The post-merge worktree-cleanup loop in quick.md issued bare `git diff`,
`git merge`, and related commands relying on CWD = project root. An LLM
orchestrator that reformats bash across separate tool calls can leak CWD
into a worktree, causing the merge to silently no-op.
At the top of each iteration body, resolve PROJECT_ROOT via
`git -C "$WT" rev-parse --git-common-dir` and `cd "$PROJECT_ROOT"`.
If the root cannot be resolved or reached, log a skip message and
continue to the next manifest entry. All existing guards (pre-merge
deletion guard #1756, STATE.md/ROADMAP.md backup/restore, resurrection
guard #2501/#3195) remain intact after the CWD pin.
Closes#3521 (bug 1 — CWD safety only; bug 2 / resurrection guard was
already fixed in a6beac40 / PR #3201 and is pending reporter retest).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Simulates orchestrator-leaked CWD in the post-merge cleanup loop and
asserts that PROJECT_ROOT is resolved via `git -C "$WT" rev-parse
--git-common-dir` before any bare git command, that a missing root
causes a logged skip/continue, and that the existing pre-merge deletion
guard (#1756) and STATE.md/ROADMAP.md backup/restore remain in-place
after the CWD pin.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The grep -v alternation in the git-enhanced two-way merge step was missing
the 'gsd-update' arm after the slash-command rename from /gsd:update to
/gsd-update. Commits authored by the current update flow fell through the
filter and were misclassified as user customizations, causing spurious merge
prompts during /gsd-update --reapply.
Adds 'gsd-update' between 'gsd:update' and 'GSD update'. The legacy
'gsd:update' arm is preserved for back-compat; 'GSD update' and 'gsd-install'
exclusions are unchanged.
Adds bug-3516-reapply-patches-gsd-update-filter.test.cjs — 7 tests that
assert all four exclusion patterns (gsd:update, gsd-update, GSD update,
gsd-install) are present in the git-enhanced two-way merge filter inside
get-shit-done/workflows/reapply-patches.md.
Two tests fail before the fix: 'filter excludes renamed gsd-update commits'
and 'all four expected exclusion patterns are present in the filter'.
Fixes two root causes behind bug #3517:
1. Idempotency: completed_phases was blindly incremented (parseInt + 1),
causing phase.complete N run twice to double-count (4 → 5 → 6).
Now derives from ROADMAP progress table Complete-row count, making
the operation idempotent.
2. Field coverage: eight STATE.md fields were left stale after phase
completion. Now updates in the same atomic lock section:
- frontmatter: stopped_at, last_updated, total_plans, completed_plans
- body: Current focus, Status line, By Phase table row
completed_plans = count of *-SUMMARY.md files across all phase dirs
total_plans = sum of M/N plan counts from ROADMAP progress table
percent = recomputed from fresh derived counts
Closes#3517
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extract composeStatusline() helper from duplicated inline template logic in
runStatusline() and renderStatusline(). Both call sites now route through the
helper, which accepts a position param ('end' | 'front', default 'end').
- 'end' (default) preserves byte-identical output to v1.38.x and earlier
- 'front' renders ctx immediately after model name, before the first │
- Invalid values silently coerce to 'end' at runtime (belt-and-suspenders;
config-set rejects invalid values upfront via enum validator)
Adds statusline.context_position to VALID_CONFIG_KEYS in both CJS and TS
schemas, enum validator in config.cjs, docs row in CONFIGURATION.md,
and a changeset. Closes#2937.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
On machines where os.tmpdir() returns a path with a space (e.g.
/Volumes/Mini Me/tmp), runGsdTools() string args were whitespace-split by
the helper tokeniser, truncating paths at the first space. Switch all
calls that embed a dynamic path into the argument list to the array form
of runGsdTools() so execFileSync receives each path as a single argv slot.
Fixes#3509
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds boolean config key `git.create_tag` (default: true, fully backcompat)
so projects with their own release flow can disable GSD's automatic
`git tag -a v[X.Y]` on milestone completion. Also adds tag-collision
pre-check to prevent silent failure on re-run. Closes#3086
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds a second job to .github/workflows/stale.yml scoped via only-issue-labels
to awaiting-retest and needs-reproduction. days-before-issue-stale=5,
days-before-issue-close=0 — when the inactivity threshold trips the stale
+ close messages fire in the same run. PR side of the action is disabled
(-1 sentinel) so this job touches issues only.
The existing broad 28+14 job adds awaiting-retest and needs-reproduction to
its exempt-issue-labels list so the two jobs do not fight over the same
issues.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
migrateCodexHooksMapFormat re-emitted the raw `[hooks.<X>]` path segment as
the leaf TOML key of the new `[[hooks.<EVENT>]]` block. When the legacy
table key was a `<file>:<event>:<line>:<col>` location identifier and the
real event lived in an `event = "..."` body field, the migration emitted a
header like `[[hooks."C:\\Users\\helen\\.codex\\config.toml:session_start:0:0"]]`
that Codex 0.124.0+ refuses to load — causing `npx get-shit-done-cc@latest`
to abort the Codex runtime install on Windows configs that pre-date AoT.
Mirror the flat-AoT branch in the map-format and stale-namespaced-AoT
branches: when the section body declares `event = "..."`, that name wins
as the leaf key and `event` is excluded from the re-emitted handler body.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The check at tests/path-replacement.test.cjs:163 used a naive
content.includes(normalizedHomedir) to detect resolved homedir leaks in
installed .md files. When os.homedir() is short (e.g. /root inside a
Docker container), the substring false-matches inside ordinary tokens
such as `</root_cause_analysis>` in agents/gsd-debug-session-manager.md,
producing spurious failures with no actual path leak.
Real path leaks are always followed by a path separator, so require
`normalizedHomedir + '/'` instead. Extracted the predicate into a
testable `containsResolvedHomedir` helper and added regression tests
covering the /root case, a genuine /home/alice leak, /root followed by
an actual separator, and the $HOME placeholder short-circuit.
Fixes#3503
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The `has_git` boolean returned by `init new-project` and `init ingest-docs`
was derived from a shallow `pathExists(cwd, '.git')` check, so a subdirectory
of an existing repo reported `has_git: false`. The workflow then ran
`git init`, creating a nested `.git` inside the outer worktree and silently
diverting subsequent `gsd-sdk commit` calls into the nested repo.
Replace the shallow check with `git rev-parse --is-inside-work-tree`
semantics in both CJS (`get-shit-done/bin/lib/init.cjs`) and TS
(`sdk/src/query/init.ts`, `sdk/src/query/init-complex.ts`) handlers via a new
shared `gitWorktreeInfoInternal` helper, and expose `git_worktree_root` +
`in_nested_subdir` so the workflows can refuse `git init` inside an existing
worktree and warn that planning files will track to the outer repo.
Regression test: `tests/bug-3491-nested-git-worktree.test.cjs`.
Fixes#3491
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The DAG resolver in phase-plan-index only matched full-stem
('03-01-auth-hardening') and canonical-prefix ('03-01') forms when
resolving `depends_on` references, so plans in decimal phases
(e.g. `99.9-test`, `02.2-cross-repo`) declaring short-form
`depends_on: [01]` had their edges silently dropped. Dependents
collapsed into wave 1 and the SDK emitted a misleading
"declared wave: N but depends_on DAG places it in wave 1" warning
that pointed at the wave declaration rather than the broken reference.
Add a tertiary short-form index keyed on the trailing `-NN` of each
plan's canonical ID — derived per plan via `lastIndexOf('-')` so it
handles integer, letter-suffixed, and decimal phase IDs uniformly.
Emit a dedicated `Plan X: unresolved depends_on reference 'NN' — no
matching plan in phase` warning whenever a dep fails all three lookup
forms, so a dropped edge can no longer hide behind the wave-mismatch
warning.
Regression coverage added in `sdk/src/query/phase.test.ts` for the
decimal-phase short-form case, the integer-phase short-form case, and
the unresolved-reference warning.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When a roadmap places shared phase-detail bodies under a non-version-prefixed
`## Phase Details` heading AFTER a `### 📋 vX.Y+ (Planned)` sibling in document
order, the entire Phase Details section fell outside the slice returned by
`extractCurrentMilestone`. `gsd-sdk query phase.insert N` then reported
"Phase N not found in ROADMAP.md" even though `### Phase N:` was unambiguously
present.
PR #2455 (closing #2422) added a same-version `continue` branch that already
handles the version-prefixed variant (e.g. `## v2.0 Phase Details`). This fix
extends the same intent to the generic-label variant: after the initial
boundary scan, look for a literal `^#{1,3}\s+Phase\s+Details\b` heading past
`sectionEnd` and, if found, append the Phase Details block (up to the next
real milestone boundary — version-bearing or milestone-emoji-bearing heading —
or EOF) to the returned slice. The intervening planned-milestone content is
skipped so it does not leak into the active-milestone view.
Bounded to a single append so a malformed roadmap can't loop. Only matches the
literal `Phase Details` label (canonical per GSD ROADMAP template); anything
else continues to terminate the slice. Does not regress the v2.0/v2.1+
version-prefixed handling shipped by #2455 (existing `bug-2422` test still
passes).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Re-invoking `state complete-phase --phase <N>` on a phase that was
already marked complete in STATE.md silently rolled STATE.md back to
that phase's moment-of-completion — clobbering Status, Last Activity,
Last Activity Description, and the ## Current Position body. The bug
fired whenever a follow-up phase had been inserted (or the next phase
had begun) and a downstream workflow re-ran complete-phase on the
already-closed phase. Damage was silent: handler reported
{"updated":["Status","Last Activity","Current Position"]} and no error.
Root cause: cmdStateCompletePhase wrote unconditionally — it never
consulted STATE.md to detect that the requested phase had been
superseded. The handler is a legacy-bridge fallback (no native SDK
registration), so the SDK CLI fell through to gsd-tools.cjs.
Fix: add an idempotency guard at the top of cmdStateCompletePhase.
If STATE.md's canonical Current Phase field already names a phase
distinct from the one we are being asked to mark complete, return a
no-op payload ({updated:[], phase:"<N>", idempotent:true, note:"phase
already superseded; no-op"}) without writing to STATE.md.
The guard is conservative — it only fires when Current Phase is set
and differs from the resolved target. First-time completion (Current
Phase == target, or Current Phase absent) is unaffected, so the four
existing complete-phase test cases (#2761, #3063) continue to pass.
Regression test: tests/bug-3489-complete-phase-idempotent.test.cjs
- re-running complete-phase --phase 02.2 with Current Phase=02.2.1
in STATE.md leaves the file byte-identical and reports idempotent:true
- normal first-time completion is NOT flagged idempotent
Scope: handler-level idempotency only. Does not address the related
stopped_at filename-sort ordering issue called out in the bug report
(filed under suggested fix#2) or porting to the native registry.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(execute-phase): classify quota/rate-limit failures across runtimes (#3095)
Dispatched executor subagents that die from provider quota or rate-limit
errors currently look identical to a crashed agent to the orchestrator —
so step 7's recovery prompt offers "retry now" when the right action is
"wait for reset and resume". This adds a runtime-agnostic classifier and
wires execute-phase step 7 to it.
- `agent.classify-failure` SDK query returns
`{class: 'quota-exceeded' | 'classify-handoff-bug' | 'unknown-failure',
sentinel?, retryAfterSeconds?}`. Sentinels cover Claude Code
(`usage limit`, `429`), Copilot CLI (`rate_limit`,
`user_weekly_rate_limited`), Codex (`usage_limit_reached`,
`too many requests`), and Gemini (`RESOURCE_EXHAUSTED`,
`exceeded your`).
- `execute-phase.md` step 7 now branches on the class. Quota-exceeded
presents a wait-for-reset prompt and points at the safe-resume gate
landing in #3212 instead of re-dispatching a fresh executor.
- `docs/research/provider-rate-limit-signals.md` records the proactive
(header / SDK event) signals each provider exposes and the upstream
Claude Code / Copilot / Codex issues blocking hook-side detection —
the forward path once host runtimes surface them.
Resume-from-partial-worktree and context-load metrics from the original
report are deliberately out of scope; they overlap #3212's
`state.verify-against-disk` work already in flight.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(execute): render quota retry hint and refresh alias artifacts
* fix(workflow): restore slash namespace and execute-phase size budget
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
