`{ "i": "__review", "type": "review" }`, like any send. Take it up with the
same post minus the `items` line, plus `"finish": null` — that takes the
session off its ending and back to work.
Either way, the post that takes the review up carries `"run": { "event":
"review", "id": "review" }`: the review runs inside this skill's session, so no
launch scores it on the session meter. Never resume the session under
`--workflow review` for it as well, or it counts twice.
A finish carrying `review` keeps the session alive for the answer, so a
`quick-task` that just posted one keeps waiting, for up to about ten minutes.
A quick-task finish carries both `review` and `"dashboard": true` and waits
for whichever comes first:
| What arrives | What you do |
| --- | --- |
| An action `{ "i": "__review", "type": "review" }` | Take the review up (below). |
| An action `{ "i": "__dashboard", "type": "dashboard" }` | Take the session back as the dashboard (console.md → Finishing). |
| An action `{ "i": "__done", "type": "done" }` | `stop` the server. The session is over. |
| `wait` exit `20` | The server is gone, not necessarily the tab. Resume it (`open --resume <sid> --no-open --workflow <yours>`) and keep waiting. |
| Nothing after about ten minutes | `stop` the server. The page then shows the terminal command for a review instead of a button that would go nowhere. |
**Read references/review.md** and run Review Mode with these overrides:
- **Scope is Focused, and already decided:** the exact change set of this
session (files created or modified by the phase or the quick task, plus the
tests and docs changed for it). Do not ask about scope. Use overview.md and
the phase file for spec compliance; they do not widen the scope.
- **No commit instructions and no next-conversation routing** mid-review. The
way back out carries both.
- The page takes the whole report, so the "top 10 of 15+" rule does not apply,
and tables go in the doc.
### The findings
The full Review Mode report as a doc, and the fix choice as one `multi` item:
```jsonc
{
"docs": [{ "id": "review", "title": "Review findings", "version": 1, "blocks": [
{ "id": "report", "state": "settled", "md": "## Overview\n\n...\n\n## Findings\n\n| # | Severity | Issue | File |\n..." }
]}],
"items": [{
"id": "fixes", "kind": "multi", "title": "Which to fix", "required": false, "doc": "review",
"body": "Tick what should be fixed now. Anything left unticked is written up as still open. Nothing changes until you send.",
"options": [
{ "k": "1", "text": "Critical: token is logged in plain text", "detail": "src/auth.ts:42. Fix: redact before logging." },
{ "k": "2", "text": "Warning: missing null check on the user lookup", "detail": "src/users.ts:88. Fix: ..." },
{ "k": "3", "text": "Suggestion: repeated query could be batched", "detail": "src/report.ts:17. Fix: ..." }
],
"presets": [
{ "label": "Critical and warnings", "ks": ["1", "2"] },
{ "label": "All of them", "ks": ["1", "2", "3"] }
],
"token": { "1": "1", "2": "2", "3": "3" },
"fallbackText": "Reply H to fix Critical + Warning findings, A to fix all findings, or S 1,3,5 to fix specific findings."
}],
"agent": { "status": "waiting", "quietMinutes": null }
}
```
`"doc": "review"` puts a **Read "Review findings" in full** button on the
card, so the report is one click from the list rather than behind a tab
nobody notices. Always carry it.
Read the reply as Review Mode's fix choice: `Which to fix: 1, 2` is `S 1,2`,
every finding ticked is `A`, and `skip` (or nothing ticked) is "fix none,
keep them as unresolved". Leave out `presets` a set would not change (no
warnings, one finding). With **zero findings** there is no item: say so in the
doc and go straight to the way back out.
### Fixing
Settle the `fixes` item in the same post that goes back to work —
`"status": "answered"` with the `answer` they sent (`{ "ks": ["1", "2"] }`,
or `{ "ks": [] }` when nothing was ticked). The server records their send on
the item, but only you decide it is settled; left open it stays on the page
as an outstanding question, beside the card you ask next. Then follow Review
Mode's Post-Fix Flow, posting `activity` per fix ("Fix 2 of 3: the null
check") and `quietMinutes` as in a build.
### The way back out
**`implement` — back to the sign-off.** The review was the preamble to an
approval that has not happened yet, so this is where it lands. Fold the review
results and any still-open findings into the report doc and bump it, then
reopen the sign-off **without the review verdict** — re-offering a review that
just ran reads as though nothing happened:
```jsonc
{
"items": [{ "id": "signoff", "status": "reopened",
"body": "Reviewed and ready for sign-off. The review found 3 issues: 2 fixed, 1 still open in the Review findings tab. The report is updated.",
"verdicts": [
{ "id": "approve", "label": "Approve this phase" },
{ "id": "changes", "label": "I want changes" }
],
"thread": [{ "who": "agent", "text": "Review done: 2 of 3 findings fixed; the rest is written up in the findings tab." }]
}],
"agent": { "status": "waiting", "quietMinutes": null }
}
```
Review Mode's own closing gate ("Reply approved to commit") is this card:
approving the phase now covers the fixes too, so there is no separate "Check
the fixes" step. On approval, continue at "After approval: the finish".
**`quick-task` — the closing gate, then the finish.** Review Mode's closing
gate ("Reply approved to commit, or request changes") becomes a `review` item
titled "Check the fixes", `doc: "review"`, verdicts `approve` ("The fixes look
right") and `changes`, token `approved` / `needs changes`. When it clears,
settle it in the same post as the new finish — `"status": "answered"` plus
`"answer": { "verdict": "approve" }` (or `"changes"`) — post the finish
**without** `review`, then `stop`. An item left open survives on the page as
a question still owed, above the ending itself:
- `headline`: "The review is done"
- `body`: findings by severity, what was fixed, what is still open, and the
commit command for the fixes, in the project's commit format.
- `command`: the same next step as the finish before it.
- `doc`: `"review"`.
---
## Implement + Review (`implement-review`)
The same build, then Stage 2 publishes the findings exactly as above (the
`review` doc and the `fixes` item). Stage 3 fixes, with no "Check the fixes"
item: Stage 4 is the only gate. The post that starts Stage 2's review carries
`"run": { "event": "review", "id": "review" }`, as above: in this workflow it
scores the review stage's 1 point on the meter as it begins, and the phase's
final `run` leaves that point out. Stage 4 is the sign-off above **without the
review verdict** — the phase was already reviewed — with the review results
and unresolved findings in the report doc, header "Reviewed and ready for
sign-off" in the item's `body`. The finish never carries `review`, and its
post carries `"run": { "event": "implement-review-phase", "id": "phase-N", "tasks": N }`
with the count of completed tasks.
---
## Quick task (`quick-task`)
- Open with the first clarifying question. Clarifying questions go out as
`text` or `choice`, at most three at a time.
- Post `"stopWarning": "A quick task saves nothing until it is built, so stopping now loses what we have worked out."`
at the start, and `"stopWarning": ""` once the build is under way.
- **Scope check:** a `choice`, "Too big for a quick task?", options
`1` Continue (lightweight format) and `2` Escalate to a full plan, with the
measured counts against the thresholds in `body`.
- **Feature name** (escalating): `text` with
`"pattern": "^[a-z0-9]+(-[a-z0-9]+)*$"` and a kebab-case placeholder.
- **The plan:** a `plan` doc (Summary, Files to change, Steps, Verify) and a
`review` item "Build this plan?" with verdicts `yes` ("Build it"), `modify`
("Change the plan"), `escalate` ("Plan it properly instead") and `abort`
("Stop here", `"danger": true`), token equal to the ids.
- **Building** after `yes`: settle the verdict (`"status": "answered"` plus
`"answer": { "verdict": "yes" }`) in the post that starts the build, and
open a **Build progress** tab in that same post, the way an Implement phase
gets its tasks tab. The plan's steps are the task list and `total` is their
number:
```jsonc
{
"stopWarning": "",
"items": [{ "id": "<the Build this plan? item's id>", "status": "answered", "answer": { "verdict": "yes" } }],
"headline": { "stage": "Building", "cleared": 0, "total": 4 },
"docs": [{ "id": "tasks", "title": "Build progress", "version": 1, "blocks": [
{ "id": "tasks", "state": "settled", "md": "- [ ] Step one\n- [ ] Step two\n- [ ] Step three\n- [ ] Step four" }
]}],
"agent": { "status": "working", "activity": "Step 1 of 4: step one", "quietMinutes": 10 }
}
```
Then post after every step, exactly as "While you build": bump the doc's
`version` and repost the whole list with that step ticked, raise
`headline.cleared`, and set `agent.activity` to the next step ("Step 2 of 4:
step two"). Never go silent for the whole build: a quick task with many steps
is the case this tab is for. Check for a Quick question between steps, and do
not call `wait` until the build is done.
- **The finish after the build:** `command` is the commit command, filled in,
in the project's commit format (AGENTS.md), with
`"where": "When you are happy with it, commit it from your terminal:"`, and
`review` on offer — the **Review it now** button, since a quick task has no
sign-off card to put it on — and `"dashboard": true`, so the finished screen
also offers **Back to the dashboard**. Both are required: a finish without
`dashboard` leaves the person on a dead end. Keep waiting for whichever
button is pressed, and everything after that is "The review, mid-session"
above.
- **Escalated / aborted:** settle the verdict the same way in the finish post —
finish with `"command": "/plan2code-1-plan"` and the
PLAN-DRAFT's path in `body`, no review; or `"command": "/plan2code-quick-task"`.
Both set `"dashboard": true`: an abort of a quick task saves nothing to
resume, so it is not a pause.
---
## Stop requests during a build
The Stop button only works on the person's turn, so it never lands mid-task.
When `__stop` arrives:
- **implement / implement-review:** Implementation Mode's abort, without the
"are you sure" (the page asked): the phase stays `[/]`, list completed and
remaining tasks, post `finish` with the resume command
(`/plan2code-3-implement specs/<feature>/overview.md`), then `stop`. An
implement abort is a pause: a headline starting "Paused", no `dashboard`,
`stop` at once.
- **quick-task:** say in `finish.body` what was worked out, and that nothing was
saved unless it was built; `"command": "/plan2code-quick-task"`.
A mid-build `__dashboard` (the top bar's triangle, console.md → Stop requests →
"Back to the dashboard, mid-workflow") lands the same way, only on the person's
turn. Save exactly as for `__stop` above (an implement phase stays `[/]`), but
post no finish and do not `stop`: resume as the dashboard instead, with one
terminal line naming what was saved and the resume command.
File diff suppressed because it is too large. Load diff
File diff suppressed because it is too large. Load diff
File diff suppressed because it is too large. Load diff
@@ -0,0 +1,265 @@
// Plan2Code Web Console - the native folder picker behind Workspace's Browse…
//
// A port of the launcher's picker (src/launcher/plan2code.js). The launcher
// is self-contained CommonJS copied alone to ~/.plan2code/bin/, and this
// console is ESM bundled into every skill, so the two cannot share a module:
// this is a copy, and a drift test in scripts/test-web-console.mjs fails the
// moment the launcher's picker changes without it. The prompt is a parameter
// here rather than the launcher's constant.
//
// Always spawned asynchronously: spawnSync would freeze the server's polling,
// chat and wait for as long as the dialog stayed open.
//
// Node built-ins only.
import fs from "node:fs";
import path from "node:path";
import { spawn } from "node:child_process";
const IS_WINDOWS = process.platform === "win32";
const IS_MAC = process.platform === "darwin";
export const WORKSPACE_PICKER_PROMPT = "Choose a folder to add to the workspace";
// Copied verbatim from src/launcher/plan2code.js; see the drift test.
export const WINDOWS_PICKER_SOURCE = `
using System;
using System.Runtime.InteropServices;
using System.Threading;
public static class Plan2CodeFolderPicker {
[ComImport, Guid("DC1C5A9C-E88A-4DDE-A5A1-60F82A20AEF7")]
private class FileOpenDialog {}
[ComImport, Guid("42F85136-DB7E-439C-85F1-E4075D135FC8"), InterfaceType(ComInterfaceType.InterfaceIsIUnknown)]
private interface IFileOpenDialog {
[PreserveSig] int Show(IntPtr owner);
void SetFileTypes(uint count, IntPtr specs);
void SetFileTypeIndex(uint index);
void GetFileTypeIndex(out uint index);
void Advise(IntPtr events, out uint cookie);
void Unadvise(uint cookie);
void SetOptions(uint options);
void GetOptions(out uint options);
void SetDefaultFolder(IShellItem item);
void SetFolder(IShellItem item);
void GetFolder(out IShellItem item);
void GetCurrentSelection(out IShellItem item);
void SetFileName([MarshalAs(UnmanagedType.LPWStr)] string name);
void GetFileName([MarshalAs(UnmanagedType.LPWStr)] out string name);
void SetTitle([MarshalAs(UnmanagedType.LPWStr)] string title);
void SetOkButtonLabel([MarshalAs(UnmanagedType.LPWStr)] string label);
void SetFileNameLabel([MarshalAs(UnmanagedType.LPWStr)] string label);
void GetResult(out IShellItem item);
}
[ComImport, Guid("43826D1E-E718-42EE-BC55-A1E261C37BFE"), InterfaceType(ComInterfaceType.InterfaceIsIUnknown)]
private interface IShellItem {
void BindToHandler(IntPtr bindContext, ref Guid handler, ref Guid iid, out IntPtr result);
void GetParent(out IShellItem parent);
void GetDisplayName(uint form, [MarshalAs(UnmanagedType.LPWStr)] out string name);
}
[DllImport("shell32.dll", CharSet = CharSet.Unicode, PreserveSig = false)]
private static extern void SHCreateItemFromParsingName(string path, IntPtr bindContext, ref Guid iid, out IShellItem item);
[DllImport("user32.dll", CharSet = CharSet.Unicode)]
private static extern IntPtr FindWindow(string className, string title);
[DllImport("user32.dll")]
private static extern bool SetWindowPos(IntPtr hWnd, IntPtr after, int x, int y, int w, int h, uint flags);
[DllImport("user32.dll")]
private static extern bool SetForegroundWindow(IntPtr hWnd);
[DllImport("user32.dll")]
private static extern int GetSystemMetrics(int index);
private const uint FOS_PICKFOLDERS = 0x20;
private const uint FOS_FORCEFILESYSTEM = 0x40;
private const uint FOS_PATHMUSTEXIST = 0x800;
private const uint SIGDN_FILESYSPATH = 0x80058000;
private const int ERROR_CANCELLED = unchecked((int)0x800704C7);
private static void Raise(string title) {
for (int i = 0; i < 100; i++) {
IntPtr hwnd = FindWindow(null, title);
if (hwnd != IntPtr.Zero) {
int w = Math.Min(1100, GetSystemMetrics(0) - 80);
int h = Math.Min(750, GetSystemMetrics(1) - 80);
int x = (GetSystemMetrics(0) - w) / 2 + 40;
int y = (GetSystemMetrics(1) - h) / 2 + 40;
SetWindowPos(hwnd, new IntPtr(-1), x, y, w, h, 0x40);
SetForegroundWindow(hwnd);
return;
}
Thread.Sleep(50);
}
}
public static string Pick(IntPtr owner, string title, string start) {
IFileOpenDialog dialog = (IFileOpenDialog)new FileOpenDialog();
try {
uint options;
dialog.GetOptions(out options);
dialog.SetOptions(options | FOS_PICKFOLDERS | FOS_FORCEFILESYSTEM | FOS_PATHMUSTEXIST);
dialog.SetTitle(title);
if (!string.IsNullOrEmpty(start)) {
try {
Guid iid = typeof(IShellItem).GUID;
IShellItem folder;
SHCreateItemFromParsingName(start, IntPtr.Zero, ref iid, out folder);
dialog.SetFolder(folder);
} catch (Exception) {}
}
Thread raiser = new Thread(delegate() { Raise(title); });
raiser.IsBackground = true;
raiser.Start();
int hr = dialog.Show(owner);
if (hr == ERROR_CANCELLED) return null;
Marshal.ThrowExceptionForHR(hr);
IShellItem result;
dialog.GetResult(out result);
string path;
result.GetDisplayName(SIGDN_FILESYSPATH, out path);
return path;
} finally {
Marshal.ReleaseComObject(dialog);
}
}
}
`;
// Resolve a command on PATH the way the shell would, including PATHEXT on Windows.
function findOnPath(command) {
const dirs = (process.env.PATH || "").split(path.delimiter).filter(Boolean);
const extensions = IS_WINDOWS ? (process.env.PATHEXT || ".EXE;.CMD;.BAT;.COM").split(";").filter(Boolean) : [""];
for (const dir of dirs) {
for (const ext of extensions) {
const candidate = path.join(dir.replace(/^"|"$/g, ""), command + ext);
try {
if (fs.statSync(candidate).isFile()) return candidate;
} catch {}
}
}
return null;
}
/**
* The native folder picker for this OS, as a command to run: it prints the
* chosen folder, or nothing when the person cancels. Null when this OS has no
* picker we can drive.
*/
export function pickerCommand(start, prompt) {
if (IS_WINDOWS) {
const script = [
"$ErrorActionPreference = 'Stop'",
"try {",
" [Console]::OutputEncoding = [Text.Encoding]::UTF8",
" Add-Type -AssemblyName System.Windows.Forms",
" [System.Windows.Forms.Application]::EnableVisualStyles()",
// A TopMost owner keeps the dialog in front of the console window that launched it.
" $owner = New-Object System.Windows.Forms.Form -Property @{ TopMost = $true; ShowInTaskbar = $false }",
" try {",
" Add-Type -TypeDefinition $env:PLAN2CODE_PICKER_SOURCE",
" $picked = [Plan2CodeFolderPicker]::Pick($owner.Handle, $env:PLAN2CODE_PICKER_PROMPT, $env:PLAN2CODE_PICKER_START)",
" } catch {",
// Add-Type is blocked under Constrained Language Mode and some AppLocker policies; the
// old tree dialog still works there.
" $dialog = New-Object System.Windows.Forms.FolderBrowserDialog",
" $dialog.Description = $env:PLAN2CODE_PICKER_PROMPT",
" $dialog.SelectedPath = $env:PLAN2CODE_PICKER_START",
" $picked = if ($dialog.ShowDialog($owner) -eq [System.Windows.Forms.DialogResult]::OK) { $dialog.SelectedPath } else { $null }",
" }",
" if ($picked) { $picked }",
"} catch { exit 3 }",
].join("\n");
return {
command: "powershell.exe",
args: [
"-NoProfile",
"-NonInteractive",
"-STA",
"-ExecutionPolicy",
"Bypass",
"-EncodedCommand",
Buffer.from(script, "utf16le").toString("base64"),
],
env: {
PLAN2CODE_PICKER_PROMPT: prompt,
PLAN2CODE_PICKER_START: start,
PLAN2CODE_PICKER_SOURCE: WINDOWS_PICKER_SOURCE,
},
};
}
if (IS_MAC) {
return {
command: "osascript",
args: [
"-e",
"on run argv",
"-e",
`POSIX path of (choose folder with prompt "${prompt}" default location (POSIX file (item 1 of argv)))`,
"-e",
"end run",
start,
],
};
}
if (!process.env.DISPLAY && !process.env.WAYLAND_DISPLAY) return null;
if (findOnPath("zenity")) {
return {
command: "zenity",
args: ["--file-selection", "--directory", `--title=${prompt}`, `--filename=${start}${path.sep}`],
};
}
if (findOnPath("kdialog")) {
return { command: "kdialog", args: ["--title", prompt, "--getexistingdirectory", start] };
}
return null;
}
/**
* Show the picker and settle on what the person did: `{ path }` for a pick,
* `{ cancelled: true }` when they closed it, `{ error }` when the picker
* itself failed or this OS has none. Pickers exit 0 or 1 with nothing printed
* when cancelled.
*/
export function pickFolder(start, prompt, { spawnImpl = spawn } = {}) {
const picker = pickerCommand(start, prompt);
if (!picker) return Promise.resolve({ error: "picker-failed" });
return new Promise((resolve) => {
let settled = false;
const settle = (result) => {
if (settled) return;
settled = true;
resolve(result);
};
let child;
try {
child = spawnImpl(picker.command, picker.args, {
env: { ...process.env, ...(picker.env || {}) },
// stderr is ignored, not piped: a pipe nobody reads fills up and
// blocks a chatty picker (GTK warnings) before it can exit.
stdio: ["ignore", "pipe", "ignore"],
windowsHide: true,
});
} catch {
return settle({ error: "picker-failed" });
}
let stdout = "";
if (child.stdout) {
child.stdout.setEncoding?.("utf8");
child.stdout.on("data", (chunk) => (stdout += chunk));
}
child.on("error", () => settle({ error: "picker-failed" }));
child.on("close", (code) => {
const folder = stdout.trim();
if (code === 0 && folder) return settle({ path: path.resolve(folder) });
if (!folder && (code === 0 || code === 1)) return settle({ cancelled: true });
settle({ error: "picker-failed" });
});
});
}
File diff suppressed because it is too large. Load diff
File diff suppressed because it is too large. Load diff
File diff suppressed because it is too large. Load diff
@@ -0,0 +1,138 @@
// Planny's wake-up on a freshly opened dashboard: he sits powered down centre
// stage until the first click or key, gets nudged awake (he rocks side to side
// for one snore), then about five seconds of him powering on, then a glide
// into the rail as the menu comes in.
//
// The beats before the glide are pure CSS (keyframes on delays under
// `.wake.booting`, see app.css). The glide is the one part CSS cannot know on
// its own, because it ends wherever the rail's Planny sits right now, so this
// measures it once and hands CSS a single transform through --wake-glide.
//
// No DOM access at module scope, so `node --test` can import it.
// When the glide starts, and when the run is over at the latest. The CSS
// delays are the beats before GLIDE_AT_MS; the glide itself is 0.6 s.
export const GLIDE_AT_MS = 4300;
export const DONE_AT_MS = 5000;
// Plays the wake-up and resolves `{ skipped }` once the dashboard is fully
// awake. He waits powered down until the first click or key anywhere, which
// nudges him (`.stirring`) and calls `stir()` -- the page's cue to play one
// full snore, since a browser only lets a page make noise after a real click
// or keypress. The boot starts once the promise `stir()` returns settles (at
// once without one), and calls `onStart()`, the page's cue for the boot-up
// sound. `onGlide({ instant })` is the page's cue to bring the menu in: called
// at the glide, or with `instant: true` if the run ends before one (a skip).
// From the nudge on, clicks and keys are swallowed, so the run always plays
// out to the end with its sound; only `skipEl` (the muted Skip button, or the
// Escape key, for anyone without a mouse) and `signal` aborting (the page
// leaving the dashboard, dormant or mid-run) cut it short, as a skip.
// `onSkip` is the page's cue to silence the intro sounds
// once a skip lands. Under reduced motion nothing is shown and it resolves at
// once.
export function playWake({ stage, bot, target, skipEl, stir, onStart, onGlide, onSkip, signal }) {
return new Promise((resolve) => {
if (matchMedia("(prefers-reduced-motion: reduce)").matches || signal?.aborted) {
resolve({ skipped: false });
return;
}
const timers = [];
let glided = false;
let over = false;
const onGlideEnd = (e) => {
if (e.target === bot && e.propertyName === "transform") end(false);
};
const skip = () => end(true);
// The Skip button's events must reach it: the document handlers below run
// in the capture phase, ahead of the button's own click.
const hitsSkip = (e) => Boolean(skipEl && e.target instanceof Node && skipEl.contains(e.target));
const onSkipClick = () => end(true);
// Browser and OS shortcuts (reload, tab switch, devtools) still go through;
// Escape skips, so the run can be left by keyboard too.
const swallow = (e) => {
if (hitsSkip(e)) return;
if (e.key === "Escape") {
onSkipClick();
return;
}
if (e.ctrlKey || e.metaKey || e.altKey) return;
e.preventDefault();
e.stopPropagation();
};
function glide() {
glided = true;
// Measured untransformed, as the sketch's glideToSlot() does: the
// transform-origin is the drawing's top-left, so the move is the gap
// between the two top-left corners and the scale is the width ratio.
// The target's own box, not its bounding rect: the rail's Planny is
// usually tilted mid-flight, and a tilted box measures wider than he
// is. Centred on the rect, which the tilt barely moves.
const from = bot.getBoundingClientRect();
const rect = target.getBoundingClientRect();
const w = target.clientWidth || rect.width;
const h = target.clientHeight || rect.height;
const left = rect.left + (rect.width - w) / 2;
const top = rect.top + (rect.height - h) / 2;
const scale = from.width ? w / from.width : 1;
bot.style.setProperty("--wake-glide", `translate(${left - from.left}px, ${top - from.top}px) scale(${scale})`);
bot.addEventListener("transitionend", onGlideEnd);
stage.classList.add("gliding");
onGlide({ instant: false });
}
function end(skipped) {
if (over) return;
over = true;
for (const t of timers) clearTimeout(t);
document.removeEventListener("pointerdown", start, true);
document.removeEventListener("keydown", start, true);
document.removeEventListener("pointerdown", swallow, true);
document.removeEventListener("keydown", swallow, true);
bot.removeEventListener("transitionend", onGlideEnd);
signal?.removeEventListener("abort", skip);
skipEl?.removeEventListener("click", onSkipClick);
if (skipped) onSkip?.();
stage.hidden = true;
stage.classList.remove("dormant", "stirring", "booting", "gliding");
bot.style.removeProperty("--wake-glide");
if (!glided) onGlide({ instant: true });
resolve({ skipped });
}
// The waking click is the only input the run takes: from here on every
// click and key is swallowed until it ends.
function start(e) {
if (hitsSkip(e)) return;
if (e.key === "Escape") {
onSkipClick();
return;
}
e.preventDefault();
document.removeEventListener("pointerdown", start, true);
document.removeEventListener("keydown", start, true);
stage.classList.replace("dormant", "stirring");
document.addEventListener("pointerdown", swallow, true);
document.addEventListener("keydown", swallow, true);
Promise.resolve(stir?.()).then(boot);
}
function boot() {
if (over) return;
stage.classList.replace("stirring", "booting");
onStart?.();
timers.push(setTimeout(glide, GLIDE_AT_MS));
// The glide's transitionend normally ends the run a touch earlier; this
// is the fallback for a transition that never fires (a hidden tab).
timers.push(setTimeout(() => end(false), DONE_AT_MS));
}
document.addEventListener("pointerdown", start, true);
document.addEventListener("keydown", start, true);
skipEl?.addEventListener("click", onSkipClick);
signal?.addEventListener("abort", skip);
stage.hidden = false;
stage.classList.add("dormant");
});
}
@@ -0,0 +1,226 @@
// Plan2Code Web Console - the Quick question chat's shared rules.
//
// Imported by the server (its guards) and by the page (its composer), so the
// two can never disagree about a limit. No DOM access at module scope, and no
// imports but answers.js: the server and `node --test` import this file too.
import { MAX_ATTACHMENTS, STALE_MS, quietLimitMs } from "./answers.js";
// Typed messages per conversation. Approve / Decline and resets never count.
export const CHAT_LIMIT = 10;
export const CHAT_MAX_CHARS = 4000;
// Images and documents together. The old name stays for existing imports.
export const CHAT_MAX_ATTACHMENTS = MAX_ATTACHMENTS;
export const CHAT_MAX_IMAGES = CHAT_MAX_ATTACHMENTS;
export const CHAT_LIMIT_TEXT = "That's 10 questions in this conversation — start a new one to keep going.";
export const CHAT_OFFLINE_TEXT = "No agent available — wait for one to become available (your terminal might be working on a task or waiting a permission approval)";
export const CHAT_LAST_WAIT_TEXT = "That was the last question in this conversation — its answer will land here.";
export const CHAT_WAITING_TEXT = "Plan2Code will answer at its next check-in";
export const CHAT_WORKING_TEXT = "Plan2Code is working on the answer…";
const EMPTY_CHAT = { conversation: 0, messages: [], typed: 0, limit: CHAT_LIMIT, offline: false, pickedUp: 0 };
const chatOf = (frameChat) => ({ ...EMPTY_CHAT, ...(frameChat || {}) });
// Person messages in the frame, each marked with whether an agent reply comes
// after it. The server puts a reply straight after the entry it answers, so a
// reply further down also covers the follow-ups sent before it.
function answeredFlags(messages) {
const flags = new Map();
let replyLater = false;
for (let i = messages.length - 1; i >= 0; i--) {
const m = messages[i];
if (m.from === "agent") replyLater = true;
else if (m.kind === "message") flags.set(m.seq, replyLater);
}
return flags;
}
/**
* What the Ask pane draws, in order: the conversation from the frame, then
* any send of this page's the server has not echoed yet.
*/
export function chatView(frameChat, ui = {}) {
const chat = chatOf(frameChat);
const messages = Array.isArray(chat.messages) ? chat.messages : [];
const answered = answeredFlags(messages);
const decisions = new Map(
messages.filter((m) => m.from === "person" && m.kind === "decision").map((m) => [m.re, m.decision])
);
const activity = typeof ui.agentActivity === "string" ? ui.agentActivity.trim() : "";
// The last seq the agent collected: an unanswered message at or below it is
// in the agent's hands, not still waiting for a check-in.
const pickedUp = Number.isInteger(chat.pickedUp) ? chat.pickedUp : 0;
const rows = [];
for (const m of messages) {
if (m.from === "agent") {
rows.push({
kind: "agent",
id: m.id,
md: m.md,
proposal: m.proposal || null,
changed: m.changed || null,
decided: decisions.get(m.id) || null,
});
} else if (m.kind === "decision") {
rows.push({ kind: "decision", decision: m.decision, text: m.text || "", re: m.re });
} else if (m.kind === "message") {
const row = {
kind: "person",
seq: m.seq,
text: m.text,
about: m.about || null,
images: m.images || [],
files: m.files || [],
};
if (!answered.get(m.seq)) {
row.waiting = true;
row.working = m.seq <= pickedUp;
row.waitingText = row.working ? CHAT_WORKING_TEXT : CHAT_WAITING_TEXT;
if (activity) row.activity = activity;
}
rows.push(row);
}
}
// A pending send is echoed once the server has it: by the seq its POST came
// back with, or, while that response is still on its way, by its text
// against a server message no other pending send has claimed.
const serverPerson = messages.filter((m) => m.from === "person" && m.kind === "message");
const serverSeqs = new Set(serverPerson.map((m) => m.seq));
const claimed = new Set((ui.pendingSends || []).filter((p) => p.seq).map((p) => p.seq));
for (const p of ui.pendingSends || []) {
if (p.seq) {
if (serverSeqs.has(p.seq)) continue;
} else {
const echo = serverPerson.find((m) => m.text === p.text && !claimed.has(m.seq));
if (echo) {
claimed.add(echo.seq);
continue;
}
}
rows.push({
kind: "person",
pending: true,
tempId: p.tempId,
text: p.text,
about: p.about || null,
images: p.images || [],
files: p.files || [],
failed: Boolean(p.failed),
});
}
return {
conversation: chat.conversation,
rows,
counterText: `${chat.typed} of ${chat.limit}`,
typed: chat.typed,
limit: chat.limit,
};
}
/**
* Whether the composer can take a message, and what to say when it cannot.
* Offline first, so a finished page never talks about the limit. The limit
* lands once the 10th question is answered; until then the 10th is simply
* waiting, and nothing more can be typed.
*/
export function sendState({ frameChat, gone, pendingSendCount = 0 }) {
const chat = chatOf(frameChat);
if (chat.offline || gone) return { state: "offline", text: CHAT_OFFLINE_TEXT, canType: false };
if (chat.typed + pendingSendCount >= chat.limit) {
const allAnswered = pendingSendCount === 0 && [...answeredFlags(chat.messages || []).values()].every(Boolean);
if (allAnswered) return { state: "limit", text: CHAT_LIMIT_TEXT, canType: false };
return { state: "ok", text: CHAT_LAST_WAIT_TEXT, canType: false };
}
return { state: "ok", text: "", canType: true };
}
const ABOUT_HISTORY = 3;
/**
* What a message is about, attached on purpose. Visiting a question card, doc
* section or spec only records it in `history` (the last 3 places, newest
* first, one entry per place); `about` is set only when the person picks one
* from + Add context, and × or a send clears it.
*
* The tracker is `{ history, about }`; each entry and `about` are
* `{ kind, id, label }` (`about` may be null), in the exact shape POST /chat
* takes. An old `{ about, frozen, dropped }` tracker normalises to an empty
* history.
*/
export function nextAbout(prev, event) {
const p = prev || {};
const state = { history: Array.isArray(p.history) ? [...p.history] : [], about: p.about ?? null };
const visit = (entry) => ({
...state,
history: [entry, ...state.history.filter((h) => h.kind !== entry.kind || h.id !== entry.id)].slice(0, ABOUT_HISTORY),
});
switch (event && event.type) {
case "card":
return visit({ kind: "card", id: String(event.id), label: String(event.label || "") });
case "section":
return visit({ kind: "section", id: `${event.docId}#${event.blockId}`, label: String(event.label || "") });
case "spec":
return visit({ kind: "spec", id: String(event.dir), label: String(event.label || "") });
case "pick":
return event.about ? { ...state, about: event.about } : state;
case "drop":
case "sent":
return { ...state, about: null };
default:
return state;
}
}
// What + Add context lists: the recent places, then "This spec" unless it is
// already one of them.
export function contextOptions(state, spec) {
const options = [...((state && state.history) || [])];
if (spec && spec.dir && !options.some((o) => o.kind === "spec" && o.id === spec.dir)) {
options.push({ kind: "spec", id: spec.dir, label: "This spec" });
}
return options;
}
const newestReply = (frameChat) => {
const messages = chatOf(frameChat).messages || [];
for (let i = messages.length - 1; i >= 0; i--) if (messages[i].from === "agent") return messages[i];
return null;
};
// The Ask tab's dot: a reply landed that the person has not seen.
export function unreadDot({ lastSeenReplyId, frameChat, viewIsAsk }) {
if (viewIsAsk) return false;
const reply = newestReply(frameChat);
return Boolean(reply) && reply.id !== lastSeenReplyId;
}
export function newestReplyId(frameChat) {
const reply = newestReply(frameChat);
return reply ? reply.id : null;
}
// One chime per reply: only the moment the dot appears.
export function shouldChime(prevDot, nextDot) {
return !prevDot && Boolean(nextDot);
}
// The page's own "has drifted away" threshold for an agent that is waiting.
const ADRIFT_MS = 3 * 60 * 1000;
/**
* Whether nobody is there to answer a chat message. A finished session's agent
* is often still listening through its last wait (the review and dashboard
* buttons on the hand-off card live as long as it does), so finish alone does
* not end the chat — going quiet does. Offline once the agent has been silent
* longer than the page would tolerate before calling it stuck (working) or
* adrift (anything else), which is also the only signal left once the server
* is gone.
*/
export function chatOffline({ finish, agent, agentLastSeenMs, now }) {
const silent = now - agentLastSeenMs;
if (agent && agent.status === "working") return silent > quietLimitMs(agent, STALE_MS);
return silent > quietLimitMs(agent, ADRIFT_MS);
}
@@ -0,0 +1,90 @@
// Plan2Code Web Console - the tab icon's status bar.
//
// The favicon follows the status line with a bar along its bottom edge: dashed
// amber while Plan2Code is busy, solid green when it is the person's move,
// solid red when it needs attention, and none (the plain Planny) once the
// session is finished. It reads from the same calls
// renderStatus() already makes for Planny's mood, so the icon and the line
// beside him can never disagree. No DOM access at module scope: `node --test`
// imports this file.
// Planny exactly as public/favicon.svg draws him (a test keeps the two in
// step), minus the <svg> wrapper so the bar can be layered on top.
export const FAVICON_BODY = `<rect x="4" y="4.6" width="24" height="30" rx="5" fill="#fff4ef" stroke="#3b332c" stroke-width="2.2" />
<path d="M16 9.8 L17.41 13.26 L21.14 13.53 L18.28 15.94 L19.17 19.57 L16 17.6 L12.83 19.57 L13.72 15.94 L10.86 13.53 L14.59 13.26 Z" fill="#ae4f2b" />
<path d="M11.6 22.2 Q16 26.4 20.4 22.2" fill="none" stroke="#3b332c" stroke-width="1.8" stroke-linecap="round" />`;
// The bar's color per state. `plain` has none.
export const FAVICON_COLORS = Object.freeze({
busy: "#f5a524",
ready: "#22c55e",
attention: "#e5484d",
});
/**
* Which icon a status line calls for. `mood` and `spin` are what renderStatus()
* handed to say(): `spin` marks something really happening at the other end
* right now, so that is busy. A working mood without the spin is the stale
* line ("has not checked in for a few minutes"), which, like adrift and gone,
* needs attention. `point` is the person's move. A finished session is plain
* unless it is visibly busy again (opening the dashboard, starting a review).
*/
export function faviconState({ mood, spin = false, finished = false } = {}) {
if (spin) return "busy";
if (finished) return "plain";
if (mood === "adrift" || mood === "gone" || mood === "work") return "attention";
if (mood === "point") return "ready";
return "plain";
}
/**
* The whole SVG for a state: Planny, plus a bar across the bottom edge, where
* he is cropped anyway. Busy's bar is dashed on a white strip (a progress bar
* mid-run) and the others solid, so busy differs by pattern as well as color:
* amber and green are close to one color for red-green color-blind eyes.
*/
export function faviconSvg(state) {
const color = FAVICON_COLORS[state];
let badge = "";
if (state === "busy") {
badge = `\n <rect x="0" y="27" width="32" height="5" fill="#ffffff" />\n <path d="M1.5 29.5 L30.5 29.5" stroke="${color}" stroke-width="4" stroke-dasharray="4 2.5" />`;
} else if (color) {
badge = `\n <rect x="0" y="27" width="32" height="5" rx="1" fill="${color}" />`;
}
return `<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 32 32">\n ${FAVICON_BODY}${badge}\n</svg>\n`;
}
let shown = "plain";
/**
* Point the page's two icon links at `state`'s icon, only when it changed. The
* SVG link takes the variant as a data URL; the PNG link, for browsers that
* pick the bitmap, gets the same variant drawn through a canvas. Plain goes
* back to the served files.
*/
export function setFavicon(state) {
if (state === shown || typeof document === "undefined") return;
shown = state;
const svgLink = document.querySelector('link[rel="icon"][type="image/svg+xml"]');
const pngLink = document.querySelector('link[rel="icon"][type="image/png"]');
if (state === "plain") {
if (svgLink) svgLink.href = "/favicon.svg";
if (pngLink) pngLink.href = "/favicon.png";
return;
}
const url = "data:image/svg+xml," + encodeURIComponent(faviconSvg(state));
if (svgLink) svgLink.href = url;
if (!pngLink) return;
const img = new Image();
img.onload = () => {
// A newer state won the race while this one was loading.
if (shown !== state) return;
try {
const canvas = document.createElement("canvas");
canvas.width = canvas.height = 64;
canvas.getContext("2d").drawImage(img, 0, 0, 64, 64);
pngLink.href = canvas.toDataURL("image/png");
} catch {}
};
img.src = url;
}
Binary file not shown.

After

Width:  |  Height:  |  Size: 3.5 KiB

@@ -0,0 +1,5 @@
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 32 32">
<rect x="4" y="4.6" width="24" height="30" rx="5" fill="#fff4ef" stroke="#3b332c" stroke-width="2.2" />
<path d="M16 9.8 L17.41 13.26 L21.14 13.53 L18.28 15.94 L19.17 19.57 L16 17.6 L12.83 19.57 L13.72 15.94 L10.86 13.53 L14.59 13.26 Z" fill="#ae4f2b" />
<path d="M11.6 22.2 Q16 26.4 20.4 22.2" fill="none" stroke="#3b332c" stroke-width="1.8" stroke-linecap="round" />
</svg>

After

Width:  |  Height:  |  Size: 443 B

@@ -0,0 +1,65 @@
// Plan2Code Web Console - the Help dialog's tab list and rules.
//
// The source of truth for which tabs Help has, in what order, and which one
// opens. The content itself is HTML in <template id="help-content">
// in index.html. No DOM access at module scope and no imports: the page and
// `node --test` both import this file.
const tab = (id, label, diagram) => Object.freeze({ id, label, diagram });
export const HELP_TABS = Object.freeze([
tab("about", "What this page is", true),
tab("around", "Finding your way around", true),
tab("workspace", "Your workspace", false),
tab("planny", "Reading Planny", true),
tab("answering", "Answering questions", false),
tab("sending", "Sending and what happens next", true),
tab("notes", "Notes and attachments", false),
tab("docs", "Documents and printing", false),
tab("dashboard", "The dashboard", true),
tab("meter", "Session meter", false),
tab("build", "Watching a build", false),
tab("ask", "Ask a quick question", false),
tab("finishing", "Stopping and finishing", false),
tab("stuck", "When something looks stuck", true),
tab("prefs", "Keyboard and preferences", false),
tab("privacy", "Privacy and where things are saved", false),
]);
// Not labels.js's BUILD_WORKFLOWS: that one leaves out quick-task, which is a
// build as far as the person watching it is concerned.
const HELP_BUILD_WORKFLOWS = new Set(["implement", "implement-review", "quick-task"]);
// Which tab Help opens on, from where the person is. First match wins.
export function helpTabFor({ view, dashboard, gone, adrift, finished, workflow, workspace, meterRed } = {}) {
const v = view || "";
// Finished comes first: a finished session's server is stopped on purpose,
// which also sets `gone`, and that is not something being stuck.
if (finished || v === "end") return "finishing";
if (gone || adrift) return "stuck";
if (workspace) return "workspace";
if (v === "ask") return "ask";
if (dashboard && meterRed) return "meter";
if (dashboard) return "dashboard";
if (v === "overview" || v.startsWith("doc:")) return "docs";
if (HELP_BUILD_WORKFLOWS.has(workflow)) return "build";
return "about";
}
// The tab index a key moves to, or -1 when the key is not a tab key.
export function helpKeyTarget(key, index, count) {
switch (key) {
case "ArrowDown":
case "ArrowRight":
return (index + 1) % count;
case "ArrowUp":
case "ArrowLeft":
return (index - 1 + count) % count;
case "Home":
return 0;
case "End":
return count - 1;
default:
return -1;
}
}
File diff suppressed because it is too large. Load diff
Loaded 100 of 668 files, more files were not shown because too many files have changed in this diff. Show more