Commit graph

318 commits

Author SHA1 Message Date
Facu_Viñas
5a2ed4eb05 feat: add startup update check with 24h cache
Queries npm registry at most once per 24h to check if a newer version
of gsd-pi is available. Displays a non-blocking banner in interactive
mode when an update exists. The check is fire-and-forget — network
errors or timeouts never block startup.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 14:28:43 -03:00
TÂCHES
1fc3a4ca77 Merge pull request #207 from gsd-build/feat/130-prefs-wizard
feat: add interactive wizard mode for /gsd prefs
2026-03-13 11:01:47 -06:00
TÂCHES
24f8cbc845 Merge pull request #206 from gsd-build/feat/129-thinking-command
feat: add /thinking slash command for toggling thinking level
2026-03-13 11:01:44 -06:00
TÂCHES
f6a942afd6 Merge pull request #205 from gsd-build/fix/177-milestone-id-generation
fix: use max-based milestone ID generation instead of length+1
2026-03-13 11:01:41 -06:00
Lex Christopherson
c622e1d3a6 fix: address audit findings for prefs wizard (#130)
- YAML-safe string quoting for values with special characters
- Preserve existing body content below frontmatter on wizard save
- Preserve existing version instead of hardcoding version: 1
- Update unknown-command fallback to include wizard|setup subcommands
- Show warning when timeout input is not a valid whole number

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:59:27 -06:00
Lex Christopherson
401397362f fix: deduplicate maxNum logic and add nextMilestoneId tests (#177)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:59:10 -06:00
Lex Christopherson
135390542a fix: handle non-thinking models correctly in /thinking command (#129)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:58:49 -06:00
Lex Christopherson
46a083fc5e feat: add interactive wizard mode for /gsd prefs (#130)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:52:55 -06:00
Lex Christopherson
bb10aacb23 feat: add /thinking slash command for toggling thinking level (#129)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:52:53 -06:00
Lex Christopherson
d2fd92f8fc fix: use max-based milestone ID generation instead of length+1 (#177)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:52:52 -06:00
TÂCHES
d9548cdf26 fix: pause auto-mode on provider errors to prevent reassess-roadmap loop (#95) (#202)
When a provider returns a fetch error, the agent_end hook now detects
stopReason === "error" and pauses auto-mode. This prevents the state
machine from silently re-dispatching the same phase until stuck
detection fires.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:39:00 -06:00
TÂCHES
7d64aac6bb fix: show TAB hint for notes input in discuss-mode survey (#192) (#203)
The "None of the above" option description said "add details in notes
below" without telling users to press TAB to reveal the notes input.
Updated the description to "Press TAB to add optional notes." and made
the footer "tab to add notes" hint always visible in single-select mode
(previously hidden until a selection was committed).

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:38:47 -06:00
Adam Dry
2ed7c830d9 fix: merge slice branches to integration branch instead of main (#200)
When working on a feature branch (e.g. f-123-new-thing), GSD creates
slice branches correctly from the current branch but merges them back
to main instead of the feature branch. This is because getMainBranch()
always resolved to the repo default branch with no concept of where
the user started.

Fix: record the current branch as the "integration branch" in a
per-milestone metadata file (.gsd/milestones/<MID>/<MID>-META.json)
when auto-mode starts. getMainBranch() checks this metadata before
falling back to repo defaults, so switchToMain() and mergeSliceToMain()
target the correct branch.

Key details:
- Integration branch is captured once per milestone (idempotent)
- Committed immediately so it survives branch switches (.gsd/ files
  are discarded during checkout)
- main_branch preference still takes highest priority
- Falls back to existing detection if metadata missing (backward compat)
- Per-milestone: different milestones can target different branches
- Validates branch still exists before using it

Tests: 41 new assertions across git-service.test.ts and worktree.test.ts
covering the full lifecycle, multi-slice workflows, resume scenarios,
backward compatibility, and edge cases.
2026-03-13 10:34:28 -06:00
Lex Christopherson
74e9f366bb 2.8.3 2026-03-13 10:30:05 -06:00
Lex Christopherson
4079a92146 docs: update changelog for v2.8.3
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:30:01 -06:00
TÂCHES
3084366d87 fix: handle undefined result from custom() in RPC mode for ask_user_questions (#156, #165, #171) (#199)
In RPC mode, `ctx.ui.custom()` returns `undefined as never`, causing
`showInterviewRound` to return undefined and `Object.keys(result.answers)`
to throw TypeError.

When `showInterviewRound` returns undefined (RPC mode), fall back to
sequential `ctx.ui.select()` calls for each question, forwarding the
abort signal (#171) and supporting `allowMultiple` (#165).

- Add `allowMultiple` to `ExtensionUIDialogOptions`
- Widen `select()` return type to `string | string[] | undefined`
- Add `allowMultiple` to RPC select request and `values` array to response
- Update RPC `select()` to forward `allowMultiple` and parse array responses
- Guard existing `ctx.ui.select()` callers against the widened return type

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:24:57 -06:00
TÂCHES
af27c5dd3c fix: provider-aware model resolution for per-phase preferences (#149) (#198)
Auto mode's model resolution used `allModels.find(m => m.id === modelId)`
which returns the first match regardless of provider. With 30+ duplicate
model IDs across providers, user preferences silently resolved to the
wrong provider.

Three fixes:
- Use `getAvailable()` instead of `getAll()` so only authenticated
  models are considered
- Support `provider/model` format (e.g. "google/gemini-2.5-pro") for
  explicit provider targeting
- For bare IDs, prefer the current session's provider, then first
  available match, with an ambiguity warning
- Store and restore original model provider instead of hardcoding
  "anthropic" when restoring the user's model after auto-mode

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:23:31 -06:00
TÂCHES
f2c6cad4dd fix: align execute-task artifact verification with deriveState and add self-repair (#133) (#195)
verifyExpectedArtifact only checked for the task summary file, but
deriveState determines the next task by finding the first unchecked
checkbox in the slice plan. When the agent writes the summary but
doesn't mark the checkbox, the dispatch loop re-sends the same unit
and gets stuck after max retries.

Part 1: verifyExpectedArtifact now also checks that the task checkbox
is marked [x] in the slice plan for execute-task units.

Part 2: At retry time, if the summary exists but the checkbox is
unmarked, the dispatch logic self-repairs by marking the checkbox
programmatically (via skipExecuteTask) and re-derives state instead
of re-dispatching the same unit.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:23:23 -06:00
TÂCHES
e68f486d8b fix: break research phase infinite loop and sync state on stop (#126) (#193)
Bug 1: resolveMilestoneFile/resolveSliceFile already check file existence
via readdirSync, so the additional loadFile content check was redundant.
Empty research files (exists on disk but no content) caused a loop where
verifyExpectedArtifact marked research complete while dispatch re-triggered
it because loadFile returned falsy for empty files.

Bug 2: stopAuto now calls rebuildState to synchronize disk state, matching
pauseAuto's pattern and preventing stale state on next resume.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:23:15 -06:00
TÂCHES
22cbd2bec3 fix: auto-resolve merge conflicts on .gsd/ runtime files (#189) (#194)
Auto mode exits when mergeSliceToMain() hits conflicts on runtime files
like completed-units.json that were manually committed via `gsd queue`.

Two-part fix:
- Untrack RUNTIME_EXCLUSION_PATHS from the index before merge starts
- If merge conflicts are limited to runtime files, auto-resolve by
  taking ours and removing from index instead of aborting

Closes #189

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:23:07 -06:00
TÂCHES
738444aeeb fix: auto-switch model after /login and /logout to prevent API key errors (#124) (#197)
After /login, if the current model has no valid API key, auto-switch to
a model from the newly authenticated provider. After /logout, if the
current model belongs to the logged-out provider, auto-switch to a
fallback model from a different provider.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:23:00 -06:00
TÂCHES
c344b0af54 fix: use provider field instead of model name prefix for Anthropic detection (#142) (#196)
The before_provider_request hook used model.startsWith("claude") to gate
native web search injection. This matched claude-* models served by any
provider (GitHub Copilot, AWS Bedrock, etc.), incorrectly injecting
Anthropic-only web_search_20250305 tool definitions into non-Anthropic
API requests.

The fix checks the isAnthropicProvider flag (set by model_select via the
provider field) instead of sniffing the model name.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:22:52 -06:00
Lex Christopherson
c87f15868a 2.8.2 2026-03-13 10:09:25 -06:00
Lex Christopherson
7ea7b025f1 docs: update changelog for v2.8.1 and v2.8.2
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:09:21 -06:00
Lex Christopherson
69d8baf17b chore: untrack .claude/ and .gsd/ directories, gitignore *.tgz
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:08:17 -06:00
Lex Christopherson
d9a9a73ab2 fix: replace hardcoded forward-slash path ops with node:path stdlib (#184)
Three locations used lastIndexOf("/") or includes("/") for path
manipulation, which fails on Windows where paths use backslashes.

- auto.ts: writeBlockerPlaceholder directory extraction → dirname()
- interactive-mode.ts: parent directory traversal → path.dirname() loop
- path-utils.ts: non-null assertion on MSYS drive letter access

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 10:02:20 -06:00
Lex Christopherson
74278ad865 fix: use relative paths in prompts to prevent Windows drive letter mangling (#184)
On Windows, LLMs convert absolute paths like F:\Projects\.gsd\... to
Unix-style /f/Projects/.gsd/... which Node's path.resolve interprets
as drive-root-relative, creating F:\f\Projects\.gsd\... instead.

Replace all *AbsPath template variables in prompt templates with
relative .gsd/... paths that resolve correctly on all platforms.
Add MSYS path normalization in resolveToCwd as defense-in-depth.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:55:15 -06:00
TÂCHES
83af235d86 Merge pull request #188 from gsd-build/worktree-agent-a3ed52d1
fix: untrack runtime files already in git index to prevent merge conflicts
2026-03-13 09:54:30 -06:00
Lex Christopherson
2846f9fcf7 fix: untrack runtime files already in git index to prevent merge conflicts (#187)
Repos that started tracking .gsd/completed-units.json before the
gitignore rule was added continue to see squash-merge conflicts because
.gitignore only prevents new tracking. This adds a bootstrap step that
runs `git rm --cached` on all RUNTIME_EXCLUSION_PATHS, eliminating the
conflict at its source.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:45:25 -06:00
TÂCHES
dde84fa248 Merge pull request #185 from Jamie-BitFlight/docs/preferences-clarification
docs: clarify preferences semantics and best practices
2026-03-13 09:34:36 -06:00
Lex Christopherson
02e2f97054 fix: respect HTTP_PROXY/HTTPS_PROXY env vars for all outbound requests (#131)
GSD bypasses pi-coding-agent's CLI entry point (which sets undici's
EnvHttpProxyAgent as the global dispatcher), so proxy env vars were
ignored — causing 403 errors for users behind HTTPS proxies.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:33:41 -06:00
Lex Christopherson
98c2d23ce6 fix: sanitize Windows NUL redirects to /dev/null in Git Bash (#157)
LLM-generated commands with `> NUL` create undeletable files on Windows
because Git Bash treats NUL as a literal filename. Rewrite NUL redirects
to /dev/null at all three bash spawn sites.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:33:28 -06:00
Lex Christopherson
273d6b9cf9 2.8.1 2026-03-13 09:26:23 -06:00
Lex Christopherson
6a960a4897 fix: bump fsevents to ~2.3.3 for Node 25 compatibility (#186)
fsevents 2.3.2 lacks prebuilt binaries for Node 25, causing node-gyp
build failures on install. Bumping to 2.3.3 which ships compatible
prebuilds while preserving the optionalDependency hoisting that fixes
Linux installs (#182).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:25:27 -06:00
Jamie McGregor Nelson
5d33805182 docs: add code references to clarify semantics and best practices
Adds verification against actual code:

- Empty arrays deleted: preferences.ts lines 583-587
- Arrays concatenated: preferences.ts lines 504-508, 659-667
- Scalar fields use ??: preferences.ts lines 503, 510, 514
- Object fields shallow merge: preferences.ts line 509
- skill_discovery independent from skill preference fields
2026-03-13 11:25:12 -04:00
Lex Christopherson
498614c95f fix: inject observability warnings into agent prompt for enforcement (#174)
emitObservabilityWarnings only called ctx.ui.notify — the agent never
saw the warnings and ignored them entirely. Validator caught real issues
(missing observability sections, placeholder diagnostics) but had zero
enforcement.

Rename to collectObservabilityWarnings (returns issues), add
buildObservabilityRepairBlock to format issues as actionable prompt
instructions. Appended to the unit prompt so the agent reads flagged
files and fixes gaps before proceeding with the unit.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:20:19 -06:00
Lex Christopherson
788c356a25 fix: auto-detect headless environment for Playwright browser launch (#183)
Browser launch was hardcoded to headless: false, crashing on Linux
servers without a display server ($DISPLAY). Auto-detect headless
environments and also support FORCE_HEADLESS=true override.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:16:26 -06:00
Lex Christopherson
66196b4a4f fix: add configurable merge_strategy preference for slice completion (#167)
Squash merge was hardcoded, causing auto-mode to hard-stop when conflicts
arose from long-lived branches or frequently-changing .gsd/* artifacts.

Add git.merge_strategy preference ("squash" | "merge", default: squash).
"merge" uses --no-ff which preserves branch history and handles conflicts
from divergent branches more gracefully. Users hitting repeated squash
merge failures can set merge_strategy: merge in .gsd/preferences.md.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:16:26 -06:00
Lex Christopherson
71d3a69646 fix: verify UAT artifact before marking complete-slice done (#176, #175)
complete-slice verification only checked for the SUMMARY file, so when
the LLM skipped writing the UAT, the unit was marked complete and UAT
was never produced. Users saw doctor-created placeholder UATs instead
of real test scripts.

- verifyExpectedArtifact now checks both SUMMARY and UAT for complete-slice
- complete-slice prompt strengthened: step 7 requires concrete test cases,
  MUST line lists all three required artifacts with enforcement warning

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:16:26 -06:00
Lex Christopherson
b07d34b448 chore: remove leftover investigation doc and .bg-shell directory
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:16:26 -06:00
Lex Christopherson
94bd622f0c fix: update stale comment to reflect 7 runtime exclusion paths
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 09:16:26 -06:00
dan bachelder
064a15f988 fix(gsd-auto): require prior slices complete on main (#160) 2026-03-13 08:56:34 -06:00
deseltrus
87a1e51bc0 fix: smartStage fallback bypasses runtime exclusions when .gsd/ is gitignored (#168)
Three-layer fix for runtime files leaking into git commits:

1. Stage-then-unstage: replace pathspec excludes with git add -A followed
   by git reset HEAD for each exclusion. The old approach failed when .gsd/
   was in .gitignore — git exited non-zero before evaluating the excludes,
   and the catch fallback staged everything unconditionally.

2. Auto-cleanup: on first smartStage call per session, remove any runtime
   files that are already tracked in the index (from the fallback bug) via
   a dedicated commit. This is a one-time migration that self-heals repos
   where runtime files were accidentally committed.

3. Pre-checkout discard: after pre-switch auto-commits that exclude .gsd/,
   run git checkout -- .gsd/ to clear dirty runtime files that would
   otherwise block git checkout when the target branch has different
   tracked versions.

Also adds completed-units.json to RUNTIME_EXCLUSION_PATHS and
BASELINE_PATTERNS (was missing — metrics.json was listed but
completed-units.json was not).
2026-03-13 08:53:58 -06:00
deseltrus
dcd993064c feat(gsd): add discussion depth verification and context write-gate (#181)
- Prompt enhancements: sparring patterns, enrichment reflection, depth
  verification checkpoint, depth-preservation guidance for context.md
- Write-gate: block milestone CONTEXT.md writes until depth verification
  is confirmed via ask_user_questions with depth_verification id
- Discussion persistence: auto-save exchanges to DISCUSSION.md during
  discuss phase with structured markdown formatting
- Depth state management: track verification status, reset on discuss
  phase completion, expose getDiscussionMilestoneId() for gate checks
- 7 unit tests covering write-gate logic (pure function, no mocks needed)
2026-03-13 08:53:00 -06:00
TÂCHES
9009e5dd78 Merge pull request #158 from deseltrus/feat/guided-flow-escape-hatch
feat: add skip/discard escape hatches to no-roadmap wizard
2026-03-13 08:50:04 -06:00
Adam Dry
c18f7d1f69 fix: add fsevents as optional dep to fix install on Linux (#182)
fsevents is a macOS-only native module that comes as a transitive
optional dependency of playwright. On Linux, npm tries to build it
via node-gyp which fails fatally, breaking 'npm install -g gsd-pi'.

Adding fsevents as an explicit optionalDependency at the root level
ensures npm gracefully skips it on non-macOS platforms. This is the
standard pattern used by vite, webpack, and chokidar.
2026-03-13 08:48:14 -06:00
Ryan Harrington
9f58583888 fix/gsd-graceful-exit: make /exit use graceful shutdown (#134)
* fix/gsd-graceful-exit: make /exit use graceful shutdown

* fix/gsd-graceful-exit: restore auto cleanup in exit command
2026-03-13 08:47:55 -06:00
TÂCHES
789a6645da feat: TTSR + blob/artifact storage (ported from oh-my-pi)
* docs(M002): context, requirements, and roadmap

* feat: port TTSR and blob/artifact storage from oh-my-pi

Phase 1 — TTSR (Time Traveling Stream Rules):
- TtsrManager: regex-based stream monitoring with scope filtering,
  repeat gating, and buffer isolation (picomatch replaces Bun.Glob)
- Rule loader: scans ~/.gsd/agent/rules/*.md and .gsd/rules/*.md
  with YAML frontmatter parsing; project rules override global
- TTSR extension: wires into pi event lifecycle (session_start,
  turn_start, message_update, turn_end, agent_end) to abort on
  match and inject violation as system reminder via sendMessage
- Interrupt template for rule violation injection

Phase 2 — Blob/Artifact Storage:
- BlobStore: content-addressed storage at ~/.gsd/agent/blobs/ using
  Node crypto (sha256), sync I/O, automatic deduplication
- ArtifactManager: session-scoped sequential artifact files stored
  alongside session JSONL (lazy dir creation, resume-safe ID scan)
- Session manager integration: prepareForPersistence externalizes
  images ≥1KB to blob store before JSONL write; resolveBlobRefs
  rehydrates on session load; truncates strings >500KB
- Bash tool artifact spill: uses ArtifactManager instead of temp
  files when available, includes artifact:// references in output

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* fix: harden blob store, TTSR manager, and dep classification

- Validate SHA-256 hex format in BlobStore.get/has/parseBlobRef to
  prevent path traversal via crafted blob references
- Cap TTSR per-stream buffers at 512KB to prevent unbounded memory growth
- Move picomatch from devDependencies to dependencies (runtime import)
- Warn on invalid regex in TTSR rule conditions instead of silent skip
- Remove .gsd/ planning files that were force-added past .gitignore
- Add trailing newline to ttsr-interrupt.md

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* test: add tests for blob store, artifact manager, TTSR manager, and rule loader

55 tests covering:
- BlobStore put/get/has, idempotency, path traversal rejection
- parseBlobRef/isBlobRef validation, externalize/resolve round-trips
- ArtifactManager sequential IDs, lazy dir creation, session resume
- TtsrManager rule matching, scope filtering, buffer isolation,
  repeat gating, buffer size cap, injection persistence
- Rule loader frontmatter parsing, directory scanning, merge logic

Also fixes BlobStore constructor to avoid TS parameter property syntax
(incompatible with Node's strip-only TypeScript mode).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 08:43:56 -06:00
Lex Christopherson
c88b54711f 2.8.0
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 08:05:20 -06:00
Lex Christopherson
171e711498 docs: update changelog and README for v2.8.0
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-13 08:05:14 -06:00