Two defensive hardening changes:
1. state.cjs: Replace two identical inline extractField closures
(cmdStateSnapshot and buildStateFrontmatter) with a shared
stateExtractField(content, fieldName) helper at module scope.
The helper uses escapeRegex on fieldName before interpolation
into RegExp constructors, preventing breakage if a field name
ever contains regex metacharacters. Net removal of duplicated
logic.
2. gsd-plan-checker.md: Bound the "relevant" definition in the
exhaustive cross-check instruction. A requirement is "relevant"
only if ROADMAP.md explicitly maps it to this phase or the phase
goal directly implies it — prevents false blocker flags for
requirements belonging to other phases.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Two hardening changes to cmdMilestoneComplete:
1. Replace 27 lines of inline isDirInMilestone logic (roadmap parsing,
normalization, and matching) with a single call to the shared
getMilestonePhaseFilter(cwd) from core.cjs. The inline copy was
identical to the core version — deduplicating prevents future drift.
2. Handle empty MILESTONES.md files. Previously, an existing but empty
file would fall into the headerMatch branch and produce malformed
output. Now an empty file is treated the same as a missing one,
writing the standard "# Milestones" header before the entry.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Three files emit path.join or path.relative results directly into
JSON output without normalizing backslashes on Windows. Wrap these
with toPosixPath (already used in core.cjs and init.cjs) so agents
receive consistent forward-slash paths on all platforms.
Files changed:
- phase.cjs: wrap directory path in cmdPhasesFind
- commands.cjs: wrap todo path in cmdListTodos, relPath in cmdScaffold
- template.cjs: wrap relPath in cmdTemplateFill (both exists and
created branches)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
cmdRequirementsMarkComplete interpolates user-supplied reqId strings
directly into RegExp constructors. If a reqId contains regex
metacharacters (e.g. parentheses, brackets, dots), the patterns break
or match unintended content.
Import escapeRegex from core.cjs (already used in state.cjs and
phase.cjs) and apply it to reqId before interpolation into all four
regex patterns in the function.
Same class of fix as gsd-build/get-shit-done#741 (state.cjs).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace general-purpose workaround pattern with named subagent types:
- plan-phase: researcher and planner now spawn as gsd-phase-researcher/gsd-planner
- new-project: 4 researcher spawns now use gsd-project-researcher
- research-phase: researcher spawns now use gsd-phase-researcher
- quick: planner revision now uses gsd-planner
- diagnose-issues: debug agents now use gsd-debugger (matches template spec)
Removes 'First, read agent .md file' prompt prefix — named agent types
auto-load their .md file as system prompt, making the workaround redundant.
Preserves intentional general-purpose orchestrator spawns in discuss-phase
and plan-phase (auto-advance) where the agent runs an entire workflow.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When users select "Other" to provide freeform input, Claude re-prompts
with another AskUserQuestion instead of dropping to plain text. This
loops 2-3 times before finally accepting freeform input.
Add explicit freeform escape rule to questioning.md reference, and
update new-milestone.md and discuss-phase.md to switch to plain text
when users signal freeform intent.
Closes#778
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Phases listed in ROADMAP.md but not yet planned (no directory on disk)
were excluded from total_phases, causing premature milestone completion.
Now uses Math.max(diskDirs, roadmapCount) via filter.phaseCount.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Extract getMilestonePhaseFilter() from milestone.cjs closure into core.cjs
as a shared helper. Apply it in buildStateFrontmatter and cmdPhaseComplete
so multi-milestone projects count only current milestone's phases instead
of all directories on disk.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Prevents re-asking questions already decided in earlier phases by reading
PROJECT.md, REQUIREMENTS.md, STATE.md, and all prior CONTEXT.md files
before generating gray areas. Prior decisions annotate options and skip
already-decided areas.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
zsh's built-in echo interprets \n escape sequences in JSON strings,
converting properly-escaped \\n back into literal newlines. This breaks
jq parsing with "control characters U+0000 through U+001F must be escaped".
printf '%s\n' preserves the JSON verbatim.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The auto-advance chain (discuss → plan → execute) was spawning each
subsequent phase as a Task(subagent_type="general-purpose"), creating
3-4 levels of nested agent sessions. At that nesting depth, the
Claude Code runtime hits resource limits or stdio contention, causing
the execute-phase to freeze or attempt to shell out to `claude` as a
subprocess (which is explicitly blocked).
The fix replaces Task spawns with Skill invocations for phase
transitions:
- discuss-phase auto-advance: Skill("gsd:plan-phase") instead of
Task(general-purpose)
- plan-phase auto-advance: Skill("gsd:execute-phase") instead of
Task(general-purpose)
The Skill tool runs in the same process context as the caller,
keeping the entire auto-advance chain flat at a single nesting level.
Each phase still spawns its own worker agents (gsd-executor,
gsd-planner, etc.) as Tasks, but the orchestration chain itself
no longer creates unnecessary depth.
Closes#686
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Three files had hardcoded `.claude` paths that break OpenCode and
Gemini installations where the config directory is `.config/opencode`,
`.opencode`, or `.gemini` respectively.
Changes:
- hooks/gsd-check-update.js: add detectConfigDir() helper that checks
all runtime directories for get-shit-done/VERSION, falling back to
.claude. Used for cache dir, project VERSION, and global VERSION.
- commands/gsd/reapply-patches.md: detect runtime directory for both
global and local patch directories instead of hardcoding ~/.claude/
and ./.claude/
- workflows/update.md: detect runtime directory for local and global
VERSION/marker files, and clear cache across all runtime directories
Closes#682
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Expand dual-format parsing to cover all 8 field extraction and
replacement locations in state.cjs. The previous fix only covered
cmdStateSnapshot and buildStateFrontmatter (read path), leaving 6
write-path functions with bold-only regex that silently failed on
plain-format STATE.md files.
Functions fixed:
- stateExtractField: shared read helper (cascades to cmdStateAdvancePlan,
cmdStateRecordSession)
- stateReplaceField: shared write helper (cascades to all state mutations)
- cmdStateGet: individual field lookup
- cmdStatePatch: batch field updates
- cmdStateUpdate: single field updates
- cmdStateUpdateProgress: progress bar writes
- cmdStateSnapshot session section: Last Date, Stopped At, Resume File
Each function now tries **Field:** bold format first (preserving
existing behavior), then falls back to plain Field: format. This
eliminates the read/write asymmetry where state-snapshot could read
plain-format fields but state-update/state-patch could not modify them.
Closes#730
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The commit case in the CLI router reads only args[1] for the message,
which captures just the first word when the shell strips quotes before
passing arguments to Node.js. This silently truncates every multi-word
commit message (e.g. "docs(40): create phase plan" becomes "docs(40):").
Collect all positional args between the command name and the first
flag (--files, --amend), then join them. Works correctly whether the
shell preserves quotes (single arg) or strips them (multiple args).
Closes#733
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
extractField in state-snapshot and buildStateFrontmatter only matches
the **Field:** bold markdown format, but STATE.md may use plain
Field: format depending on how it was generated. When all fields
return null, progress routing freezes with no matching condition.
Add dual-format parsing: try **Field:** first, fall back to plain
Field: with line-start anchor. Both instances in cmdStateSnapshot
and buildStateFrontmatter are updated consistently.
Closes#730
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
getMilestoneInfo matches the first version string in ROADMAP.md, which
is typically the oldest shipped milestone. For list-format roadmaps
that use emoji markers (e.g. "🚧 **v2.1 Belgium**"), the function now
checks for the in-progress marker first before falling back to the
heading-based and bare version matching.
Closes#700
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
cmdPhaseComplete determines is_last_phase and next_phase by scanning
.planning/phases/ directories on disk. Phases defined in ROADMAP.md
but not yet planned (no directory created) are invisible to this scan,
causing premature is_last_phase:true when only the first phase has
been scaffolded.
Add a fallback that parses ROADMAP.md phase headings when the
filesystem scan finds no next phase. Uses the existing comparePhaseNum
utility for consistent ordering with letter suffixes and decimals.
Closes#709
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Without --no-index, git check-ignore only reports files as ignored if
they are untracked. Once .planning/ files enter git's index (e.g., from
an initial commit before .gitignore was set up), check-ignore returns
"not ignored" even when .gitignore explicitly lists .planning/.
This means the documented safety net — "if .planning/ is gitignored,
commit_docs is automatically false" — silently fails for any repo where
.planning/ was ever committed. The --no-index flag checks .gitignore
rules regardless of tracking state, matching user expectations.
Closes#703
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The SessionStart hook always writes to ~/.claude/cache/ via os.homedir()
regardless of install type. The update workflow previously only cleared
the install-type-specific path, leaving stale cache at the global path
for local installs.
Clear both ./.claude/cache/ and ~/.claude/cache/ unconditionally.
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
Claude Code subagents sometimes rewrite ~/. paths to relative paths,
causing MODULE_NOT_FOUND when CWD is the project directory. $HOME is a
shell variable resolved at runtime, immune to model path rewriting.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add lightweight codebase scanning before gray area identification:
- New scout_codebase step checks for existing maps or does targeted grep
- Gray areas annotated with code context (existing components, patterns)
- Discussion options informed by what already exists in the codebase
- Context7 integration for library-specific questions
- CONTEXT.md template includes code_context section
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* fix: use `.claude/skills/` instead of `.agents/skills/` in agent and workflow skill references
Claude Code resolves project skills from `.claude/skills/` (project-level)
and `~/.claude/skills/` (user-level). The `.agents/skills/` path is the
universal/IDE-agnostic convention that Claude Code does not resolve, causing
project skills to be silently ignored by all affected agents and workflows.
Fixes#758
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix: support both `.claude/skills/` and `.agents/skills/` for cross-IDE compatibility
Instead of replacing `.agents/skills/` with `.claude/skills/`, reference both
paths so GSD works with Claude Code (`.claude/skills/`) and other IDE agents
like OpenCode (`.agents/skills/`).
Addresses review feedback from begna112 on #758.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Stephen Miller <Stephen@betterbox.pw>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
DEBUGGER_MODEL and CHECKER_MODEL used uppercase bash convention but the
Task() calls referenced {debugger_model} and {integration_checker_model}
(lowercase). The mismatch caused Claude to skip substitution and fall back
to the parent session model, ignoring the configured GSD profile.
Co-authored-by: Ethan Hurst <ethan.hurst@outlook.com.au>
stateReplaceField properly escaped fieldName before building the regex,
but stateExtractField did not. Field names containing regex metacharacters
could cause incorrect matches or regex errors.
Co-authored-by: Ethan Hurst <ethan.hurst@outlook.com.au>
loadConfig() didn't include nyquist_validation in its return object, so
cmdInitPlanPhase always set nyquist_validation_enabled to undefined. The
plan-phase workflow could never detect whether Nyquist validation was
enabled or disabled via config.
Co-authored-by: Ethan Hurst <ethan.hurst@outlook.com.au>
cmdPhasePlanIndex had 3 mismatches with the canonical XML plan format
defined in templates/phase-prompt.md:
- files_modified: looked up fm['files-modified'] (hyphen) but plans use
files_modified (underscore). Now checks underscore first, hyphen fallback.
- objective: read from YAML frontmatter but plans put it in <objective>
XML tag. Now extracts first line from the tag, falls back to frontmatter.
- task_count: matched ## Task N markdown headings but plans use <task>
XML tags. Now counts XML tags first, markdown fallback.
All three fixes preserve backward compat with legacy markdown-style plans.
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
- Add toPosixPath() helper to normalize output paths to forward slashes
- Use string concatenation for relative base paths instead of path.join()
- Apply toPosixPath() to all user-facing file paths in init.cjs output
- Use array-based execFileSync in test helpers to bypass shell quoting
issues with JSON args and dollar signs on Windows cmd.exe
Fixes 7 test failures on Windows: frontmatter set/merge (3), init
path assertions (2), and state dollar-amount corruption (2).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
cmdMilestoneComplete previously appended new milestone entries at the
end of MILESTONES.md. Over successive milestones this produced
chronological order (oldest first), which is counterintuitive — users
expect the most recent milestone at the top.
This fix detects the file header (h1-h3) and inserts the new entry
immediately after it, pushing older entries down. Files without a
recognizable header get the entry prepended. New files still get a
default '# Milestones' header.
Adds 2 tests: single insertion ordering assertion and three-sequential-
completions verification (v1.2 < v1.1 < v1.0 in file order).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
cmdMilestoneComplete previously iterated ALL phase directories in
.planning/phases/, including phases from prior milestones that remain on
disk. This produced inflated stats, stale accomplishments from old
summaries, and --archive-phases moving directories that belong to
earlier milestones.
This fix parses ROADMAP.md to extract phase numbers for the current
milestone, builds a normalized Set for O(1) lookup, and filters all
phase directory operations through an isDirInMilestone() helper.
The helper handles edge cases: leading zeros (01 -> 1), letter suffixes
(3A), decimal phases (3.1), large numbers (456), and excludes
non-numeric directories (notes/, misc/).
Adds 5 tests covering scoped stats, scoped archive, prefix collision
guard, non-numeric directory exclusion, and large phase numbers.
Complements upstream PRs #756 and #783 which fix getMilestoneInfo()
milestone detection — this fix addresses milestone completion scoping.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
loadConfig() never returned model_overrides, so resolveModelInternal()
could never find per-agent overrides — they were silently ignored.
Additionally, cmdResolveModel duplicated model resolution logic but
skipped the override check entirely. Now delegates to resolveModelInternal
so both code paths behave consistently.
Phase number parsing only matched single-decimal (e.g. 03.2) but crashed
on multi-level decimals (e.g. 03.2.1). Requirement IDs with regex
metacharacters (parentheses, dots) were interpolated raw into RegExp
constructors, causing SyntaxError crashes.
- Add escapeRegex() utility for safe regex interpolation
- Update normalizePhaseName/comparePhaseNum for multi-level decimals
- Replace all .replace('.', '\\.') with escapeRegex() across modules
- Escape reqId before regex interpolation in cmdPhaseComplete
- Update all phase dir matching regexes from (?:\.\d+)? to (?:\.\d+)*
- Add regression test for phase complete 03.2.1
Closes#621
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* Fix /gsd:update to pin installer to latest
* Harden /gsd:update install detection with global fallback
---------
Co-authored-by: Colin <colin@solvely.net>
Reduces token overhead of Dimension 8 and related agent additions by ~37%
with no behavioral change. Removes theory explanation, dead XML tags
(<manual>, <sampling_rate>), aspirational execution tracking, and
documentation-density prose from runtime agent bodies.
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Add requirements_completed field to summary-extract output, mapping
from SUMMARY frontmatter requirements-completed key. Enables
/gsd:audit-milestone cross-check to receive data from SUMMARY source.
Re-applied from #631 against refactored codebase (commands.cjs + split tests).
Co-authored-by: Colin Johnson <Solvely-Colin@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
When orphaned SUMMARY.md files cause totalSummaries > totalPlans, the
progress percentage exceeds 100%, making String.repeat() throw RangeError
on negative arguments. Clamp percent to Math.min(100, ...) at all three
computation sites (state, commands, roadmap).
Closes#633
Co-authored-by: vinicius-tersi <vinicius-tersi@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Records test generation in project state via state-snapshot call,
keeping add-tests consistent with other GSD workflows that track
their operations.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
New command that generates unit and E2E tests for completed phases:
- Analyzes implementation files and classifies into TDD/E2E/Skip
- Enforces RED-GREEN gates for both unit and E2E tests
- Uses phase SUMMARY.md, CONTEXT.md, and VERIFICATION.md as test specs
- Presents test plan for user approval before generating
Test classification rules:
- TDD: pure functions where expect(fn(input)).toBe(output) is writable
- E2E: UI behavior verifiable by Playwright (keyboard, navigation, forms)
- Skip: styling, config, glue code, migrations, simple CRUD
Ref: Issue #302
Add --cwd <path> / --cwd=<path> support so sandboxed subagents running
outside the project root can target a specific directory. Invalid paths
return a clear error. Tests ported to tests/state.test.cjs (the old
monolithic test file was split into domain files on main).
Closes#622
Co-Authored-By: Colin Johnson <colin@solvely.net>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
