* feat(commands): add safe git revert command /gsd-undo
Adds a new /gsd-undo command for safely reverting GSD phase or plan
commits. Uses phase manifest lookup with git log fallback, atomic
single-commit reverts via git revert --no-commit, dependency checking
with user confirmation, and structured revert commit messages including
a user-provided reason.
Three modes: --last N (interactive selection), --phase NN (full phase
revert), --plan NN-MM (single plan revert).
Closes#1730
* fix(commands): address review feedback for /gsd-undo
- Add dirty-tree guard before revert operations (security)
- Fix manifest schema to use manifest.phases[N].commits (critical)
- Extend dependency check to MODE=plan for intra-phase deps
- Handle mid-sequence conflict cleanup with reset HEAD + restore
- Fix unbalanced grep alternation pattern for phase scope matching
- Remove Write from allowed-tools (never needed)
* feat(workflows): add stall detection to plan-phase revision loop
Adds issue count tracking and stall detection to the plan-phase
revision loop (step 12). When issue count stops decreasing across
iterations, the loop escalates to the user instead of burning
remaining iterations. The existing 3-iteration cap remains as a
backstop. Uses normalized issue counting from checker YAML output.
Closes#1716
* fix(workflows): add parsing fallback and re-entry guard to stall detection
phases clear now checks for phase dirs before deleting. If any exist and
--confirm is absent, the command exits non-zero with a message showing the
count and how to proceed. Empty phases dir (nothing to delete) succeeds
without --confirm unchanged.
Updates new-milestone.md workflow to pass --confirm (intentional programmatic
caller). Updates existing new-milestone-clear-phases tests to match new API.
Closes#1826
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
Add three hard-stop checks to /gsd-next that prevent blind advancement:
1. Unresolved .continue-here.md checkpoint from a previous session
2. Error/failed state in STATE.md
3. Unresolved FAIL items in VERIFICATION.md
Also add a consecutive-call budget guard that prompts after 6
consecutive /gsd-next calls, preventing runaway automation loops.
All gates are bypassed with --force (prints a one-line warning).
Gates run in order and exit on the first hit to give clear,
actionable diagnostics.
Closes#1732
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Lightweight alternative to /gsd-map-codebase that spawns a single
mapper agent for one focus area instead of four parallel agents.
Supports --focus flag with 5 options: tech, arch, quality, concerns,
and tech+arch (default). Checks for existing documents and prompts
before overwriting.
Closes#1733
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Integrate lightweight thinking partner analysis at two workflow decision
points, controlled by features.thinking_partner config (default: false):
1. discuss-phase: when developer answers reveal competing priorities
(detected via keyword/structural signals), offers brief tradeoff
analysis before locking decisions
2. plan-phase: when plan-checker flags architectural tradeoffs, analyzes
options and recommends an approach aligned with phase goals before
entering the revision loop
The thinking partner is opt-in, skippable (No, I have decided),
and brief (3-5 bullets). A third integration point for /gsd-explore
will be added when #1729 lands.
Closes#1726
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Add a fourth model profile preset that assigns models by agent role:
opus for planning and debugging (reasoning-critical), sonnet for
execution and research (follows instructions), haiku for mapping and
checking (high volume, structured output).
This gives solo developers on paid API tiers a cost-effective middle
ground — quality where it matters most (planning) without overspending
on mechanical tasks (mapping, checking).
Per-agent overrides via model_overrides continue to take precedence
over any profile preset, including adaptive.
Closes#1713
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Three locations in execute-phase.md and quick.md used raw `git add
.planning/` commands that bypassed the commit_docs config check. When
users set commit_docs: false during project setup, these raw git
commands still staged and committed .planning/ files.
Add commit_docs guards (via gsd-tools.cjs config-get) around all raw
git add .planning/ invocations. The gsd-tools.cjs commit wrapper
already respects this flag — these were the only paths that bypassed it.
Fixes#1783
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
When a worktree branch outlives a milestone transition, git merge
silently overwrites STATE.md and ROADMAP.md with stale content and
resurrects archived phase directories. Fix by backing up orchestrator
files before merge, restoring after, and detecting resurrected files.
Fixes#1761
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* test: add stale /gsd: colon reference regression guard
Fixes#1748
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix: replace 39 stale /gsd: colon references with /gsd- hyphen format
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Replace AI self-identification with env var checks (ANTIGRAVITY_AGENT,
CLAUDE_CODE_ENTRYPOINT) to correctly determine which review CLI to skip.
Fixes incorrect skip behavior when running non-Claude models inside
the Antigravity client.
* feat(state): add programmatic gates for STATE.md consistency
Adds four enforcement gates to prevent STATE.md drift:
- `state validate`: detects drift between STATE.md and filesystem
- `state sync`: reconstructs STATE.md from actual project state
- `state planned-phase`: records state after plan-phase completes
- Performance Metrics update in `phase complete`
Also fixes ghost `state update-position` command reference in
execute-phase.md (command didn't exist in CLI dispatcher).
Closes#1627
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix(state): By Phase table regex ate next section when table body was empty
The lazy [\s\S]*? with a $ lookahead in byPhaseTablePattern would
match past blank lines and capture the next ## section header as table
body when no data rows existed. Replaced with a precise row-matching
pattern ((?:[ \t]*\|[^\n]*\n)*) that only captures pipe-delimited
lines. Added regression assertion to verify row placement.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Allows users to run autonomous mode up to a specific phase number.
After the target phase completes, execution halts instead of advancing.
Closes#1644
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* feat(i18n): add response_language config for cross-phase language consistency
Adds `response_language` config key that propagates through all init
outputs via withProjectRoot(). Workflows read this field and instruct
agents to present user-facing questions in the configured language,
solving the problem of language preference resetting at phase boundaries.
Usage: gsd-tools config-set response_language "Portuguese"
Closes#1399
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* test(security): allowlist discuss-phase.md for size threshold
discuss-phase.md legitimately exceeds 50K chars due to power mode
and i18n directives — not prompt stuffing.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: replace /gsd: command format with /gsd- skill format in all suggestions
All next-step suggestions shown to users were still using the old colon
format (/gsd:xxx) which cannot be copy-pasted as skills. Migrated all
occurrences across agents/, commands/, get-shit-done/, docs/, README files,
bin/install.js (hardcoded defaults for claude runtime), and
get-shit-done/bin/lib/*.cjs (generate-claude-md templates and error messages).
Updated tests to assert new hyphen format instead of old colon format.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix: migrate remaining /gsd: format to /gsd- in hooks, workflows, and sdk
Addresses remaining user-facing occurrences missed in the initial migration:
- hooks/: fix 4 user-facing messages (pause-work, update, fast, quick)
and 2 comments in gsd-workflow-guard.js
- get-shit-done/workflows/: fix 21 Skill() literal calls that Claude
executes directly (installer does not transform workflow content)
- sdk/prompt-sanitizer.ts: update regex to strip /gsd- format in addition
to legacy /gsd: format; update JSDoc comment
- tests/: update autonomous-ui-steps, prompt-sanitizer to assert new format
Note: commands/gsd/*.md frontmatter (name: gsd:xxx) intentionally unchanged
— installer derives skillName from directory path, not the name field.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix(plan-phase): preserve --chain flag in auto-advance sync and handle ui-phase gate in chain mode
Bug 1: step 15 sync-flag check only guarded against --auto, causing
_auto_chain_active to be cleared when plan-phase is invoked without
--auto in ARGUMENTS even though a --chain pipeline was active. Added
--chain to the guard condition, matching discuss-phase behaviour.
Bug 2: UI Design Contract gate (step 5.6) always exited the workflow
when UI-SPEC was missing, breaking the discuss --chain pipeline
silently. When _auto_chain_active is true, the gate now auto-invokes
gsd-ui-phase --auto via Skill() and continues to step 6 without
prompting. Manual invocations retain the existing AskUserQuestion flow.
* fix: remove <sub>/clear</sub> pattern and duplicate old-format command in discuss-phase.md
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
The gsd-verifier was reporting false-positive gaps for items explicitly
scheduled in later phases of the milestone (e.g., reporting a Phase 5
item as a gap during Phase 1 verification). This adds Step 9b to
cross-reference gaps against later phases using `roadmap analyze` and
move matched items to a `deferred` list that does not affect status.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Address review feedback: LLMs cannot enforce timing between tool calls.
Replace "stagger by ~2 seconds" with concrete, enforceable pattern:
dispatch each Task() one at a time with run_in_background: true. The
round-trip latency of each tool call provides natural spacing for
worktree creation, while agents still run in parallel once created.
Explicitly warn against sending multiple Task() calls in a single
message (which causes simultaneous git worktree add).
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The Next Up blocks showed the main command first, then /clear as a <sub>
footnote saying 'first'. Users would copy-paste the command before noticing
they should have cleared first. This restructures all 41 instances across
19 workflow files and 2 reference files so /clear appears before the
command as a clear sequential instruction.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The --chain flag was not checked in plan-phase's auto-advance guard,
causing the ephemeral _auto_chain_active config to be cleared when
discuss-phase chained into plan-phase. This broke the discuss→plan→
execute auto-advance pipeline, requiring manual intervention at each
transition.
Two fixes applied to plan-phase.md step 15:
- Sync-flag guard now checks for both --auto AND --chain before
clearing _auto_chain_active (matching discuss-phase's pattern)
- Added chain flag persistence (config-set _auto_chain_active true)
before auto-advancing, handling direct invocation without prior
discuss-phase
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add severity column (blocking/advisory) to Critical Anti-Patterns table
in .continue-here.md template in pause-work.md
- Document that blocking anti-patterns trigger a mandatory understanding
check when discuss-phase or execute-phase resumes work
- Add check_blocking_antipatterns step to discuss-phase.md: parses
.continue-here.md for blocking rows and requires three-question
understanding demonstration before proceeding
- Add identical enforcement step to execute-phase.md
- Tests: tests/anti-pattern-enforcement.test.cjs (12 assertions, all pass)
Closes#1491
* test(#1488): add failing tests for methodology artifact type
- Test artifact-types.md exists with methodology type documented
- Test shape, lifecycle, location fields are present
- Test discuss-phase-assumptions.md consumes METHODOLOGY.md
- Test pause-work.md Required Reading includes METHODOLOGY.md
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* feat(#1488): add methodology artifact type with consumption mechanisms
- Create get-shit-done/references/artifact-types.md documenting all GSD
artifact types including the new methodology type
- Methodology artifact: standing reference of named interpretive lenses,
located at .planning/METHODOLOGY.md, lifecycle Created → Active → Superseded
- Add load_methodology step to discuss-phase-assumptions.md so active lenses
are read before assumption analysis and applied to surfaced findings
- Add METHODOLOGY.md to pause-work.md Required Reading template so resuming
agents inherit the project's analytical orientation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
When Playwright-MCP is available in the session, GSD UI verification
steps can be automated via screenshot comparison instead of manual
checkbox review. Falls back to manual flow when Playwright is not
configured.
Closes#1420
Analyzes ROADMAP.md phases for file overlap and semantic dependencies,
then suggests Depends on entries before running /gsd:manager. Complements
the files_modified overlap detection added in the executor (PR #1600).
Closes#1530
* fix: correct STATE.md progress counter fields during plan/phase completion (#1589)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* ci: re-run CI with Windows pointer lifecycle fix in main
* fix: orchestrator owns STATE.md/ROADMAP.md writes in parallel worktree mode (#1571)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: detect files_modified overlap and enforce wave ordering for dependent plans (#1587)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: trim gsd-planner.md below 50000-char prompt-injection limit
The assign_waves section added in this branch pushed agents/gsd-planner.md
to 50271 chars, triggering the security scanner's prompt-stuffing check on
all CI platforms. Condense prose while preserving all logic and validation
rules; file is now 49754 chars.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: clear phases directory when creating new milestone (#1588)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* ci: re-run CI with Windows pointer lifecycle fix in main
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
When EnterWorktree creates a branch from main instead of the current HEAD
(a known issue on Windows), executor agents now detect the mismatch and
reset their branch base to the correct commit before starting work.
- execute-phase: capture EXPECTED_BASE before spawning, inject
<worktree_branch_check> block into executor prompts
- execute-plan: document Pattern A worktree_branch_check requirement
- quick.md: inject worktree_branch_check into executor prompt
- diagnose-issues: inject worktree_branch_check into debugger prompts
- settings: add workflow.use_worktrees option so Windows users can
disable worktree isolation via /gsd:settings without editing files
Closes#1510
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Detect runtime from execution_context path or env vars at workflow start,
then set INSTRUCTION_FILE (AGENTS.md for Codex, CLAUDE.md for all others).
Pass --output $INSTRUCTION_FILE to generate-claude-md so the helper writes
to the correct file instead of always defaulting to CLAUDE.md.
Also add .codex to skipDirs in init.cjs so Codex runtime directories are
not mistaken for project content during brownfield codebase analysis.
Closes#1521
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Addresses review feedback — checks if opencode output file is
non-empty after invocation, writes a failure message if empty
to prevent blank sections in REVIEWS.md.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Remove --copilot alias to avoid collision with existing Copilot concept
- Remove hardcoded model (-m) and variant flags; let user's OpenCode
config determine the model, consistent with other reviewer CLIs
- Use generic "OpenCode Review" section header since model varies by config
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds `manager.flags.{discuss,plan,execute}` config keys so users can
configure flags that /gsd:manager passes to each step when dispatching.
For example, `manager.flags.discuss: "--auto --analyze"` makes every
discuss dispatched from the manager include those flags.
Closes#1400
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds a new `audit_test_quality` step between `scan_antipatterns` and
`identify_human_verification` that catches test-level deceptions:
- Disabled tests (it.skip/xit/test.todo) covering phase requirements
- Circular tests (system generating its own expected values)
- Weak assertions (existence-only when value-level proof needed)
- Expected value provenance tracking for parity/migration phases
Any blocker from this audit forces `gaps_found` status, preventing
phases from being marked complete with inadequate test evidence.
Fixes#1457
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Bring the latest main branch updates into feat/kilo-runtime-support while preserving KILO_CONFIG resolution, Kilo agent permission conversion, and relative .claude path rewrites.
Honor KILO_CONFIG across installer and workflow resolution, preserve Claude agent tool intent as explicit Kilo permissions, and rewrite relative .claude references during Kilo conversion.
Read context_window config (default 200000) in execute-phase and plan-phase workflows. When >= 500000 (1M-class models), subagent prompts include richer context: executor agents receive CONTEXT.md, RESEARCH.md, and prior wave SUMMARYs; verifier agents receive all PLANs, SUMMARYs, and REQUIREMENTS.md; planner receives prior phase CONTEXT.md for cross-phase decision consistency.
At 200k (default), behavior is unchanged.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Both the gsd-executor and gsd-verifier Task() invocations were missing
the required `description` parameter, causing InputValidationError when
spawning agents in parallel during /gsd:execute-phase.
Adds OpenCode CLI as a 5th reviewer option, enabling teams with GitHub
Copilot subscriptions to leverage Copilot-routed models (e.g.
gpt-5.3-codex) for cross-AI plan reviews.
Closes#1520
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The autonomous workflow previously had two extremes: full auto-answer
mode (--auto discuss, lean context but no user input) or manager mode
(interactive but bloated context from accumulating everything inline).
The --interactive flag bridges this gap:
- Discuss runs inline via gsd:discuss-phase (asks questions, waits for
user answers — preserving all design decisions)
- Plan and execute dispatch as background agents (fresh context per
phase — no accumulation in the main session)
- Pipeline parallelism: discuss Phase N+1 while Phase N builds in the
background
This keeps the main context lean (only discuss conversations accumulate)
while preserving user input on all decisions. Particularly helpful for
users hitting context limits with /gsd:manager on multi-phase milestones.
Usage:
/gsd:autonomous --interactive
/gsd:autonomous --interactive --from 3
/gsd:autonomous --interactive --only 5
Also adds --only N flag parsing to the upstream workflow (previously only
in PR #1444's branch).
Closes#1413
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The map-codebase workflow had a hardcoded 300000ms (5 minute) timeout
for parallel subagent tasks. On large codebases or with slower models
(e.g. GPT via Codex), subagents can need 10-20+ minutes, causing the
parent to kill still-working agents and fall back to sequential mode.
Changes:
- Add workflow.subagent_timeout config key (default: 300000ms)
- Register in VALID_CONFIG_KEYS (config.cjs)
- Add to loadConfig() defaults and return object (core.cjs)
- Emit in map-codebase init context (init.cjs)
- Update map-codebase.md to use config value instead of hardcoded 300000
- Document in planning-config.md reference
Users can now increase the timeout via:
/gsd:settings workflow.subagent_timeout 900000
Closes#1472
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
