mirror of
https://github.com/backnotprop/plannotator.git
synced 2026-09-14 14:17:26 +08:00
v0.26.4
541 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Michael Ramos
|
d5ae439f7a | chore: bump version to 0.26.4 | ||
|
Michael Ramos
|
c760fc522b | chore: bump version to 0.26.3 | ||
|
Michael Ramos
|
7ad4d39ed9 |
feat(comments): reference agent skills with / or $ in plan review and annotate comments (#1229)
* feat(comments): reference agent skills with / or $ in plan and annotate comments
Typing / or $ at the start of a word in the document-UI comment composer
opens a picker of the user's global agent skills (~/.claude/skills,
~/.codex/skills, ~/.agents skills roots), served by a new GET /api/skills
on the plan and annotate servers in both runtimes (Bun + Pi mirror).
Multiple references per comment are supported; references live in the
comment text itself and are appended to exported feedback as a
'Skills referenced' block so the acting agent knows which skills to apply.
Human-invocation-only skills (disable-model-invocation: true frontmatter)
stay listed and selectable but render dimmed with a badge, warn in the
menu and composer, and are marked in the export so the agent is never
asked to invoke something it cannot.
Discovery reuses the review-skill loader (same roots, precedence, and
skip-and-log discipline), reads only an 8KB head per SKILL.md, caps the
catalog at 500 skills, takes no client input, and is never persisted;
any failure degrades to plain typing.
* fix(comments): harden skill references per review (trigger, IME, seam, fail-closed frontmatter)
Blockers:
- B1: a trigger now requires at least one query character. A bare / or $
no longer opens the catalog, so Enter stays a newline and Tab still
leaves the field ("This costs $" + Enter, "cd /" + Tab, bullets).
- B2: the menu ignores keys mid-IME-composition (nativeEvent.isComposing),
matching the 16 existing guards; Enter committing a Pinyin/Telex/Korean
candidate can no longer insert a skill.
- B3: the catalog request is a host seam (skillCatalogTransport via
configurePlannotatorUI), defaulting to the existing GET /api/skills.
- B4: resetSkillCatalogCache() invalidates outstanding requests
(generation counter), and a late-resolving stale request can no longer
overwrite a newer cached value or the export registry. The catalog
tests reset in beforeEach, so they hold in any file order.
Also:
- F1: skillReferences={false} is fully inert — the human-only notice memo
and the cache seed are gated on the prop.
- F4: frontmatter flag parsing no longer fails open: trailing YAML
comments are stripped, on/1 (and TRUE/yes etc.) read as true, the head
read is 64KB, and truncated unterminated frontmatter fails CLOSED on
disable-model-invocation.
- F5: extraction ignores markdown link destinations ([x](/name)), shell
redirects (cat /x > out), and /-triggered FHS root names (/run, /tmp);
menu insertion switches / to $ for those names so inserted references
always survive extraction.
- F6: the 500-skill cap slices after sorting, so which skills survive no
longer depends on readdir order.
- F3: /api/skills wiring guards for the Bun and Pi plan + annotate
servers (skills-endpoint.test.ts).
- Keyboard state machine tests against the real CommentPopover in
happy-dom (bare trigger, insertion, composition, Escape, highlight
bounding, opt-out inertness), added to the CI DOM step.
- The insertion path dismisses the trigger start so the menu close is
ordering-safe against React's select-plugin re-reading a stale caret.
* feat(comments): redesign the skill reference menu (bare triggers, no preselection, highlighted tokens)
Per maintainer direction, reversing the earlier bare-trigger opt-out
deliberately: typing a bare / or $ at the start of a word now opens the
full skill catalog immediately, and the safety story moves from the
trigger to the menu itself.
No preselection (the load-bearing rule): the menu opens with NO row
active, and while nothing is active every key behaves exactly as if the
menu were closed. "This costs $" + Enter is a newline; "cd /" + Tab
leaves the field (the proven regression that must never return). A row
activates only via ArrowDown/ArrowUp (Down from none lands on the first
row, Up on the last); only then do Enter/Tab insert. Pointer hover never
activates a row, because the menu floats exactly where the mouse rests
over the composer; a click inserts directly and never arms Enter.
Continuing to type re-filters and disarms any active row. Escape clears
the active row and dismisses when the user engaged (query typed or row
active); an unengaged bare-trigger menu passes Escape through so closing
the composer still costs one press.
Menu redesign to the reference look: icon, bold name, dimmed inline
description with ellipsis, right-aligned source column (Agents / Claude
/ Codex from the discovery roots), rounded generously padded rows, and a
subtle active-row background; human-only rows stay dimmed with their
badge and the warning now shows while such a row is ACTIVE.
Inserted references render highlighted in the composer via a mirrored
aria-hidden overlay behind a transparent-text textarea (identical font,
padding and wrapping metrics; scroll synced; tokens change color and
background only, drawn from the --primary theme token so every palette
works in light and dark). The caret keeps --foreground, selection uses a
translucent primary wash, and IME composition temporarily restores
native textarea text so composition underlines render normally.
skillReferences={false} still renders the plain pre-feature textarea.
Also, per review:
- extraction: dropped the over-broad shell-redirect exclusion (false
negatives on prose like "use /animate <- this one"; the motivating
case stays covered by the reserved-path rule)
- frontmatter: an unterminated frontmatter block now fails CLOSED on
disable-model-invocation even in complete (untruncated) files
- the reserved-path / to $ insertion switch stays: extraction still
reads /run as a path, and the new token highlight makes the switch
self-explanatory (an unhighlighted insert would look broken)
The composition guard, transport seam, catalog generation counter,
enabled gating, and export rules are unchanged and re-covered by the
rewritten DOM test matrix.
* fix(comments): give the skill reference menu adaptive, viewport-clamped placement
The menu rendered bottom-full with a fixed max-h-64: always upward, up to
256px, with no viewport awareness. With the comment popover near the top of
the viewport (annotating near the top of a document), typing a trigger ran
the menu off the top of the screen with its upper rows unreachable.
Placement now mirrors the popover's own computePosition idiom: measure the
space above and below the composer wrapper against window.innerHeight,
prefer above (the shipped direction; keeps the action row and human-only
notice visible), flip below when the list fits below but not above, and when
neither side fits pick the roomier side. The list's max height is clamped to
the available space (still capped at the former 256px), so the menu never
extends past a viewport edge. Recomputes on every commit (drag moves,
popover flips, filtering changing the item count, warning-footer toggles)
plus capture-phase scroll and resize listeners, matching the popover's
tracking. Visual design of the menu and rows is unchanged.
* feat(comments): inject human-only skill instructions into exported feedback
A human-only skill (disable-model-invocation: true) referenced in a review
comment used to export as a dead name the agent could do nothing with. A
human referencing a human-only skill IS the human invocation, so the export
now injects the skill's SKILL.md body verbatim (frontmatter stripped) inside
clearly delimited BEGIN/END SKILL INSTRUCTIONS markers, with the absolute
skill directory and the resolve-relative-paths pointer so references/,
scripts/, and assets/ stay actionable. Model-invocable skills keep exporting
as names the agent can invoke itself.
Transport is lazy: a new GET /api/skills/content?name= endpoint (Bun and Pi)
serves one discovered skill's body, capped at 20k chars with an explicit
truncation notice pointing at the file; the client fetches contents only for
the human-only skills actually referenced, keyed off comment state, and the
catalog now carries each skill's absolute dir so every failure path (deleted
skill, unreadable file, race with submit) degrades to naming the skill plus
its directory. Names are matched against discovery only and never used as
paths, so traversal cannot escape the skill roots. A per-export dedupe
injects each skill once even when several comments reference it, and
GLOBAL_COMMENT annotations run through the same block.
The referenced-skills header now says the reviewer is asking for the
invocation, and the human-only menu footer and composer notice explain that
the skill's instructions will be included with the feedback instead of
warning that the reference will not work.
* polish(comments): quiet, progressive human-only skill treatment
The human-only surfaces shipped with too much emphasis: a dimmed row plus
a bordered uppercase badge, an amber warning footer, and a persistent
amber notice in the composer after insertion. Human-only is a property of
a skill, not an error state, so the treatment is now quiet and
progressively disclosed:
- Menu rows render at full strength with a small muted 'human-only' pill
(bg-muted / muted-foreground tokens; no border, no dimming).
- The plain-language explanation (a model cannot invoke it, so its
instructions will be included with your feedback) appears as a muted
footer only while a human-only row is active (keyboard) or hovered
(pointer). Hover disclosure is purely visual state local to the menu;
it never touches activeIndex, so the no-preselection invariant and the
hover-never-arms-Enter rule are unchanged and re-asserted by a new test.
- When not disclosed, the same sentence stays in the DOM sr-only and
human-only rows point at it with aria-describedby, so the state reaches
assistive tech as text rather than as a purely visual badge (this does
not attempt the #1233 combobox semantics, and does not worsen them).
- After insertion, the highlighted token itself carries the quiet inline
marker (a dotted primary underline; text-decoration cannot move glyphs,
so overlay alignment is untouched) and the standing amber notice is
replaced by a native <details> disclosure: a single muted 'Includes
skill instructions' summary line that expands to the full accurate
sentence, operable by pointer, keyboard, and AT alike.
No amber remains; every color is a theme token (muted, muted-foreground,
border, primary, ring), so the treatment follows every palette in light
and dark. Copy is unchanged where it was accurate. Behavior is unchanged:
human-only skills stay selectable and injection still happens.
* fix(comments): harden human-only skill injection per adversarial review
Three findings on the injection path, each with tests that fail pre-fix:
1. Marker forgery: an injected SKILL.md body containing our own
`--- BEGIN/END SKILL INSTRUCTIONS ---` markers (or an
`[Instructions truncated:` notice) could close the block early — making
everything after it read as the reviewer's own words — forge a block for
a skill nobody referenced, or forge a truncation notice pointing at an
attacker-chosen path. Body lines matching the structural marker forms
(leading-whitespace and case variants included) are now visibly
neutralized before injection: kept verbatim but prefixed, never silently
deleted (neutralizeSkillMarkerLines).
2. Forged human invocation: POST /api/external-annotations is
unauthenticated on localhost, so any local process could submit a
comment referencing a human-only skill and cause its instructions to be
injected "at the reviewer's request". Annotations carrying a `source`
now still LIST their skill references but never cause verbatim
injection — human-only references fall back to naming the skill plus
its directory, with an honest reason. The content-prime effect skips
external texts for the same reason. A human referencing a human-only
skill IS the human invocation; a tool is not.
3. Unbounded read: readReferenceSkillContent read the whole SKILL.md
before slicing to the 20k cap, so an unauthenticated no-cors fetch loop
could balloon RSS by file size per request (measured +64.4MB for a 64MB
file). It now uses the same bounded readFileHead as the catalog,
reading only frontmatter allowance + 4 bytes per capped char + slack;
truncation detection is unchanged for any file whose frontmatter fits
the catalog bound, and frontmatter that overflows the read falls back
to null rather than serving raw YAML. Measured: 12 reads of a 64MB
SKILL.md now cost +5.1MB total.
Also: the fast-fail guard no longer rejects legitimately discovered names —
`name.includes("..")` 404'd a real `v1..2` skill dir forever (and `\` is
legal in POSIX names) while defending nothing, since the name is only ever
matched against discovery output and never joined into a path. It now
rejects exactly the names that can never be a readdir entry: empty, `.`,
`..`.
|
||
|
Michael Ramos
|
b69742c3bf |
feat: add PLANNOTATOR_URL_HOST display-only override for advertised URLs (#1225)
* feat: add PLANNOTATOR_URL_HOST display-only override for advertised URLs Remote mode binds 0.0.0.0 but every advertised URL hardcoded http://localhost:<port>, so a session opened from another device (e.g. a phone on the same tailnet) got an unopenable link (#657). - resolveUrlHost() in packages/shared/config.ts: PLANNOTATOR_URL_HOST env var over config.json urlHost, validated host-only (bare hostname, IPv4, bracketed IPv6); invalid values warn once and fall back to localhost. - buildAdvertisedUrl(port) in packages/server/remote.ts and its Pi mirror in apps/pi-extension/server/network.ts; all 7 construction sites use it. - Strictly display-only: binding stays governed by PLANNOTATOR_REMOTE, and agent-review jobs get a pinned http://127.0.0.1:<port> API URL. - Remote-ready copy says "open on your device" when the host is overridden; local sessions with an override warn it is unreachable. - Tests for validation, precedence, and URL composition in both runtimes; docs in CLAUDE.md and the marketing site. * fix(review): ignore urlHost in local sessions, harden warning output Review follow-ups on #1225: - Local (loopback-bound) sessions no longer honor the advertised-host override: honoring it auto-opened http://<host>:<port> against a server nothing was listening on, openBrowser still reported success, and the agent blocked on waitForDecision. Local sessions now advertise and open localhost, warning once that PLANNOTATOR_REMOTE=1 is required. - The invalid-host warning JSON-encodes the echoed value so an embedded newline cannot forge extra stderr lines (hosts surface session-ready lines as clickable links); warn-once is now per value. - Docs: local-session behavior reworded, the empty-env-suppresses-config semantic documented, secure-context note generalized. |
||
|
Michael Ramos
|
bbae458e5a | chore: bump version to 0.26.2 | ||
|
Michael Ramos
|
2dd5b7b25d |
docs: self-hosting page no longer claims a bundled Highlight.js (#1221)
highlight.js was removed in #1218; the portal bundles Shiki via the diff renderer. Caught by the v0.26.2 dependency audit. |
||
|
Michael Ramos
|
c08b188812 |
perf(ui): single Shiki highlighter, palette-matched code blocks, drop highlight.js (#1218)
* perf(build): stub out the dead Oniguruma WASM in every bundle
@pierre/diffs picks its Shiki engine with a runtime ternary:
engine: preferredHighlighter === "shiki-wasm"
? createOnigurumaEngine(import("shiki/wasm"))
: createJavaScriptRegexEngine()
Plannotator pins `preferredHighlighter: 'shiki-js'` (and Pierre's own
default is 'shiki-js'), so the Oniguruma branch never executes. Because
the choice is a runtime ternary, bundlers keep the `import("shiki/wasm")`
edge anyway and inline `@shikijs/engine-oniguruma/wasm-inlined`, a
~622 KB base64 blob, into the single-file HTML builds. The review app
paid for it twice: once on the main thread (via
`highlighter/shared_highlighter.js`) and once inside the `?worker&inline`
Pierre worker.
Alias `shiki/wasm` to a stub that throws if it is ever reached. Wired via
`resolve.alias` rather than a plugin because `resolve.alias` is shared
with Vite's worker build and `plugins` are not.
Highlighting output is unchanged: the JS regex engine was already the one
doing the work. Opting back into 'shiki-wasm' now fails loudly instead of
silently costing every user a megabyte of dead bytes.
apps/review/dist/index.html 19,424,646 -> 18,180,545 (-1,244,101 raw / -463,348 gzip)
apps/hook/dist/index.html 23,032,467 -> 22,410,416 (-622,051 raw / -233,485 gzip)
* perf(ui): consolidate code highlighting onto Shiki, drop highlight.js
The app shipped two highlighters. Shiki already tokenised the code-review
diff pane (via @pierre/diffs, JavaScript regex engine); highlight.js
separately coloured markdown fences and review suggestion snippets at
~982 KB minified for a full build of ~190 grammars. That second
highlighter is now gone.
Every call site moves onto `packages/ui/utils/codeHighlight.ts`, a thin
wrapper over Pierre's SHARED Shiki instance:
CodeBlock, Viewer, PlanCleanDiffView markdown fences
InlineMarkdown code-file hover preview
HighlightedCode review suggestion snippets
Reusing Pierre's instance rather than standing up a second fine-grained
one is deliberate. Pierre imports Shiki's full bundle, so every grammar
and theme is ALREADY inlined in the single-file builds: a separate
highlighter with a curated language list would have duplicated a subset
of bytes that are already there. Sharing costs nothing, gives every
language Shiki bundles instead of a shortlist, and — the point of the
change — guarantees fences resolve the exact same theme the diff pane
resolves.
Theming. `SHIKI_THEME_MAP` / `resolveSyntaxTheme` move from
`packages/review-editor/hooks/usePierreTheme.ts` to
`packages/ui/utils/syntaxTheme.ts`; usePierreTheme re-exports them, so
the review editor's imports are unchanged. `useFenceTheme()` feeds the
components and re-highlights on palette or mode change. Code blocks now
follow the active palette across all ~52 themes in both light and dark,
instead of always rendering github-dark and relying on hand-written
`.hljs-*` override stacks to stay legible. Those stacks are deleted:
`packages/editor/index.css`'s light-mode token palette, and
`colorblind.css`'s hand-tuned tokens which existed to APPROXIMATE
@pierre/theme's protanopia-deuteranopia themes that are now simply used.
Behaviour held fixed:
- Language-less fences stay plain text (#1212). No auto-detection
anywhere, including the hover preview, which previously called
`hljs.highlightAuto`. `HighlightedCode` derives its language from
the caller's file path; an unknown extension renders plain.
- `applyHighlight(el, ...)` keeps the imperative `hljs.highlightElement`
DOM contract the annotation layer reaches into, and writes plain text
at final size first so async highlighting causes no layout shift.
Already-attached grammars highlight synchronously — no flicker on
cached highlights.
- It also verifies the rendered text is byte-identical to the source
and falls back to plain otherwise, because annotations address code
blocks by text offset.
- `@plannotator/ui`'s public API is unchanged: the highlighter is a
module-level default like the package's other seams, no new props.
The `hljs` class on fenced `<code>` becomes `pn-code` (it is a
structural hook for blockTargeting, vim navigation and print.css, and it
named a library we no longer ship). `language-*` stays.
apps/review/dist/index.html 18,180,545 -> 17,270,889 (-909,656 raw / -291,921 gzip)
apps/hook/dist/index.html 22,410,416 -> 21,704,434 (-705,982 raw / -238,096 gzip)
Verified the diff pane is untouched: the rendered Pierre shadow-DOM
markup is byte-for-byte identical between an origin/main build and this
one (SHA-256 aa1ee88a…).
* fix(ui): strip stray NUL bytes from the code-highlight source
Two U+0000 bytes slipped into comments in the previous commit, which made
git treat the file as binary. Replaced with spaces; no behaviour change.
* fix(ui): keep code-block annotation marks across highlight swaps
Fenced code is annotated by hand: one `<mark data-bind-id>` inside the
`<code>` element, which `applyHighlight` also owns. Every highlight swap
(palette change, dark/light toggle, or the first async grammar attach
after load) replaces that element's children, so the mark was silently
wiped and nothing put it back. Annotation state, the sidebar panel and
exports were unaffected; the loss was purely visual, and deterministic.
`applyHighlight` now publishes every write through `onCodeHighlightSwap`,
synchronously, immediately after it. `Viewer` subscribes and re-paints the
fence's mark, so a swapped block ends up with BOTH the new theme's tokens
and its annotation. The shared painter (`paintCodeBlockMark`) moves the
token spans into the mark instead of flattening them to text, so creating
an annotation no longer costs a block its colours either.
Being driven by the swap also fixes the cousin race by ordering rather
than timing: share/draft restore runs on a timer after load, and on a slow
machine the first async swap could land after it and wipe the restored
marks per block. A restore that painted before the swap is now
re-established in the same task the swap ran in, and one that runs after
finds the mark already there.
Removal tombstones the id before re-highlighting, because the host drops
the annotation from state a tick later — without it the swap listener
would paint the just-removed annotation back in, and a fence carrying a
second annotation would end up bare.
Also closes the named gap in the WASM coverage: entry-assets only grepped
source, so a future @pierre/diffs bump could reintroduce the inlined blob
through a different import specifier unnoticed. It now greps the built
`apps/{review,hook}/dist/index.html` for the base64 WASM magic, skipping
on an unbuilt checkout and running for real in the CI job that builds the
bundles.
|
||
|
Michael Ramos
|
2d65c65596 |
feat(ui): pair a light theme and a dark theme, switched by mode (#1217)
* feat(ui): pair a light theme and a dark theme, switched by mode
ThemeProvider stored one palette plus a mode, so picking a dark-only
palette pinned the mode and greyed out the Light/System buttons. Store a
pair instead: { mode, light, dark }, resolved as pair[preferredMode], so
System flips between the two choices as the OS scheme changes.
The Settings Theme tab now assigns one half at a time. A Light/Dark
switch decides which half the grid is filling, the grid lists only the
palettes that can render that half (from the registry's modeSupport), and
a summary line names both halves with each side clickable. Every mode
button is permanently enabled: a dark-only palette simply never occupies
the light slot, so no mode coercion is left to do.
The pair round-trips through the SETTINGS registry to the `theme` key in
~/.plannotator/config.json the way diffOptions does. A user upgrading
seeds both halves from their stored single palette, and the legacy
plannotator-color-theme key keeps tracking the active palette so a
downgrade never lands on an unstyled first frame.
Addresses part 1 of #1211.
* fix(ui): make the theme pair seed local, and keep the legacy API non-destructive
Review of #1217 found a data-loss path and three published-API regressions.
Seeding: ThemeProvider handed its resolved pair to the config store through
set(), which queues a debounced POST. configStore.init() applies the server
config but never cancelled that queued write, so a single cookie-less visit
(fresh profile, incognito, cleared cookies) flushed a default pair to
~/.plannotator/config.json AFTER the real one had arrived, and the next
session restored those defaults over the user's cookies. The provider now
uses a new configStore.seed(): memory plus cookie, never the server, and
never over a value init() already applied. init() additionally retracts
queued writes for the leaves the server just spoke for, which closes the
same race for every server-synced setting rather than this one key.
Deprecated APIs: isThemeModeAvailable() and normalizeThemeMode() are back as
one-line wrappers with @deprecated notes, since packages/ui exports utils/*.
setColorTheme: assigns exactly one half and nothing else. A both-mode palette
goes to the half on screen instead of clobbering both; a mode-restricted one
goes to its half without yanking a System user to an explicit mode (render
time already resolves that). It persists through configStore.setLocal(), so
it stays cookie-only as it was before the pair, unless a host installed its
own serverSync transport.
storageKey / colorThemeStorageKey are honored on the read path, so a host's
stored pre-pair preference is migrated rather than discarded. The two halves
have no pre-pair equivalent and stay on fixed keys, documented on the props.
Tests: a fresh-mount case that pins zero POSTs (the previous helper pre-seeded
cookies, which is why this was invisible), a case that pins a real choice
still reaching config.json, direct setColorTheme cases for all three
semantics, a host-storage-keys migration case, and configStore seed/retract
unit tests. All of them fail against the code they replace.
|
||
|
Michael Ramos
|
50a54c872b | chore: bump version to 0.26.1 | ||
|
Michael Ramos
|
f555168deb | chore: bump version to 0.26.0 | ||
|
Michael Ramos
|
9c693ae348 |
fix(opencode): drop the bun peerDependency that made npm download a 50MB runtime (#1204)
npm >= 7 auto-installs peer dependencies, and the npm registry package named bun ships the full Bun binary, so every install of @plannotator/opencode pulled a useless ~50MB second copy of Bun. Express the runtime requirement as an informational engines field instead, which npm never installs. Also update the stale fixture comment that referenced the peer dependency's install weight. |
||
|
Michael Ramos
|
10a5104888 |
fix(install): repair the skills checkout guard, add --skip-skills (#1201)
* fix(install): make a failed skills checkout stop reporting success The skills/commands checkout runs in a subshell written as `( set -e; ... ) || checkout_failed=1`. POSIX ignores `set -e` for every command of an AND-OR list except the last, and bash 3.2.57 (what `curl | bash` gets on macOS), bash 5.3, dash, zsh and ksh all carry that suppression into the subshell. The `set -e` was inert. A failed clone therefore ran the whole block anyway, the subshell exited with the status of its trailing `if` (0), `checkout_failed` stayed 0, and the installer printed "YOU'RE ALL SET!" with no skills installed. It also blamed the wrong thing, printing "Tag vX.Y.Z predates the per-agent skill layout" when the real cause was a failed clone. Drop the inert `set -e` and guard the four fetch steps with explicit `|| exit 1`. Everything after the checkout stays best-effort, matching install.cmd, which only checks git clone and lets every xcopy run unchecked. A local cp/mkdir/rm failure must not surface as the "network or git error" message. Verified on bash 3.2.57 in a sandboxed HOME: the failure case now exits 1 with the fetch error and no success banner, and original vs patched success runs produce byte-identical trees (74 paths, 35 files) and identical logs. * feat(install): add --skip-skills opt-out The skills and slash commands come from a sparse `git clone` of the release tag. There was no way to decline that fetch short of --minimal, which also drops the sem sidecar, the agent-terminal runtime, the hooks, and every per-agent config. Anything that installs a tag github.com cannot serve had no option at all. Add --skip-skills to all three installers, following the existing --skip-codex / --skip-gemini / --skip-kiro / --skip-opencode family: CLI flag (-SkipSkills in PowerShell), PLANNOTATOR_SKIP_SKILLS_INSTALL env var, skipInstall.skills config key, resolved flag > env > config. It is not a per-agent switch; it covers every scope the checkout writes (Claude, ~/.agents, OpenCode, Gemini, Kiro), the extras, and the skill-scope cleanup sweeps. Skip means do-not-write: nothing already installed is replaced or removed, and git stops being a hard requirement. The run reports "Skills: skipped (<source>)" and the closing banner no longer claims the /plannotator-* commands are ready, which is the same false-success the checkout guard exists to prevent. Use it in the install-script-smoke job. That job installs a synthetic v9.9.9: the fake curl serves the freshly built binary for any URL, but the skills clone goes to real github.com, where the tag does not and cannot exist. That clone has always failed; it only went unnoticed while the broken guard let the installer exit 0 anyway. The job asserts Codex hook config, not skills, so it opts out rather than ignoring a real error. Both run_installer call sites go through the one function definition. Verified in an env -i sandbox on bash 3.2.57 (what `curl | bash` gets on macOS) with a fake curl and a local stand-in remote. Flag, env var, and config each skip and name their own source; flag beats env=0; env=0 beats config true; an explicit "skills": false stays a veto. Without the flag, pre-change and post-change runs produce byte-identical trees (39 entries) and identical logs. A bad clone URL without the flag still exits 1 with the fetch error and no success banner. A --skip-skills re-run over an existing install leaves all 12 skill and command files byte-identical. The CI step was reproduced locally: both run_installer calls exit 0 and every Codex assertion still passes. * test(install): cover --skip-skills and repoint the pinned source strings scripts/install.test.ts asserts against exact install-script source text, so six assertions broke when --skip-skills landed. Each is repointed at the new string with its intent preserved, not weakened: - The "hook/config writing happens before the git hard-fail" ordering test keeps proving the ordering; it just matches the gate's new conditional form. git being a hard requirement is now a narrower invariant (it applies only when the checkout actually runs), so that is asserted separately rather than dropped. - The skipInstall walk assertions follow codex/gemini/kiro/opencode gaining a skills entry, in install.sh's `for _agent` loop and install.cmd's PowerShell key list. - The three "never remove" sweep assertions follow the Codex stale-skill cleanup gaining its skills-opt-out arm, in all three installers. Add nine tests covering --skip-skills itself in the same style as the per-agent family: flag/switch parsing, PLANNOTATOR_SKIP_SKILLS_INSTALL, skipInstall.skills, and flag > env > config precedence by textual layering, for each installer. Each installer also gets a test that the opt-out bails before the clone and leaves the checkout guard intact (#1201's fix must keep failing a real fetch error), and one that the run reports honestly, never prints the "commands are ready" banner over an empty skills dir, and suspends the extras, the model-invocation rewrite, and the stale-stub sweeps rather than applying them partially. bun test scripts/: 116 pass, 6 skip, 0 fail. Full bun test: 2884 pass, 234 skip, 0 fail. bun run typecheck clean. |
||
|
Michael Ramos
|
d7be3406f6 |
fix(ci): repair the OpenCode 2 installed-package smoke (#1202)
The smoke's wait budgets were sized on a warm macOS dev box (5s to a healthy server, 20s to plugin activation). On a cold Linux runner OpenCode 2 needs longer to boot and has to install the packed plugin plus its whole dependency closure through the fixture's throwaway registry first, so the job has failed on every run since it was introduced. Measured, same opencode2 build and the same packed tarball: macOS, warm caches: healthy 0.8s, plugin activated 6.3s linux/amd64 container: healthy 8.2s, plugin activated 45.3s Raise the budgets to 120s and 300s (overridable via PLANNOTATOR_SMOKE_HEALTH_TIMEOUT_MS / PLANNOTATOR_SMOKE_PLUGIN_TIMEOUT_MS) and give the job a 25 minute backstop. The assertion is untouched: the smoke still requires the plugin registry to report the plannotator plugin. Also make a failure legible and prompt. Each poll gets a per-request timeout so one wedged request cannot swallow the budget, waits report progress, failures carry the elapsed time and the last HTTP status/body, and teardown escalates to SIGKILL and force-closes the registry. The CI failure previously burned five minutes in teardown before printing anything. |
||
|
Michael Ramos
|
308e4ba8a5 |
fix(opencode): drop runtime dependency on prerelease plugin nightly (#1199)
* fix(opencode): drop runtime dependency on prerelease plugin nightly
`@opencode-ai/plugin` was a runtime dependency pinned to the exact nightly
0.0.0-next-16775, so every `npm install @plannotator/opencode` resolved a
prerelease snapshot sitting inside npm's 72h unpublish window and pulled
95MB across 101 packages (effect@4.0.0-beta.101 alone is 47MB). None of it
is executed by OpenCode 1 users.
The only runtime use was `Plugin.define`, which is an identity function
(`export function define(plugin) { return plugin; }`, verified identical
across 0.0.0-next-16775, next-16600, next-16797 and stable 1.18.13). The
import is now type-only and the plugin is a plain object literal checked
with `satisfies Plugin.Plugin`. OpenCode 2's loader validation is purely
structural (`Schema.Struct({ id: String, setup: function })` in its
supervisor), so an object literal satisfies it.
The package moves to devDependencies. Built `dist/index.js` and
`dist/embedded.js` are byte-identical to the pre-change build;
`dist/server.js` differs only by the dropped import and the two
`Plugin.define(...)` wrapper lines.
* docs(opencode): explain why the V2 logReady callback is empty
The old one-liner read as an unfinished TODO. The empty function is
correct: OpenCode 2's server-plugin Context exposes no `log` or `tui`
domain, `@opencode-ai/client` has no `tui` namespace and zero `/tui/*`
routes (checked on 0.0.0-next-16775 and next-16797), and `createV2Client`'s
`app.log` bottoms out in `console.error`, the same stderr stream
`handleServerReady` already writes to. Wiring it would print the session
URL twice in remote mode and add a stray line locally. V1 targets
`client.app.log` and `client.tui.showToast`, which are HTTP surfaces
distinct from stderr, so V1 never repeats itself. `tui.toast.show` exists
in V2 only as a subscribe-only event and on the separate
`@opencode-ai/plugin/tui` context, a different plugin kind in a different
process, so a real toast needs an upstream OpenCode API.
Comment only, no behavior change.
|
||
|
Sergiy Dybskiy
|
050dfcda9a |
feat(opencode): add OpenCode 2 plan review adapter (#1194)
* feat(opencode): add OpenCode 2 plan review adapter * fix(opencode): harden V2 review lifecycle * fix(opencode): address V2 review feedback * fix(opencode): update V2 target and isolate tests * test(opencode): assert prompt composition invariants |
||
|
Michael Ramos
|
7682628db7 |
feat(install): Codex opt-out and credential-free attestation verification (#1197)
* feat(install): Codex opt-out and credential-free attestation verification Implements both asks from #1178 (reported and designed by @astradevkin): - Per-agent installer opt-outs: --skip-codex / --skip-gemini / --skip-kiro flags, PLANNOTATOR_SKIP_{CODEX,GEMINI,KIRO}_INSTALL env vars, and config.json skipInstall.{codex,gemini,kiro} keys, with flag > env > config precedence mirroring verifyAttestation. Detected-but-skipped is reported as its own honest state, never conflated with not-detected, and a skipped agent's home is neither written nor cleaned up. - Credential-free provenance verification: when --verify-attestation is active, the Sigstore bundle is fetched from GitHub's public attestations endpoint (single unauthenticated attempt, no retry) and verified via gh attestation verify --bundle with the same --repo/--source-ref/--signer-workflow constraints; gh's authenticated fetch remains the fallback. TUF trust-root failures are reported as connectivity, distinct from real provenance failures; every path stays fail-closed. Zero behavior change for users who do not opt in: the default install path is unchanged (verified by sandbox-HOME parity runs against main). * fix(install): address #1197 review round (H1 retry, M2-M7, lows) - H1: a failed gh --bundle invocation now retries once through the exact authenticated path before any classification, so an older gh (unknown flag) or a corrupt bundle never reports as a provenance failure. Pinned by a functional stub-gh test; a real failure still fails again on the retry and aborts. - M2: the sh config layer extracts the skipInstall object (awk, character indexed) before matching per-agent keys and honors explicit false as a veto; cmd now parses the real JSON via PowerShell like ps1. Functional tests cover the foreign-key collision and explicit-true cases. - M3: sh names the real cause when the bundle path cannot run (no JSON extractor vs fetch vs extraction failure) and gains python3 and jq fallback extractors; docs state the dependency. - M4: ps1/cmd gate the existing-integration note on plannotator content in hooks.json and word the skip state as what those platforms actually do (manual instructions suppressed). - M5: sh bundle lives inside a private mktemp -d, one rm -rf on every exit, and a mktemp failure degrades to the fallback instead of aborting. - M6: README, verifying-your-install, environment-variables, and installation docs updated for the credential-free path and skip flags. - M7: ps1/cmd extraction replaced with a byte-exact string scanner (no ConvertFrom/ConvertTo round trip, immune to DateTime coercion), with PowerShell-driven unit tests over the captured real attestations response plus synthetic DateTime and brace-in-string controls. - Lows: fail-closed abort pinned by a functional test; TUF beats auth in cmd classification (matches sh/ps1); skipped-state output mentions the shared ~/.agents/skills; Gemini summary is skip-aware and gains an honest not-detected state; --skip-opencode do-not-write switch added (flag, env var, config key) across all three scripts. * fix(install): review round three (EncodedCommand fetcher, mutation-proof fail-closed test, lows) - R1: install.cmd's attestation fetcher no longer touches disk. The %RANDOM%-named %TEMP% .ps1 (predictable-path code execution, the M5 class escalated) is replaced by powershell -NoProfile -EncodedCommand with a base64(UTF-16LE) payload defined next to its full REM PS: plaintext; a test decodes the blob and asserts byte equality with the documented lines plus the security-relevant shape (env-var inputs, ordinal scan, no JSON round trip, distinct exit codes). Inputs still travel via env vars. Verified end to end under pwsh: the decoded blob fetched the real attestations response, wrote 2 bundles, gh verified the real v0.25.1 binary credential-free (exit 0) and rejected a wrong binary (exit 1). - R2: the fail-closed test now asserts the output ENDS with 'Refusing to install.' - mutation-verified: with the verify-failure exit 1 deleted the mutant still exits 1 via an incidental mv failure, but the trailing mv error breaks the endsWith and the test fails; restored, it passes. - Lows: CI guard test fails loudly when process.env.CI is set and no pwsh/powershell is on PATH (scanner coverage cannot silently vanish); scanner IndexOf calls are ordinal in ps1 and the encoded cmd variant; the awk skipInstall extraction requires optional-whitespace-then-colon- then-brace after the key (string values can no longer anchor it, with non-token occurrences skipped, unit-checked against escaped-embedded payloads); install.cmd comments warn that the fallback-reason literals inside parenthesized blocks must stay parenthesis-free. |
||
|
Michael Ramos
|
84846d9b9d | chore(deps): bump @pierre/diffs to 1.3.2 (#1191) | ||
|
Michael Ramos
|
8abc685460 |
chore(deps): bump @pierre/diffs to 1.3.1 with @pierre/theme 2.0.0 and @pierre/theming 1.0.0 (#1190)
Retune buildLineBgOverrides for the 1.3.x hover pipeline: per-selector hover mix rules are gone; hover is now one central rule mixing the active-line bg 97% (light) / 91% (dark) toward --diffs-hover-mix-target. Emitted hover --mix-* values are divided by those factors so the final rendered hover bg shares match 1.2.12 exactly at normal and strong, and subtle pins the 1.2.x hover finals (deletion 80/75, addition 80/70). Also add @pierre/theme and @pierre/theming to the bunfig minimumReleaseAge excludes (review follow-up from #1188). |
||
|
Michael Ramos
|
767be3bfef |
chore(deps): bump @pierre/diffs to 1.2.12 (stage 1 of 2) (#1188)
Bump the exact pin from 1.2.8 to 1.2.12 in all six package.json files (root, packages/ui, packages/server, packages/review-editor, apps/review, apps/pi-extension) and resolve the lockfile. 1.2.12 pulls in @pierre/theme 1.1.0 (minor, transitive-only; we have no direct theme dependency) and a new transitive @pierre/theming 0.0.2. Stage 2 (1.3.x, next week once aged) carries the theme major and the hover pipeline rework. Verified: typecheck, full bun test, CI DOM set, all four builds, SSR parity at 1.2.8 vs 1.2.12 (structure identical except the intentional tabindex removal from the 1.2.12 focus fix), and a fresh registry install of the pi-extension tarball resolving a working 1.2.12 (guarding against the 1.2.9-era broken-tarball failure mode, #880). |
||
|
Michael Ramos
|
46f1e8d5b2 |
fix(annotate): recognize wrapped URLs in token probe and port #1185 coverage (#1187)
Ports five small items from the closed parallel PR #1185 into the
tolerant annotate argument resolution that landed in #1183 (#1182):
- Bug fix: the token probe tested the raw token against the URL regex,
but the pipeline strips the @ reference marker and wrapping quotes
first, so a multi-token 'annotate @https://example.com/page and
summarize it' probed to nothing and emitted the handoff instead of
opening the URL. The probe now unwraps with stripAtPrefix before the
regex and returns the unwrapped form (the pipeline re-strips
harmlessly). Tests cover @-prefixed and quote-wrapped URLs as
multi-token candidates.
- Test ports: absolute-path candidate, the wider plain-text set (.txt,
.yaml) guarding ANNOTATABLE_DOC_REGEX breadth, the scoped-package
literal-@ fallback against a real @scope/ directory, and the
whole-un-split-string preference over its own tokens ('Meeting
Notes.md' wins over a resolving 'Notes.md' token) covering
annotateInputNamesExistingTarget.
- Defensive scan: the strict-mode source-scan test now asserts the
annotate startup block gates tolerance on !strictAnnotate via
isStrictAnnotateInvocation, since an inverted gate cannot be
spawn-tested without starting a server.
- DRY: the strict predicate was defined twice (strict-annotate-result
exit-code helper and the index.ts tolerance bypass). Extracted
isStrictAnnotateInvocation with a StrictAnnotateFlags type; both
sites use it so the exit-code path and the tolerance bypass can
never drift. Behavior byte-identical; existing subprocess tests
unchanged.
- Docs: the tolerant-resolution section now cites #872 (commit
|
||
|
Raúl
|
bc5470b90d |
fix(review): bound server memory for large tracked-file diffs (#1167)
* fix(review): bound server memory for large tracked-file diffs PR #1118 renders large untracked files as binary additions, but staging one moves it into the tracked `git diff` path, which had no size guard and buffered the full multi-megabyte patch (~240 MB RSS on a 51 MB text artifact). Any large tracked text file modified in the working tree hits the same unguarded path. Add a per-invocation `git -c core.bigFileThreshold=<MAX_REVIEW_FILE_CONTENT_BYTES>` prefix to every content-producing git diff, so git renders oversized blobs as "Binary files ... differ" instead of a text patch. Their bytes never enter git's diff machinery or the server's buffered stdout, mirroring the untracked-file guard. The flag is a no-op at or below the threshold, so smaller files are byte-for-byte unaffected, and the blob hash git emits in the binary diff still changes with content, so staleness detection holds. The guard is applied in the shared cores, so the Bun and Pi runtimes inherit it identically: `review-core.ts` covers the ordinary git provider (working-tree, staged, commit, and the freshness fingerprint) and `gitbutler-core.ts` covers the GitButler object diff. The jj provider runs `jj diff`, which has no `core.bigFileThreshold` equivalent, so it is out of scope here and stays unbounded as before. * fix(review): preflight oversized tracked diffs * fix(review): batch tracked diff preflight * fix(review): restore browser-safe diff core * fix(review): preserve gitlinks and textconv * fix(review): require filesystem runtime seam Fail compilation when a runtime omits file metadata or symlink support instead of silently disabling bounded reads and expansion. |
||
|
Syahrul
|
0d9128794f |
fix: release pi review ports on session shutdown (#1160)
* fix: release pi review ports on session shutdown * fix(pi): self-preempt stale sessions on fixed-port contention A session abandoned without a decision keeps listening on the fixed remote port in this long-lived process, so the next command's bind fails (#1159). On a port-in-use or range-exhausted start failure, stop every tracked browser session and retry once. Also: guard closeAllConnections for older runtimes, downgrade the superseded session notification from error to info, and cover the preemption and stop-drain paths with tests. * fix(pi): harden session preemption per independent review Age-gate the preemption sweep so a concurrent command's fresh session is never stopped, drop the session_shutdown sweep (sessions deliberately outlive in-process session replacement so cross-session feedback keeps working; process teardown frees ports anyway), make server stop() and the sweep exception-safe, register archive sessions in the registry, type the stopped error instead of matching prose, handle a stopped plan review without a false startup error, keep the tests from launching a real browser, and pin the call-site wrappings. --------- Co-authored-by: Michael Ramos <mdramos8@gmail.com> |
||
|
Michael Ramos
|
747b5ea7e6 |
fix(annotate): resolve natural-language arguments or hand off to the agent (#1183)
* fix(annotate): resolve natural-language arguments or hand off to the agent Claude Code skills run the CLI through a bash-substitution prefix that executes before the model sees anything, so any trailing natural language in /plannotator-annotate died with 'File not found: the'. Worse, a non-zero exit from that prefix aborts the whole prompt before the model runs (verified empirically), so the error was never even visible to the agent. Three-tier resolution in the binary's annotate argument handling, shared by every host via packages/shared/annotate-target.ts: 1. Fast path: probe each whitespace-delimited token; exactly one naming an existing file, URL, or folder proceeds with it directly. 2. Ambiguity: two or more tokens resolve; error naming every candidate, never guess. 3. Handoff: nothing resolves; emit an agent-addressed message echoing the words tried and asking the reading agent to interpret the request and re-run with a concrete target, preserving flags. In plain mode it lands on stdout with exit 0, the only combination that reaches the model through the bang prefix; in --json/--hook mode it goes to stderr with exit 1 so machine stdout stays clean. Single-token invocations run the unchanged pipeline first, so bare correct invocations are byte-identical. Strict gates (--require-approval or --result-file) bypass the tolerance entirely: a typo'd path stays a startup failure with exit 2 and no agent-facing prose. The CLI resolution pipeline moves to apps/hook/server/annotate-resolution.ts (returns typed outcomes instead of exiting) so the token fallback can run it once with a selected candidate; OpenCode and Pi wire the same shared selection into their own not-found paths. Skill bodies gain one line telling the agent to re-run with a concrete target when the command reports unresolvable arguments. Closes #1182 Reported-by: @technicalpickles * fix(annotate): harden tolerant resolution per review Review fixes for the three-tier annotate argument handling: - A single unresolvable token now falls through to the legacy pipeline verbatim: 'annotate nope.md' is exit 1 with 'File not found: nope.md' again in every non-strict mode, instead of an exit-0 handoff that fail-opened scripts gating on the exit code. The handoff fires only when two or more words resolve to nothing. - Unrecognized dash-prefixed tokens disable tolerance instead of being skipped, so a typo'd flag ('--no-jna') errors the way it did on base rather than silently fetching via Jina. Known flags are stripped before selection as before. - Token selection now receives the original argv tokens, so a quoted missing path ('my notes.md') is probed as one token and can never be re-split into a silently resolving 'notes.md'. - Bare directory names only count as fast-path candidates when they are the sole argument; a stray word matching a directory (or '.') hands off instead of opening folder mode. Explicit paths like 'src/' keep resolving, and the bare-existence probe fallback is file-only. - The handoff re-run suggestion echoes content flags only (--markdown, --no-jina, --render-html), never transport flags (--gate, --json, --hook). - New subprocess suite (annotate-cli.test.ts) spawns the real CLI entry and pins the contract: single-token typo exit 1, strict invocations (--require-approval and --result-file) exit 2 with empty stdout and no handoff prose, unknown-flag error, quoted-token preservation, and the directory-hijack case. Placeholder dist files are created when a build is absent so the suite runs in CI. - The copilot and gemini annotate command bodies gain the same handoff instruction as the Claude, core, and kiro skills. - AGENTS.md documents the three tiers under Annotate Flow and corrects the strict-section sentences that claimed non-strict behavior was fully unchanged; the marketing annotate doc mentions the tolerant arguments. Refs #1182 |
||
|
Michael Ramos
|
2ab377f77d | fix(uninstall): require host cleanup (#1177) | ||
|
Michael Ramos
|
0935cf08ca | fix(marketing): avoid sharing article redirect chain (#1180) | ||
|
Michael Ramos
|
93b66e0ab2 |
feat(cli): add safe uninstall lifecycle (#1170)
* feat(cli): add safe uninstall lifecycle * fix(uninstall): harden cleanup and add Windows QA * fix(uninstall): detach Windows self-delete worker * fix(uninstall): preserve PowerShell worker syntax * fix(uninstall): harden purge and host recovery * fix(uninstall): revalidate purge boundary * fix(uninstall): unlink managed link entries safely |
||
|
Michael Ramos
|
d53cbfb373 |
fix(annotate): enforce archive read-only surfaces (#1171)
* fix(annotate): enforce archive read-only surfaces * fix(archive): close remaining read-only leaks |
||
|
Michael Ramos
|
7cd023cbc7 |
docs: correct privacy and network claims (#1163)
* docs: correct privacy and network claims * docs: address privacy review findings * docs: clarify GitLab avatar lookup concurrency |
||
|
Michael Ramos
|
8f84852f97 |
fix(review): surface partial GitLab comment submissions (#1164)
Surface partial GitLab submission outcomes and preserve narrowed, duplicate-safe retries across dialog reopen and same-tab refresh. Follow-up for an explicit blocked-recovery escape: #1166. |
||
|
tbontb
|
07c89ff145 |
fix(install): authenticate api.github.com call to avoid 60/hr rate limit (#1157)
* fix(install): authenticate api.github.com call to avoid 60/hr rate limit All three installers called api.github.com/repos/.../releases/latest unauthenticated to resolve the latest release tag. GitHub caps unauthenticated REST API requests at 60/hour per source IP (shared across NAT/CGNAT/corporate proxy users), causing opaque 'Failed to fetch latest version' errors. Each installer now detects a token (precedence: GITHUB_TOKEN env > GH_TOKEN env > 'gh auth token') and attaches it as an Authorization: Bearer header to ONLY the version-resolution API call. Falls back to anonymous (unchanged behavior) when no token is available. Release downloads and git clone are left unauthenticated - they are not subject to the 60/hr REST limit. Token variables are cleared from memory immediately after the API call (defense in depth). install.ps1 wraps the API call in try/catch to surface a friendlier error instead of a raw terminating exception under $ErrorActionPreference=Stop. Docs updated with a rate-limited-IP install section and a troubleshooting entry. Parity + negative-assertion tests added. Closes #1156 * fix(install): retry anonymously when authenticated api call fails Addresses review feedback on #1157: a stale/revoked token (expired GITHUB_TOKEN lingering in CI images, dotfiles, direnv) gets a 401 and would break an install that works fine anonymously today. Each installer now retries the api.github.com version-resolution call without the auth header when the authenticated attempt fails: - install.sh: `|| true` + `[ -z "$latest_tag" ]` gate - install.ps1: try/catch, retries with no -Headers on catch - install.cmd: ERRORLEVEL check, retries without !GH_AUTH_HEADER! A bad token costs one extra request; nobody is worse off than today. New test asserts the retry path exists in all three installers (89 pass). Also corrects the AI disclosure: the assistant was opencode, not Claude. * fix(install): address second-review findings Errorlevel check adjacency in install.cmd, delayed-expansion token reads, token charset guard, PATH-resolved gh with --hostname github.com, 401-only anonymous retry in install.sh and install.ps1, preserved original errors in install.ps1, honest cleanup comments, non-vacuous test assertions, and doc corrections (1000/hr Actions token, zero-scope guidance, -Version for PowerShell). --------- Co-authored-by: tbontb-iaq <tbontb-iaq@users.noreply.github.com> Co-authored-by: Michael Ramos <mdramos8@gmail.com> |
||
|
Michael Ramos
|
b5cf065cba | chore: bump version to 0.25.1 | ||
|
Michael Ramos
|
c353413b55 |
fix(hook): enable the annotate client lease on OpenCode's last-message bridge
#1143 wired abandoned-gate dismissal into three of the four startAnnotateServer call sites. The OpenCode annotate-last bridge takes gate from stdin JSON rather than CLI flags and was missed, so /plannotator-last --gate under OpenCode still hung on waitForDecision forever once every review tab was abandoned: exactly the hang that commit set out to close. The bridge's inputs map onto the same predicate the other three use: gate from the stdin payload, json unconditionally true because emitOpenCodeAnnotateOutcome is the branch's only output path and always writes a structured record the bridge parses back, hook false because no flags are parsed here. The new test scans every startAnnotateServer call site in index.ts rather than pinning this one line, so the next site added cannot repeat the omission. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS |
||
|
Michael Ramos
|
4c89418bad |
fix(pi): contain the pi-todos mirror inside the project root
The mirror resolved <cwd>/.pi/todos with a plain path.join, so a repo that ships .pi/todos as a symlink pointing elsewhere had its plan checklist written to the symlink target on the first plan approval. The mirror is on by default, so checking out a hostile repo and approving a plan was enough. Write-only and non-overwriting, but not something to ship knowingly. The repo-implied path must now realpath inside cwd or the provider reads as absent, which is the documented no-mirror no-op. An explicit PI_TODO_PATH stays trusted verbatim: that is the user's own choice at the same trust level as config.json, and pi-todos honors it the same way. Containment reuses the shared isWithinDirectory sink, resolved through the deepest existing ancestor so a not-yet-created todos dir under a symlinked .pi cannot be created outside either. Also corrects the detection docstring: pi-todos calls ensureTodosDir on session_start, so an installed-but-unused provider does not read as absent. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS |
||
|
Raúl
|
37acde15d5 |
fix(ai): defer Codex model discovery until a Codex session starts (#1145)
* fix(ai): defer Codex model discovery until a Codex session starts Opening any plan, annotate, or code review builds the shared AI runtime, and the runtime called every provider's fetchModels() while constructing itself. For the Codex provider that starts a throwaway `codex app-server` process, so a review launched Codex even when the user never opened Ask AI. On macOS with a quarantined Homebrew Codex payload this surfaces as a Gatekeeper confirmation dialog in front of the review the user actually asked for. Codex discovery now runs on explicit activation instead. The provider is still registered and still advertised through /api/ai/capabilities using its static fallback model metadata, so nothing about discovery is user-visible until a session is created for it. createBestEffortOnce() memoizes the discovery call so it runs at most once per runtime and a failure leaves the static fallback in place rather than blocking session creation. /api/ai/session gained a beforeProviderSession hook, invoked for the resolved provider id before the session is created. /api/ai/capabilities deliberately does not invoke it: the editor probes capabilities automatically on load, so activating a provider there would reintroduce the same eager launch through a different path. Because discovery can replace the provider's model list, the session handler compares the requested model against the pre-activation default. A caller that sent no model, or sent the pre-activation default, gets the post-activation default; an explicitly chosen model is always honored. Without this a first Codex session would pin the static fallback model that discovery just replaced. Both runtimes are changed the same way, and the other providers keep their existing eager discovery, which beforeCapabilities still awaits. Tests cover the regression with a fake Codex executable rather than a real one: runtime construction and a capabilities probe must not invoke discovery, the first Codex session must, the second must not, and a failing discovery must still create a session on the fallback metadata. * fix(ai): refresh provider metadata on explicit activation Follow-up to the deferred Codex discovery change, addressing the review findings on #1145 while keeping the deferral intact: constructing the runtime and probing /api/ai/capabilities still never spawns `codex app-server`. - /api/ai/capabilities now accepts ?activate=<providerId>: it runs the same createBestEffortOnce initializer the session path uses (no second discovery path) and responds with the refreshed capabilities payload. A plain capabilities probe still activates nothing. Both runtimes get this through the shared endpoint (packages/ai is vendored into the Pi server by vendor.sh). - The apps activate the selected provider on explicit user gestures -- opening the Ask AI surface or switching the provider picker -- via the new useAIProviderActivation hook (single-flight per provider id), then merge the refreshed models and reasoning efforts into state so the model picker and per-model reasoning-effort selector populate past the static fallback. (review finding 1) - A resolver-derived model is no longer persisted: useAIProviderConfig and AISettingsTab write the per-provider model preference only on an explicit user pick, so a saved Codex model the pre-activation fallback list doesn't include survives instead of being clobbered by the fallback id. The session request still falls back; the cookie doesn't. (review finding 2) - The session handler resolves the requested model by membership in the post-activation model list instead of comparing against the pre-activation default, so sessions after the first can no longer pin a stale fallback id that discovery already replaced. (review finding 3) Tests: activation endpoint behavior (shared endpoints plus both runtimes against a hermetic fake codex on PATH), effectiveModel membership resolution, and saved-preference no-clobber (DOM tests for useAIProviderConfig persistence). Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS --------- Co-authored-by: Michael Ramos <mdramos8@gmail.com> |
||
|
Michael Ramos
|
383ac97a8e |
fix(pi): survive pi's stale-ctx guard in the post-approval continuation
Pi >= 0.69 invalidates an extension's ctx on session teardown or replacement, and every guarded member throws thereafter. The plan continuation timer polled ctx.isIdle() from a setTimeout, so a session dying mid-poll (print-mode teardown, /new or /reload right after approval) crashed the whole pi process with an uncaught exception. Cancel the continuation instead: a sessionAlive latch off the sanctioned session_shutdown event plus an isCtxAlive() probe, and the timer body never rethrows. Also .catch() the fire-and-forget browser open, which could crash pi via unhandled rejection on a failed launcher even without staleness. Fixes #1140 Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS |
||
|
Raúl
|
c750427ab8 |
feat(annotate): dismiss abandoned gate sessions (#1143)
A direct local `plannotator annotate --gate --json` waits for one authoritative decision. If every review surface disappears without approving, sending feedback, or exiting, the caller blocks forever: the server has no notion of whether a client ever connected, whether another tab is still open, or whether a disconnect is a reload. Page lifecycle events cannot answer that. `pagehide` and `beforeunload` also fire on reload and navigation, so dismissing from them ends reviews the user expects to resume. Use connection presence instead, which is exactly what the transport can observe. Local direct structured gates advertise a client lease in /api/plan and serve /api/annotate/client-lease as SSE. One open stream is one connected review surface. The server heartbeats every 5s and, only after at least one client has connected, starts a 30s reconnect grace when the last one disconnects. A reconnect inside the grace continues the same review; expiry resolves the gate through the same path as explicit Close, so it produces an ordinary `dismissed` decision and inherits the strict-result contract unchanged. Approve, feedback, explicit exit, and server stop all cancel a pending expiry. Presence lives in two runtime-independent pieces so Bun and Pi cannot drift. createAnnotateClientLeaseTracker owns first-client, active-count, reconnect, cancellation, and one-shot expiry. createAnnotateClientLease- StreamSession owns one connected client: acquire the slot, write the ready comment, heartbeat, release exactly once. Each server passes only its own write primitive (a ReadableStream controller for Bun, res.write for Pi). A write that fails closes the session, because a stream that can no longer be written to is a client that is no longer present; holding the slot there would make the gate un-dismissable for the rest of the run, which is reachable only through a half-open connection and so is covered by unit tests rather than an integration test. Scope is deliberately narrow. The capability stays off for remote and shared sessions, where tunnel disconnects would read as abandonment, and off for hook transport, legacy plaintext, archive, plan, review, and folder-picker sessions. A session that never receives its first client never auto-dismisses, so browser-launch failures still need a caller-side timeout. Decision settlement is explicit for the same reason: a connected surface and the lease can both try to settle the session, and the awaited promise ignoring the second resolve was not enough. The loser still deleted the reviewer's draft and answered ok, so a tab reported success for a decision the caller never received. createAnnotateDecisionSettler makes the winner explicit; a loser changes nothing and answers 409. Expiry deliberately keeps the saved draft, unlike explicit Close, so an abandoned review stays recoverable. Stopping the server closes live lease streams instead of only releasing their slots, so a long-lived host process does not retain a heartbeat timer and an open response for every finished session. |
||
|
mararn1618
|
5d1544fa32 |
fix(hook): route annotate-last to the live Copilot CLI session (#1150)
* fix(hook): add Copilot session lock detection Copilot CLI exports no identifying environment variable, so nothing distinguishes a Copilot session from a plain shell. Match ancestor pids against session-state inuse locks to find the live session, and only accept a match when the lock owner still names a copilot process, since locks can outlive their session and pids get reused. * fix(hook): route annotate-last to the live Copilot session Under Copilot CLI, annotate-last silently fell back to the default transcript reader and annotated a message from a different tool. Take the Copilot branch when an ancestor process holds a session lock, or when PLANNOTATOR_ORIGIN=copilot-cli is set with the cwd heuristic as fallback, and report origin copilot-cli to the annotate server. * fix(hook): prefer ancestor lock match in copilot-last The cwd heuristic can pick a stale session when several exist for one repo. Resolve the session locked by an ancestor copilot process first and keep the heuristic as fallback. * docs(cli): document copilot-last in help The subcommand worked but was missing from the top-level usage and the per-subcommand help map. |
||
|
Souptik Chakraborty
|
26681abb8c |
fix(amp,droid): surface Approve-with-Notes feedback instead of dropping it (#1146)
The Approve with Notes flow (#1092) emits decision:"approved" with a feedback field, but both adapters collapsed every approved decision to a bare "Approved." -- the reviewer's notes were silently lost. Amp had two drop sites: formatAnnotationFeedback returned null for anything not "annotated", and handleAnnotateResult early-returned on approved before formatting was ever reached. Droid's emitAnnotateDecision printed only "Approved." Both adapters now emit the shared approved-with-notes prompt (mirroring DEFAULT_ANNOTATE_APPROVED_WITH_NOTES_PROMPT in packages/shared/prompts.ts); note-less approvals keep the old behavior. The amp config override key approvedWithNotes matches the shared prompt config shape. Fixes #1137 |
||
|
BrandonNoad
|
8d8e643976 |
fix(hook): read annotate-last from Claude Code's transcript tree, not file order (#1141)
* fix(hook): read annotate-last from the transcript tree, not file order Claude Code session logs are append-only and tree-shaped: every entry records the entry it follows in `parentUuid`. `/rewind` writes nothing at all. The next committed message simply re-parents to an earlier entry, leaving everything after it orphaned in the file forever. `extractRecentRenderedMessages` scanned bottom-up in file order, so those orphans were still offered in the annotate-last message picker even though they are no longer part of the conversation. On a rewound session in this repo the picker listed 10 messages where only 7 are live. Add `resolveActiveBranchIndices`, which walks `parentUuid` from the newest id-bearing entry back to the root. It returns indices rather than a filtered array so callers keep reporting real file line numbers, and returns null on a chain it cannot trust (no ids, dangling parent, cycle) so callers degrade to the previous file-order read instead of returning nothing. Note the newest entry is not always the last line: `last-prompt`, `ai-title`, `mode` and `file-history-snapshot` carry no ids and are often written last. Opt in at the Claude Code call site only. Droid's call site, Codex and Copilot (separate parsers), and Pi/OpenCode/Amp (live APIs, no transcript reads) are all unaffected. Both new parameters default to off. Checked against 311 local transcripts: every one walks cleanly to the root with no dangling parents or cycles, and the default pick is byte-identical under both readings in all 298 that contain a message. That is expected, since a committed rewind's new branch is always the newest lines in the file. The picker is where the difference shows up. Test fixtures previously assigned random `parentUuid`s, which left every entry an orphan and made branch resolution untestable, so `buildLog` now links them into a real chain and `buildRewoundLog` models a fork. * fix(hook): fail open when the active branch has no assistant messages A /compact boundary is written with parentUuid: null, so it is a tree root: the active-branch walk stops there and a freshly-compacted session yields zero messages. Callers treat an empty result as "wrong log file" and walk off to an older session, so fail open to the file-order read instead. Adds tests for the compaction cut, the fallback, and the post-compaction recovery. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS --------- Co-authored-by: Michael Ramos <mdramos8@gmail.com> |
||
|
Michael Ramos
|
b752080204 |
feat(ai): add Claude Opus 5 across providers and review agents (#1151)
* feat(ai): add Claude Opus 5 to the model catalogs Adds `claude-opus-5` to both static Claude model lists: - `packages/ai/providers/claude-agent-sdk.ts` — the Ask AI picker - `packages/ui/components/AgentsTab.tsx` — the review-agent picker, which `TOUR_CLAUDE_MODELS` spreads, so tour and guide pick it up too No `claude-opus-5[1m]` variant: the `[1m]` suffix opts the 4.x models into the larger context window, and the 5-series models are already 1M by default — which is why the existing `claude-fable-5` and `claude-sonnet-5` entries carry no variant either. A test pins that invariant. Guide, tour, and marker review need no change — they pass the picked model straight through to the CLI with no server-side allowlist. Tests cover the Bedrock/Vertex family matcher accepting the new bare alias (a matcher keyed on "opus-4" would have silently dropped it) and pin the existing Fable fallthrough, which has no ANTHROPIC_DEFAULT_*_MODEL of its own. Also refreshes the marketing Ask AI model list, which had gone stale at Sonnet 4.6 / Opus 4.6 / Haiku 4.5. * feat(ai): default Claude review agents to Opus 5 Bumps the two Claude review-agent defaults from Opus 4.7 to Opus 5: - `DEFAULT_CLAUDE_MODEL` (packages/ui/hooks/useAgentSettings.ts) — the review-agent picker's default for users with no saved cookie preference - `buildClaudeCommand`'s default parameter (packages/server/claude-review.ts) — the server-side fallback when a job arrives with no model BEHAVIOR CHANGE: users who have never picked a Claude review model move from Opus 4.7 to Opus 5. Saved picks are untouched — the cookie already holds an explicit model for anyone who has chosen one. Kept deliberately separate from the catalog commit so it can be reverted on its own if the maintainer prefers to ship the new models as options only. Ask AI's default (`DEFAULT_MODEL` in claude-agent-sdk.ts) stays on Sonnet 5, and the tour/guide defaults stay on the `sonnet` latest-alias — those are interactive, latency-sensitive surfaces where the cheaper tier is the intentional choice. |
||
|
jms830
|
a54b46bbee |
feat(pi): mirror plan checklist to pi-todos (#1139)
* feat(pi): mirror plan checklist into an editable todo provider Closes the one-way half of #484: on plan approval, feature-detect an editable todo provider and mirror the approved checklist into it, then reflect [DONE:n] markers onto todo status as execution proceeds. pi-todos is the first provider. It exposes no API to other extensions, so the integration surface is its on-disk format (.pi/todos/<id>.md with JSON front-matter, <id>.lock taken with O_EXCL). The provider writes through that contract, reconciles by tag so repeated syncs are idempotent, closes steps dropped from an edited plan, and skips any todo another session holds a lock on. Provider state resets on return to idle, so a second plan re-detects rather than inheriting the first plan's decision. The mirror is additive: the existing progress widget is untouched. pi-todos has no live surface of its own -- its list renders on demand in /todos -- so suppressing the widget would trade a visible tracker for files behind a keystroke. Sync is one-way, so provider-side edits can never desync plan execution. oh-my-pi's native todo panel is the obvious second provider and slots in behind the same interface, but is not implementable from an extension today: the panel repaint is gated to the built-in `todo` tool, and the session handed to an extension-registered tool is a read-only projection without the todo accessors. Details in todo-providers/index.ts. Gated by PLANNOTATOR_TODO_PROVIDER / config.todoProvider, and inert when no provider is detected. * fix(pi): harden todo provider sync --------- Co-authored-by: jms830 <jms830@noreply.github.com> |
||
|
Michael Ramos
|
68c1291cd7 |
chore: bump version to 0.25.0
Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS |
||
|
Michael Ramos
|
951175ec1c |
fix(pi): ship annotate-outcome.ts in the npm package
The #1092 approval-notes work added annotate-outcome.ts as a top-level import of index.ts but the package.json files array was never updated, so the published tarball failed to load under Pi's jiti loader for every npm-installed user. Found by the pre-release QA sweep; verified the packed tarball now contains the module and no other top-level import is missing from the files array. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS |
||
|
Reveck
|
26a5314a99 |
feat(pi): add external plan execution handoff (#1126)
* feat(pi): add external plan execution handoff * fix(pi): release phase-added tools on external plan handoff #1124 made every idle-exit path hand back the tools the phase added, but the external handoff path introduced here restored the saved model and thinking level without releasing them. Approving a plan in executionMode "external" therefore leaked the planning tool scope (grep, find, ls, plannotator_submit_plan) into the idle session. Extract the shared exit sequence into returnToIdle() and use it from all three idle transitions (manual exit, execution completion, handoff), so a new exit path cannot miss a step again. Callers keep their own messaging, session entries, and events. Also add a default-mode parity test: with executionMode unset, approval enters executing, emits no plannotator:plan-approved, appends the plannotator-execute entry, and still sends the continuation message. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS * chore(pi): warn on unknown executionMode and document the handoff payload An unrecognized executionMode was silently dropped, so a typo like "handoff" left the session in automatic execution with no explanation. Warn through the existing config warning channel (the same one malformed JSON uses) and keep the automatic fallback. README: note that planFilePath is relative to cwd and should be resolved against it, and that companions can subscribe with the raw channel string without importing Plannotator internals, matching how the plannotator:request channel is already described. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS --------- Co-authored-by: Michael Ramos <support@plannotator.ai> |
||
|
Peete
|
d26289060d |
fix(pi): preserve extension tool selections (#1124)
* fix(pi): preserve extension tool selections * fix(pi): release phase tools when the plan completes The agent_end plan-completion path dropped back to idle via restoreSavedState but never called releaseAddedPhaseTools(), so tools Plannotator activated for the executing phase stayed active in idle and a stale phaseAddedTools list was persisted. Mirror exitToIdle. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS * fix(pi): keep grep/find/ls in the default planning tool set Dropping them from the planning phase profile silently removed native search from default planning sessions. Restore them as default planning phase tools so they now flow through the new tracking mechanism: they are recorded in phaseAddedTools and released on phase exit, and a discovery tool another extension already activated is neither claimed nor removed. A user's own planning.activeTools still overrides them. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS * fix(pi): always keep the submit tool active during planning Planning tools now come purely from the resolved profile's activeTools, which uses replace semantics: a phases.planning.activeTools in ~/.pi/agent/plannotator.json or cwd/.pi/plannotator.json supersedes the built-in list wholesale. Pi's setActiveToolsByName then replaces the session's tool list outright, so such a config silently dropped plannotator_submit_plan while the planning system prompt kept telling the model to call it - a dead end with no way to submit a plan. Union PLAN_SUBMIT_TOOL into the configured list for the planning phase only (deduped if the config already lists it). It flows through phaseAddedTools like any other addition, so it is released normally on phase exit. Other tools are left to the user: overriding planning.activeTools is a deliberate choice of tool set, and only the submit tool is contractually required for the phase to function. Also corrects the README's "extra tools ... additive" wording, which described the old merge behavior. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS --------- Co-authored-by: Michael Ramos <support@plannotator.ai> Co-authored-by: Michael Ramos <mdramos8@gmail.com> |
||
|
Brad Beebe
|
ac3a84ba5a |
fix(opencode): Fix hardcoding default build OpenCode agent when sending responses (#1131)
* fix(opencode): default agent switching to disabled
* fix(opencode): keep plan-approval build handoff; default no-switch for review feedback only
The agent switch cookie is shared by plan approval and code review, so
flipping the stored default to `disabled` also removed OpenCode's
plan-approval hand-off for every user who never configured the setting.
Make the unset default surface-aware instead: `getAgentSwitchSettings('plan')`
keeps the historical build hand-off, `getAgentSwitchSettings('review')`
stays on the current agent. An explicit user choice still applies to both
surfaces. Settings and the agent warning resolve the default from the mode
they render in, and the OpenCode "agent not available" warning now names
plan approval on the plan path.
Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS
---------
Co-authored-by: Michael Ramos <mdramos8@gmail.com>
|
||
|
Raúl
|
9a450a69e7 |
feat(annotate): preserve notes on structured approval (#1092)
* feat(annotate): add strict atomic result output * feat(annotate): exit 2 for strict-gate usage and publication errors Adopt the grep convention for the strict annotate gate's exit codes: 0 = approved, 1 = negative human outcome (annotated/dismissed under --require-approval), 2 = the gate itself was misconfigured or could not start/deliver a decision. Previously all usage/startup/validation failures shared exit 1 with "reviewer did not approve", so callers could not tell a denied review from a broken gate. - parseStrictAnnotateOptions failures (bad flag combos, strict flags outside annotate --gate --json) now exit 2 - --result-file preflight failures (missing parent, pre-existing or dangling-symlink destination) now exit 2 - post-decision publication failures (destination raced into existence, hard links unavailable, stdout write failure) now exit 2: they deliver no decision record at all, so the code's own fail-closed handling presents them as environment errors, never as a reviewer outcome -- and never approval, since only 0 means approved - decision outcomes keep 0/1 exactly as before; signal deaths keep 128+n - document the contract in AGENTS.md and the annotate-gates guide Claude-Session: https://claude.ai/code/session_01YXkgsNucxDwAL4GdR4XYRk * feat(annotate): preserve notes on structured approval * test(pi): use exact annotate outcome import * fix(annotate): exit 2 for strict-gate startup failures The six startup-failure sites in the annotate path (missing path, unreachable URL, empty folder, ambiguous name, missing/unsupported file, oversized file) run after flag parsing and exited 1. Under --require-approval / --result-file, 1 is the "reviewer requested changes" signal, so a typo'd path made automation misclassify a configuration error as a legitimate rejection. Route those sites through exitAnnotateStartupFailure(), which picks its code from the already-parsed strict options via the new pure helper annotateStartupFailureExitCode(). Non-strict invocations still exit 1 with byte-identical stderr; strict invocations exit STRICT_GATE_ERROR_EXIT_CODE (2). Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS * fix(annotate): emit the strict decision on stdout before publishing it writeResultFile ran before the decision JSON reached stdout. On a filesystem without hard links (exFAT, FAT32, most SMB/NFS, some container bind mounts) publication fails deterministically, the catch exited 2 with nothing written anywhere — and the reviewer's autosaved draft had already been deleted by the feedback flow, so their completed decision was lost. Emit the stdout record first, then publish the result file. Exit semantics are unchanged: a publication failure still exits 2, but the decision has reached stdout by then. Only a stdout write failure now leaves no record at all. Correct the docs and comments that claimed exit 2 delivers no decision record: it means the result *file* was not published. Also document the two publication caveats: the 0600 mode is a no-op on Windows, and the atomic link/rename is not followed by a parent-directory fsync, so publication is atomic but not crash-durable. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS * fix(annotate): parse linked docs with the render-side frontmatter rule on export buildCompleteAnnotateFeedback re-parsed each linked document with parseMarkdownToBlocks(entry.markdown) — no options, so frontmatter stripping defaulted on. The render side parses with { frontmatter: shouldStripFrontmatter(path) }. For plain-text linked docs (.yaml/.json/.toml/…) a leading `---` is real content, not frontmatter: a multi-document YAML opens with it. Stripping it on the export side shifted every block id, so ordinary Send Feedback and deny emitted wrong `(line N)` labels — or dropped them entirely when the annotation's block no longer existed. Pass the same shouldStripFrontmatter(filepath) option at the export call site so both sides agree. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS * fix(annotate): carry the message scope through approve-with-notes /api/feedback forwards selectedMessageId and feedbackScope; /api/approve dropped them. Pi resolves the anchor message from those fields, so notes delivered on the approve path anchored to the last message instead of the one the reviewer picked in a multi-message annotate-last session — while Send Feedback in the same session anchored correctly. Forward both fields on the approve path in the Bun and Pi servers, and have the client build the approval body with the same scope resolution Send Feedback uses (extracted as getFeedbackMessageScope so the two can no longer drift). Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS * docs(annotate): tell agents an approval may carry notes The skill and slash-command files still described `"decision": "approved"` as "acknowledge and stop", with no mention of the feedback field the gate can now attach — so an agent reading them would silently drop the reviewer's approval notes. Update the Claude core/claude skills, the Copilot commands, the Gemini annotate command, and the annotate command reference so the approved branch names the optional feedback field and says what to do with it: carry it into subsequent work, do not treat it as a change request. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS * docs(annotate): document the real approvedWithNotes default The default annotate.approvedWithNotes template is `{{contextBlock}}{{feedback}}`, not `{{context}}` on its own line, and {{contextBlock}} was missing from the variable table entirely. Show the actual default, add {{contextBlock}} to the variable table, and explain why the default prefers it: it collapses to nothing for message annotations instead of leaving a stray blank line. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS --------- Co-authored-by: Michael Ramos <mdramos8@gmail.com> |
||
|
Raúl
|
5aaa420080 |
feat(annotate): add strict atomic result output (#1091)
* feat(annotate): add strict atomic result output * feat(annotate): exit 2 for strict-gate usage and publication errors Adopt the grep convention for the strict annotate gate's exit codes: 0 = approved, 1 = negative human outcome (annotated/dismissed under --require-approval), 2 = the gate itself was misconfigured or could not start/deliver a decision. Previously all usage/startup/validation failures shared exit 1 with "reviewer did not approve", so callers could not tell a denied review from a broken gate. - parseStrictAnnotateOptions failures (bad flag combos, strict flags outside annotate --gate --json) now exit 2 - --result-file preflight failures (missing parent, pre-existing or dangling-symlink destination) now exit 2 - post-decision publication failures (destination raced into existence, hard links unavailable, stdout write failure) now exit 2: they deliver no decision record at all, so the code's own fail-closed handling presents them as environment errors, never as a reviewer outcome -- and never approval, since only 0 means approved - decision outcomes keep 0/1 exactly as before; signal deaths keep 128+n - document the contract in AGENTS.md and the annotate-gates guide Claude-Session: https://claude.ai/code/session_01YXkgsNucxDwAL4GdR4XYRk * fix(annotate): exit 2 for strict-gate startup failures The six startup-failure sites in the annotate path (missing path, unreachable URL, empty folder, ambiguous name, missing/unsupported file, oversized file) run after flag parsing and exited 1. Under --require-approval / --result-file, 1 is the "reviewer requested changes" signal, so a typo'd path made automation misclassify a configuration error as a legitimate rejection. Route those sites through exitAnnotateStartupFailure(), which picks its code from the already-parsed strict options via the new pure helper annotateStartupFailureExitCode(). Non-strict invocations still exit 1 with byte-identical stderr; strict invocations exit STRICT_GATE_ERROR_EXIT_CODE (2). Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS * fix(annotate): emit the strict decision on stdout before publishing it writeResultFile ran before the decision JSON reached stdout. On a filesystem without hard links (exFAT, FAT32, most SMB/NFS, some container bind mounts) publication fails deterministically, the catch exited 2 with nothing written anywhere — and the reviewer's autosaved draft had already been deleted by the feedback flow, so their completed decision was lost. Emit the stdout record first, then publish the result file. Exit semantics are unchanged: a publication failure still exits 2, but the decision has reached stdout by then. Only a stdout write failure now leaves no record at all. Correct the docs and comments that claimed exit 2 delivers no decision record: it means the result *file* was not published. Also document the two publication caveats: the 0600 mode is a no-op on Windows, and the atomic link/rename is not followed by a parent-directory fsync, so publication is atomic but not crash-durable. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS --------- Co-authored-by: Michael Ramos <mdramos8@gmail.com> |
||
|
Raúl
|
53650f3f6b |
fix(annotate): watch open source files exactly (#1089)
* fix(annotate): watch open source files exactly * test(annotate): cover atomic watcher saves * fix(watch): survive atomic file replacement * fix(watch): disable exact-file coalescing * fix(watch): track exact file signatures * fix(annotate): tolerate undefined watcher filenames and harden watch callbacks The exact-file watcher only treated a `null` filename as "name unavailable". On Linux, Bun's fs.watch delivers `filename === undefined` for events on the watched directory itself (chmod/utimes/rename of the parent, as produced by `tar -x`, `rsync -a`, `cp -a`), so `filename.toString()` threw an uncaught TypeError and killed the annotate server for every Linux user with a watched file open. Widen the guard to `filename == null` (null and undefined) and move the listener into `createExactFileWatchListener`, whose body is wrapped in try/catch so no watcher event can ever take the server down. Mirrored in the Pi runtime, with regression tests in both. Claude-Session: https://claude.ai/code/session_01H5KQWqXqjrPxyxUNso1QHS --------- Co-authored-by: Michael Ramos <mdramos8@gmail.com> |
||
|
Ben Newman
|
6b542da8b9 |
feat(annotate): extend per-file version diff to folder sessions (#1105)
* refactor(annotate): extract per-file version history into a shared helper Move the single-file annotate-history pipeline (slug derivation, saveToHistory, previous-version lookup, degrade-on-error) out of the Bun-specific annotate server and into packages/shared/annotate-history.ts, built on node:fs/node:path/node:crypto only so other runtimes can vendor it unmodified. annotate.ts now calls computeAnnotateHistory() instead of inlining the pipeline; behavior for single-file sessions is unchanged. * feat(annotate): extend per-file version history to folder annotate sessions Eligible folder files served through /api/doc now get snapshotted into the same version history the single-file flow uses, and their doc responses carry the same previousPlan/versionInfo/diffCurrent fields /api/plan already returns for single-file sessions. The pipeline runs lazily on first open and is memoized per resolved absolute path for the life of the server, so reopening a file never re-snapshots it. Eligibility mirrors the single-file source-save gates: a local file under the session's folder root, markdown-branch documents only (.md/.txt, not HTML, not a Turndown-converted doc), under the existing 2MB annotatable-file cap, and gated by the same annotateHistory config toggle. Storage failures degrade to a plain render (never a gate on the request) via the same try/catch computeAnnotateHistory already wraps. /api/plan/version and /api/plan/versions gain an optional path (+ base) query param so folder sessions can ask for a specific file's history; the slug is always derived server-side from the resolved, containment-checked path — never accepted from the client, since it gets joined unsanitized into a filesystem path. Omitting path keeps today's single-session-binding behavior unchanged. * test(annotate): cover folder annotate version history Adds a new describe block exercising the folder-mode history pipeline added in the previous commit: first-open snapshot + same-session memoization, storage-level dedupe, cross-mode slug continuity with the single-file flow, first-ever-open field shape, the config toggle, an ineligible (HTML) file type, degrade-on-unwritable-history-dir, and the path-parameterized version endpoints (including containment rejection and the no-path fallback). * feat(ui): add a docKey seam to usePlanDiff for per-document resets usePlanDiff's diff-base state (diffBasePlan, diffBaseVersion, versions, ...) was seeded once from its constructor args and only ever synced later via a "still falsy" guard - fine for a single root document, but switching to a different document (a different previousPlan/versionInfo) would silently keep the previous document's diff base around instead of adopting the new one's. Add an optional docKey param identifying which document the current previousPlan/versionInfo belong to. When it changes between renders, reset diffBasePlan/diffBaseVersion/versions (and in-flight loading/selecting flags) to the newly-provided values. Omitting docKey (or keeping it stable) preserves exactly today's one-time-hydration behavior, so the root document's call site is unaffected until it opts in. No caller passes docKey yet - this is purely additive. * feat(ui): carry a per-document version-diff baseline through useLinkedDoc /api/doc now returns previousPlan/versionInfo/diffCurrent for eligible folder files (same shape /api/plan already returns for single-file sessions). Extend LinkedDocLoadData with those fields and carry them through the same activate/cache/back lifecycle annotations and markdown already use, so a document's diff baseline: - is captured once when the document is first opened - persists in the per-filepath cache across back()/re-open, instead of being lost or needing a re-fetch - resolves cache-first via the new resolveDiffBaseline helper, gated on whether a baseline was ever captured (versionInfo presence) rather than truthiness of previousPlan - a document at its first-ever version legitimately caches previousPlan: null, which is a resolved fact, not a cache miss The hook exposes the active document's baseline as diffPreviousPlan/ diffVersionInfo, both null when no document is active or the active one has no eligible history (every non-folder linked doc, since /api/doc never populates these fields for those). Not yet consumed by App.tsx - purely additive. * feat(editor): render folder-doc version diffs via the active document Folder annotate's version-diff UI (inline PlanDiffViewer blocks, the +N/-M badge, and the Version Browser) was root-document-coupled: usePlanDiff was fed only the root's previousPlan/versionInfo, and every render site keyed off linkedDocHook.isActive to blank out the badge/version tab whenever any linked or folder document was open. Wire the two new per-document seams together instead: - Feed usePlanDiff the active document's own previousPlan/versionInfo/ filepath (falling back to the root document's when none is active), using the document's filepath as usePlanDiff's new docKey so switching documents resets the diff base instead of inheriting the previous one's. - Add per-document fetchers (fetchVersion/fetchVersions with &path=<filepath>) so selecting a base version or listing versions targets the active document's own history, not the session-bound bare endpoints. - Replace the root-only versionInfo/showVersionsTab reads with the active document's, so the Version Browser now reflects whichever document is on screen (previously it kept showing the root document's versions while a linked doc was open). - Drop the blanket "linkedDocHook.isActive ? null/false : ..." suppression at the Viewer callsite and in DocBadges - planDiffStats/hasPreviousVersion already resolve to the active document's own (possibly absent) diff data, so the badge now shows for folder docs with history and stays hidden for every other document exactly as it did before. Root-document behavior (single-file, plan, review, HTML surfaces) is unaffected: none of those ever set a docKey or have an eligible document history, so they fall through to the same defaults as before. * feat(pi): extend per-file version history to folder annotate sessions Mirrors the Bun runtime's folder annotate history support (packages/server/annotate.ts + reference-handlers.ts) in the Pi Node server: - Vendor the shared annotate-history helper (deriveAnnotateHistorySlug, computeAnnotateHistory) from packages/shared into generated/ via vendor.sh, and delegate the single-file version-history pipeline in serverAnnotate.ts to it instead of the hand-duplicated inline block. Behavior for single-file sessions is unchanged. - Eligible folder files served through /api/doc now get snapshotted into the same version history the single-file flow uses, and their doc responses carry the same previousPlan/versionInfo/diffCurrent fields /api/plan already returns. The pipeline runs lazily on first open and is memoized per resolved absolute path for the life of the server, so reopening a file never re-snapshots it. - /api/plan/version and /api/plan/versions gain an optional path (+ base) query param so folder sessions can ask for a specific file's history; the slug is always derived server-side from the resolved, containment-checked path (resolveAllowedDocPath in reference.ts) — never accepted from the client. * test(pi): cover folder annotate version history Adds apps/pi-extension/server/annotate-history.test.ts, the Node mirror of packages/server/annotate.test.ts's folder-history describe block: first-open snapshot + same-session memoization, cross-mode slug continuity with the single-file flow, the config toggle, an ineligible (HTML) file type, degrade-on-unwritable-history-dir, and the path-parameterized version endpoints (including containment rejection and the no-path fallback). History writes land in the real ~/.plannotator data dir rather than a per-test PLANNOTATOR_DATA_DIR override: generated/storage.js caches its data directory in a module-level constant at first import, so a per-test env var override taken after that point silently no-ops. Each test uses its own unique project namespace instead, same approach as the Bun-side suite. * ci: run docKey/linked-doc DOM tests in CI usePlanDiff.test.tsx and useLinkedDoc.test.tsx use the test.skipIf(!hasDom) pattern but were never added to the DOM_TESTS step, so they silently skipped under CI's plain `bun test` and never actually ran. * refactor(annotate): drop diffCurrent from the folder /api/doc path diffCurrent equals the document's own markdown and the client never reads it off /api/doc — it only exists on /api/plan for legacy single-file shape parity, which is untouched. Stop merging it into folder /api/doc responses and stop retaining it in the per-launch folder history memo (Bun and Pi), and drop the now-unused field from LinkedDocLoadData. - packages/server/reference-handlers.ts: new FolderAnnotateHistory type (AnnotateHistoryResult minus diffCurrent); applyDocOptions no longer copies diffCurrent onto the response - packages/server/annotate.ts: the folder memo now stores/returns only slug/previousPlan/versionInfo - apps/pi-extension/server/reference.ts + serverAnnotate.ts: mirrored changes for the Pi runtime - packages/ui/hooks/useLinkedDoc.ts: removed the unused diffCurrent field from LinkedDocLoadData * test(annotate): stop leaking history dirs; update diffCurrent expectations The folder annotate history tests (Bun and Pi) minted a fresh project namespace per test but never cleaned up, leaving hundreds of directories under the real ~/.plannotator/history over repeated runs. Track every minted project and remove its history directory in afterAll — this also covers the stray non-directory artifact the "unwritable data dir" test deliberately plants inside its own project's history dir, since removing the project dir recursively takes it with it. Also update the two assertions that expected diffCurrent on the folder /api/doc response: that field is no longer propagated on the folder path (see the preceding diffCurrent-removal commit), so both now assert its absence instead. * fix(ui): remember per-document diff-base selection across navigation usePlanDiff reset diffBasePlan/diffBaseVersion to the newly-provided document's defaults on every docKey change. That discarded a manually selected base version when navigating away from a document and back (e.g. root -> linked doc -> root), regressing behavior upstream relied on keeping (nothing reset the selection before this seam existed). Track each docKey's selection in a ref-held Map (keyed by docKey, including null for the root document) and restore it on return instead of re-seeding defaults; a key visited for the first time still seeds from its own initialPreviousPlan/versionInfo exactly as before, and selections never leak between distinct keys. Adds two DOM-gated tests: restoring a manual selection after a detour to another document, and confirming distinct docKeys don't leak into each other. * fix(annotate): match folder history eligibility to the single-file plain-text set The folder /api/doc history gate was a hardcoded /\.(md|txt)$/i in both runtimes, so any other annotatable plain-text file (.mdx, .yaml, .json, .toml, ...) opened via a folder session silently skipped snapshotting — breaking the cross-mode continuity this feature advertises (a .yaml with an existing single-file version thread showed no diff when opened via its folder). Reuse the canonical predicate instead: isAnnotatableTextPath (ANNOTATABLE_TEXT_REGEX in @plannotator/core/annotatable), the exact set the single-file pipeline snapshots. HTML stays deferred and .env stays excluded, both by that same definition. Tests extended in both runtimes: .mdx mints on first open, .yaml single-file history serves as the folder baseline, .env mints nothing, .html unchanged. * feat(ui): label the folder diff badge with its baseline The in-file version-diff badge in annotate/folder sessions shows +N/-M against the file's last-reviewed snapshot, while the git badges in the file tree count uncommitted-vs-HEAD — same numbers, different baselines. Give the badge an optional baseline suffix and tooltip override (PlanDiffBadge baselineLabel/baselineTooltip, threaded through DocBadges, Viewer, and StickyHeaderLane) and have annotate mode pass 'since last review' / 'Changes since you last reviewed this file'. Plan review passes nothing and renders byte-identically to before. DOM tests cover both the labeled and the unchanged default rendering. * fix(editor): exit diff view when the active document loses its baseline Follow-up to the per-document diff baselines: with diff view active on file A, opening a history-less file B left isPlanDiffActive latched on — the diff viewer could not render for B, but the stale flag hid the annotation toolstrip and sticky header until the user pressed Escape. Auto-exit the diff view whenever the active (non-HTML) document has no baseline. The --render-html surface is explicitly gated out: its diff view is driven by htmlDiffHtml with usePlanDiff fed nulls, so hasPreviousVersion is always false there and auto-exiting would kill the HTML diff toggle. Plan review is unaffected — the root document's baseline never goes false mid-session. DOM tests cover the exit, the keep-active document switch, the HTML gate, and the no-baseline activation snap-back. --------- Co-authored-by: Michael Ramos <mdramos8@gmail.com> |