diff --git a/.agents-docs/AGENTS-architecture.md b/.agents-docs/AGENTS-architecture.md index 4729dea..a28ac18 100644 --- a/.agents-docs/AGENTS-architecture.md +++ b/.agents-docs/AGENTS-architecture.md @@ -16,7 +16,7 @@ plan2code/ │ │ ├── handoff.md # Clearing gate + PLAN-DRAFT handoff │ │ └── trail.md # Every-response map visual + pathed resume command │ ├── shared-references/ # Reference files shared across skills -│ │ ├── building-habits.md # Find the cause, test first, pin behaviour (quick-task, implement, implement-review) +│ │ ├── building-habits.md # Find the cause, test first, pin behavior (quick-task, implement, implement-review) │ │ └── helper-strength.md # How strong a model and how much effort each helper gets (pathfinder, plan, revise-plan, quick-task, implement, implement-review, review) │ ├── plan2code-review-references/ # Reference files for review skill │ │ ├── verification-protocol.md # Deep verification + confidence calibration @@ -95,11 +95,19 @@ plan2code/ | `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, Devin or Codex (installed by `I` and Advanced → `E`; also via Advanced → `D`). When the desktop app is installed (`~/.plan2code/app/plan2code-app[.exe]`, or the file `PLAN2CODE_APP` names) and a display exists (Windows and macOS always; Linux needs `DISPLAY` or `WAYLAND_DISPLAY`), it opens the app after the agent and model questions with `--project --agent [--model ]` and exits; `--classic` (never forwarded to the CLI) and a failed spawn keep the terminal flow. The app reads all three (`project::launch_choice`): the picker starts on `--agent`, and `--model` applies to that agent on every start and Restart (`AgentManager::set_model`); a model never applies to another agent | | `plan2code-app/` | Tauri desktop app (Rust host, xterm panel, the console's dashboard in a frame; its own README). `RELEASE-CHECKLIST.md` lists the by-hand checks before each release. `src-tauri/src/versions.rs` compares the app's version (`tauri.conf.json`) with the installed skills' (`~/.plan2code/bin/version.json`) for the bar's **Update skills** line, which runs `node ~/.plan2code/bin/plan2code.js update` only when pressed | +| `plan2code-app/src-tauri/src/attention.rs` | The taskbar signal's rules, pure and knowing nothing about Tauri: `Signal` (`busy`, `ready`, `attention`, `plain`), `badge_for(signal, os, enabled)` (Windows: green dot on `ready`, red on `attention`; macOS: red `!` on `attention` only; Linux: count 1 on `attention`), `should_flash(prev, next, focused, enabled)` (once, only on a change into `ready` or `attention` while unfocused) and `dot_rgba()` (the dot's pixels, in the favicon's colors, which a test checks against `public/favicon.js`). `main.rs` applies the result through `set_attention`, `set_taskbar_signal` and `clear_attention_cmd`; the page asks with the `attention` and `taskbar` messages (`public/app-bridge.js`) and the switch is `taskbarSignal` in the app's own `~/.plan2code/app.json`. | +| `plan2code-app/src-tauri/src/agents.rs` | Custom agents in the app: `CustomAgent`, the `Agent` handle over built-ins and custom entries, `load_custom_agents`, reading the Agents file below; the console side is `src/web-console/agents.mjs` | | Desktop app install (`install.js`) | `installDesktopApp()` downloads `plan2code-app---[.exe]` and `plan2code-app--SHA256SUMS.txt` from the GitHub release (`PLAN2CODE_RELEASE_BASE` replaces the URL; the tests serve a stand-in), checks the SHA-256 and places the file in `~/.plan2code/app/` (the old file is renamed aside first, since Windows cannot overwrite a running exe). `app` is an install piece (`INSTALL_PIECES`, after `launcher`), so `plan2code update` updates it with the rest. `install()` asks for it after the command (default yes; a no is kept as `skipped` in `install.json`); unattended it installs it unless it is already recorded or was declined. A failed app never fails the install. `.github/workflows/release-app.yml` builds one file per system (`tauri build --no-bundle`) and the checksum list on a published release, after `scripts/check-versions.mjs --tag` | ### Agents file -`~/.plan2code/agents.json` holds the person's custom agent CLIs (`{ "schema": 1, "agents": [...] }`). The web console is its only writer (User Preferences → Agents, through `POST /agents`); the `plan2code` launcher and the desktop app only read it. Each reader carries its own copy of the rules (`src/web-console/agents.mjs` for the console), and `scripts/fixtures/agents-cases.json` is the contract all three readers' tests load: file normalization with its warnings, `buildArgs` command lines and `makeId` ids. A bad file or entry is skipped with a warning, never a crash. Until the feature is finalized, the rules are written out in `specs/agent-setup-and-taskbar-attention/PLAN-DRAFT-20261025.md` section 4.4. +`~/.plan2code/agents.json` holds the person's custom agent CLIs. The web console is its only writer (User Preferences → Agents, through `POST /agents`); the `plan2code` launcher and the desktop app only read it. Each reader carries its own copy of the rules (`src/web-console/agents.mjs` for the console, a copy in `src/launcher/plan2code.js`, `agents.rs` in the app), and `scripts/fixtures/agents-cases.json` is the contract all three readers' tests load: file normalization with its warnings, `buildArgs` command lines and `makeId` ids. Change a rule in the fixture first. + +- **File:** `{ "schema": 1, "agents": [...] }`. Any other schema, text that is not JSON, or `agents` that is not a list gives no custom agents and one warning. A missing file is an empty list with no warning. More than 50 entries: the first 50 are used. Unknown keys, in the file and in an entry, are kept when the console saves. +- **Entry:** `id`, `name` and `command` are required; `args` (default `[]`), `prompt` (default `"/plan2code"`, blank allowed), `bypassArgs` (`[]`), `modelArgs` (`["--model", "{model}"]`) and `models` (`[]`, each `{ id, label? }`) are optional. The three argument fields are lists of text, one list item per argument. +- **Rules:** `id` matches `^[a-z0-9][a-z0-9-]{0,31}$`, is not `claude`, `devin` or `codex`, and is unique; `name` is at most 60 characters and unique ignoring case; `command` is not blank. `makeId(name, taken)` makes the id from the name in kebab case (`agent` when nothing is left), with `-2`, `-3` on a clash, at most 32 characters. +- **Bad entries:** a bad entry is skipped with a warning naming it (`agents[i]`, 0-based) and the rest still load, in all three readers. `POST /agents` instead refuses the save with `{ error, index, field }`, and answers `409` while the file holds something `GET /agents` could not read, unless the body says `replace: true`. +- **Command line:** `buildArgs` returns `args` with `{prompt}` filled (the prompt is appended when no argument holds it, and a blank prompt drops an argument that is exactly `{prompt}`), then `modelArgs` with `{model}` filled when a model is chosen, then `bypassArgs` when checks are off. Arguments travel as a list, never through a shell; the Windows `cmd` wrapper for a `.cmd` or `.bat` belongs to the launcher and the app. ## Workflow Prompts (in `src/`) @@ -115,7 +123,7 @@ plan2code/ | `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, analysed by one fresh helper when the tool has one (verified in session; the report says who reviewed) | +| `plan2code-review.md` | review | Post-implementation comprehensive review, analyzed by one fresh helper when the tool has one (verified in session; the report says who reviewed) | | `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 | | `plan2code-xray.md` | xray | Explain how and why an area works (files, commits, PRs via `gh` when present), or rank the rough spots worth fixing into `ARCHITECTURE-BRIEF-.md` for Pathfinder or Plan; never edits code | @@ -250,7 +258,7 @@ Optional local browser UI shipped with **every** skill. Source lives in `src/web | `public/styles/poses.css` | The start-screen poses, one per skill. After the dashboard on purpose: its narrow-window rules override the dashboard's | | `public/styles/responsive.css` | Print, then the narrow-window rules: last, so its overrides come after everything they override | | `public/styles/shell.css` | The page shell: banner, top bar and meter, layout, sidebar, aside, footer, waiting and arrival | -| `public/styles/tokens.css` | Colours, type, sizes and the light and dark themes: the variables every other file reads | +| `public/styles/tokens.css` | Colors, type, sizes and the light and dark themes: the variables every other file reads | | `public/help.js` | the Help dialog's PURE rules: `HELP_TABS` (18 tabs: ids, order, labels, diagram flags; every tab uses the chosen highlight color; The git board sits right after Watching a build), `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, the Subagents tab → Helpers (subagents), a red-meter dashboard → Session meter, dashboard, a doc or the Overview, a git session → The git board, a build, else What this page is) and `helpKeyTarget()` (roving arrow keys, Home, End, wrapping). No module-scope DOM, so `node --test` covers it | | `git-repo.mjs` | the one read-only reader of a git repository: the worktree status (`countPending()`, behind `pendingFiles()`), the branch table, the default branch, a merge or rebase in progress, the remotes and their web addresses, the last fetch and the pull-request link; `snapshot()` is what `open` and every `reread` store as `state.git`. Every git call carries `GIT_OPTIONAL_LOCKS=0`, so a background look never takes the index lock. Imported by `cmd-open.mjs`, `cmd-post-wait.mjs` and `sessions.mjs`; the git skill's `git.mjs` reaches it through `common.mjs` | | `public/git-board.js` | PURE, one import (`pr-url.js`): the git board's rules from a `state.git` snapshot: branch order and the Remote only fold at `REMOTE_FOLD` 50, whether a new branch name is free, which buttons show and are enabled, whether the board still waits on the agent (`boardAwaiting()`: from a click until the agent's next post, not just until `wait` collects it), the plain sentence each click sends as `reply`, the "Last refreshed" line, which files start ticked in the commit dialog, and the pull-request link; the next step (ten rules, `nextStep()`), the tab that owns it, per-tab badges (`tabBadges()`), the busy mode and per-control gate (`busyMode()`, `controlAllowed()`), the tab to show (`pickTab()`), and the create-panel sources and clean-up groups. `app.js` only renders what it says | @@ -289,17 +297,17 @@ Key design points: - **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/` (the `/version` route in `routes.mjs`). `renderHelpVersion()` in `screens/help-dialog.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 `routes.mjs` read the shipped `~/.plan2code/bin/models.json` and read/write the user's `~/.plan2code/models.json` (`SHIPPED_MODELS_FILE` / `USER_MODELS_FILE` in `files.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 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` (canceled). The agent loops. Nothing blocks for the whole time a person takes. - **A turn is one command, and the CLI does the bookkeeping.** Agent speed is model round trips and typed tokens, not the console (`bench-web-console.mjs --roundtrip`: Send→agent ~70 ms, post→page ~180 ms), so the CLI absorbs every step with one right answer: `consume()` marks the agent `working` when it hands over a submit (`markWorking()`), and `wait` hands the turn back (`markWaiting()`) when it starts with an open question, a finish, or on the dashboard or git board; `post --wait ` posts and waits in one process; `settle: true | [ids]` (`settleItems()` in `patches.mjs`) records the latest send's cards as answered from their `submitted`; a new or reopened question with no stated status sets `waiting` and waiting drops `quietMinutes` (`settleAgent()`); a doc whose content changed with no `version` moves up one (`mergeDocs()`); a chat reply with no `id` is `r`; doc blocks with `file` (and `select: "tasks"`, doc-level `refresh: true`) are read from the worktree by `resolveDocFiles()` before the merge, refusing paths outside it; and a hop (`__launch`, `__dashboard`, `__commit`, `__pickup`) comes with `resume` (the exact `open --resume` command) and `skillFile` (`hopFor()`), a pick-up also with `handoffFile`. The wait loop's health check runs beside the 50 ms polling instead of blocking it, and is aborted on exit. `help ` prints any contract file; `help topics` lists them. `token` defaults to the option key in the page's `tokenFor()`, and nothing reads `fallbackText`, so the contract no longer asks for either. - **The CLI hands the agent its terminal line.** `open`, `post` and `wait` (exits `0` and `10`) print `terminalLine: "→ Look at the web console: "` (`terminalLine()` in `sessions.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, handoffs }` 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 `/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 `routes.mjs`, which paints `data-width` on first load; `styles/tokens.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 `/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. +- **`GET /overview` is the one route that reads a project file.** It serves `/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 honors `?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, folder? }], bytes }` (`folder` on `notepad:` items only), ids `session:` (the `staleSessions()` rule, and every file in the folder that old too, never the serving sid), `console:` (anything at the top of the console home but `PROTECTED_CONSOLE_FILES`, `sessions` and `runtime`, directories too) and `root:` (anything at the top of `~/.plan2code` but `KEPT_PLAN2CODE_ENTRIES`: `console`, `bin`, `launcher.json`, `install.json`, `models.json`, `handoffs`, `notepads`, `app`, `app.json`, `agents.json`), `notepad:` (a month-old notepad, whole; see Notebook below) and `completed:` (each folder or link directly under the session worktree's `specs--completed/` whose newest mtime is older than the session age, 30 days; loose files there stay; the server passes `project: sessionWorktree() || PROJECT`, and without a `project` the archive is never scanned). 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 `..tmp`, which live for milliseconds. `root:` is only scanned when the console home is `/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//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 "/console.mjs" open --workflow ` 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 `screens/starting.js` from `