Commit Graph

110 Commits

Author SHA1 Message Date
CI
9f8d11d603 fix: preserve top-level config keys and use absolute agent paths for Codex ≥0.116
Two fixes for Codex config.toml compatibility:

1. ensureCodexHooksFeature: insert [features] before the first table header
   instead of prepending it before all content. Prepending traps bare
   top-level keys (model, model_reasoning_effort) under [features], where
   Codex rejects them with "invalid type: string, expected a boolean".

2. generateCodexConfigBlock: use absolute config_file paths when targetDir
   is provided. Codex ≥0.116 requires AbsolutePathBuf and cannot resolve
   relative "agents/..." paths, failing with "AbsolutePathBuf deserialized
   without a base path".

Fixes #1202

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-24 10:11:31 +08:00
Tom Boucher
8579a30065 feat: add Windsurf runtime support
Adds full Windsurf (by Codeium) runtime integration, following the same
pattern as the existing Cursor support. Windsurf uses .windsurf/ for
local config and ~/.windsurf/ for global config, with skills in
.windsurf/skills/ using the SKILL.md structure.

What:
- CLI flag --windsurf and interactive prompt option (8)
- Directory mapping (.windsurf local, ~/.windsurf global)
- Content converter functions (tool names, path replacements, brand refs)
- Skill copy function (copyCommandsAsWindsurfSkills)
- Agent conversion (convertClaudeAgentToWindsurfAgent)
- Install/uninstall branches
- Banner, help text, and issue template updates
- Windsurf conversion test suite (windsurf-conversion.test.cjs)
- Updated multi-runtime selection tests for 8 runtimes

Closes #1336

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-23 13:50:23 -04:00
Tom Boucher
156c008c33 Merge pull request #1296 from gsd-build/fix/gemini-before-tool-hook-1295
fix: use BeforeTool hook event for Gemini CLI prompt guard
2026-03-21 14:10:55 -04:00
Tom Boucher
4395b2ecf9 fix: use BeforeTool hook event for Gemini CLI instead of PreToolUse
Gemini CLI uses BeforeTool (not PreToolUse) for pre-tool hooks, matching
the existing pattern where AfterTool is used instead of PostToolUse. The
prompt injection guard hook was hardcoded to PreToolUse, causing Gemini
CLI to log "Invalid hook event name: PreToolUse" on startup.

Apply the same runtime-conditional mapping used for post-tool hooks, and
update the uninstall cleanup to iterate both event names.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-21 14:08:32 -04:00
Tom Boucher
87e3b41d67 Merge pull request #1291 from gsd-build/feat/multi-runtime-select-1281
feat: multi-runtime selection in interactive installer
2026-03-21 10:05:56 -04:00
Tom Boucher
f3ce0957cd feat: multi-runtime selection in interactive installer (#1281)
The interactive prompt now accepts comma-separated or space-separated
choices (e.g., "1,4,6" or "1 4 6") to install multiple runtimes in
one go, without needing --all or running the installer multiple times.

- Replaced if/else-if chain with runtimeMap lookup + split parser
- Added hint text: "Select multiple: 1,4,6 or 1 4 6"
- Invalid choices silently filtered, duplicates deduplicated
- Empty input still defaults to Claude Code
- Choice "8" still selects all runtimes

Tests: 10 new tests covering comma/space/mixed separators,
deduplication, invalid input filtering, order preservation,
and source-level assertions.

Closes #1281

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-21 09:52:52 -04:00
Tom Boucher
f9cb02e005 fix: use $HOME instead of ~ in installed shell command paths (#1284)
The installer pathPrefix for global installs replaced os.homedir()
with ~, which does NOT expand inside double-quoted shell commands
in POSIX shells. This caused MODULE_NOT_FOUND errors when executing
commands like: node "~/.claude/get-shit-done/bin/gsd-tools.cjs"

Changed pathPrefix to use $HOME instead of ~, which correctly expands
inside double quotes. Also fixed a quoted-tilde instance in do.md.

- bin/install.js: $HOME prefix instead of ~ for global installs
- get-shit-done/workflows/do.md: node "$HOME/..." instead of "~/"
- tests/path-replacement.test.cjs: updated + 3 new regression tests

Closes #1284

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-21 09:24:31 -04:00
Tom Boucher
02254db611 fix: resolve ProviderModelNotFoundError on non-Claude runtimes (#1156)
Extend resolve_model_ids to accept "omit" value: returns empty string
so non-Claude runtimes (OpenCode, Codex, Gemini, etc.) use their
configured default model instead of unresolvable Claude aliases.

- resolve_model_ids: "omit" short-circuits before alias resolution
- model_overrides still respected (checked first) for explicit IDs
- Installer sets resolve_model_ids: "omit" in ~/.gsd/defaults.json
  for non-Claude runtimes during install
- 4 new tests covering omit behavior and override passthrough
- Fix websearch test mocks for fs.writeSync output change (#1276)

Closes #1156

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-21 00:57:20 -04:00
Tom Boucher
c4b313c60f fix: stale hook detection checks wrong directory path
gsd-check-update.js looked for hooks in configDir/hooks/ (e.g.,
~/.claude/hooks/) but the installer writes hooks to
configDir/get-shit-done/hooks/. This mismatch caused false stale
hook warnings that persisted even after updating.

Also clears the update cache during install so the next session
re-evaluates hook versions with the correct path.

Closes #1249

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-20 16:24:06 -04:00
Tom Boucher
62db008570 security: add prompt injection guards, path traversal prevention, and input validation
Defense-in-depth security hardening for a codebase where markdown files become
LLM system prompts. Adds centralized security module, PreToolUse hook for
injection detection, and CI-ready codebase scan.

New files:
- security.cjs: path traversal prevention, prompt injection scanner/sanitizer,
  safe JSON parsing, field name validation, shell arg validation
- gsd-prompt-guard.js: PreToolUse hook scans .planning/ writes for injection
- security.test.cjs: 62 unit tests for all security functions
- prompt-injection-scan.test.cjs: CI scan of all agent/workflow/command files

Hardened code paths:
- readTextArgOrFile: path traversal guard (--prd, --text-file)
- cmdStateUpdate/Patch: field name validation prevents regex injection
- cmdCommit: sanitizeForPrompt strips invisible chars from commit messages
- gsd-tools --fields: safeJsonParse wraps unprotected JSON.parse
- cmdFrontmatterGet/Set: null byte rejection
- cmdVerifyPathExists: null byte rejection
- install.js: registers prompt guard hook, updates uninstaller

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-20 11:38:26 -04:00
Tom Boucher
9bf85fb97d fix(install): add matcher and timeout to context-monitor hook (#1246)
The gsd-context-monitor PostToolUse hook was configured without a
matcher or timeout, causing it to fire on every tool use including
Read, Glob, and Grep. When multiple Read calls happen in parallel,
some hook processes failed with errors.

Added matcher: 'Bash|Edit|Write|MultiEdit|Agent|Task' to limit the
hook to tools that actually modify context significantly. Added
timeout: 10 to prevent hangs.

Includes migration logic: existing installations without matcher/timeout
get them added on next /gsd:update.
2026-03-20 10:45:56 -04:00
Disaster-Terminator
b6a49163ea fix(codex-config): preserve EOL when enabling codex hooks 2026-03-20 13:24:19 +08:00
Tom Boucher
52f6c71bdf Merge pull request #1144 from ChaptersOfFloatingLife/fix/codex-agent-toml-metadata
fix(codex): include required agent metadata in TOML
2026-03-19 12:12:08 -04:00
Tom Boucher
1f2e17923a Merge pull request #1142 from medhatgalal/AI/fix-codex-markerless-gsd-merge
Fix duplicate Codex GSD agent blocks without marker
2026-03-19 12:11:48 -04:00
Tom Boucher
62a1d5186f Merge pull request #1126 from ElliotDrel/fix/track-hooks-in-manifest
fix: track hook files in manifest for local patch detection
2026-03-19 12:11:25 -04:00
Tom Boucher
a44094018b Merge pull request #1177 from matteo-michele-bianchini/fix/windows-absolute-paths
fix: use ~/ instead of resolved absolute paths in global install
2026-03-19 12:08:34 -04:00
Tom Boucher
862f6b91ba fix: prevent Codex config.toml corruption from non-boolean [features] keys (#1202)
When GSD installs codex_hooks = true under [features], any non-boolean
keys already in that section (e.g. model = "gpt-5.4") cause Codex's
TOML parser to fail with 'invalid type: string, expected a boolean'.

Root cause: TOML sections extend until the next [section] header. If
the user placed model/model_reasoning_effort under [features] (common
since Codex's own config format encourages this), GSD's installer
didn't detect or correct the structural issue.

Fix: After injecting codex_hooks, scan the [features] section for
non-boolean values and move them above [features] to the top level.
This preserves the user's keys while keeping [features] clean for
Codex's strict boolean parser.

Includes 2 regression tests:
- Detects non-boolean keys under [features] (model, model_reasoning_effort)
- Confirms boolean keys (codex_hooks, multi_agent) are not flagged

Closes #1202
2026-03-18 23:32:05 -04:00
Tom Boucher
5fd384f336 fix: universal agent name replacement for non-Claude runtimes (#766) (#1195)
* fix: universal agent name replacement for non-Claude runtimes (#766)

Adds neutralizeAgentReferences() shared function that all non-Claude
runtime converters call to replace Claude-specific references:

- 'Claude' (standalone agent name) → 'the agent'
- 'CLAUDE.md' → runtime-specific file (AGENTS.md, GEMINI.md, COPILOT.md)
- Removes 'Do NOT load full AGENTS.md' (harmful for AGENTS.md runtimes)

Preserves: 'Claude Code' (product), 'Claude Opus/Sonnet/Haiku' (models),
'claude-' prefixes (packages, CSS classes).

Integrated into: OpenCode, Gemini, Copilot, Antigravity, and Codex
converters. Claude Code converter unchanged (references are correct).

Includes 7 new tests covering all replacement rules.

Closes #766

* fix: use copilot-instructions.md instead of COPILOT.md for Copilot runtime

Addresses review from Solvely-Colin: Copilot's actual instruction file
is copilot-instructions.md, not COPILOT.md. The neutralizer was mapping
CLAUDE.md -> COPILOT.md which would reference a non-existent file.

- install.js: pass 'copilot-instructions.md' to neutralizeAgentReferences
- runtime-converters.test.cjs: update test to validate correct filename
2026-03-18 19:20:58 -04:00
Tom Boucher
a8539a1779 Merge pull request #1164 from trek-e/fix/opencode-model-inherit-1156 2026-03-18 18:06:03 -04:00
mitchelladam
69cfae2011 fix(cursor): preserve slash-prefixed gsd commands in converted markdown
Keep `/gsd-*` command examples slash-prefixed during Cursor conversion while still normalizing legacy `gsd:` syntax, and add regression coverage for Next Up markdown output.

Made-with: Cursor
2026-03-18 18:49:28 +00:00
adam.mitchell
7172d16447 Merge branch 'main' into feat/cursor-cli-support 2026-03-18 18:34:04 +00:00
mitchelladam
026a1b013e fix(cursor): write unquoted skill and subagent names
Prevent Cursor from treating frontmatter quotes as part of skill/subagent identifiers by emitting plain name scalars, and add regression tests to lock the conversion behavior.

Made-with: Cursor
2026-03-18 18:29:47 +00:00
Tom Boucher
0f112abf55 fix: remove model:inherit from OpenCode agent conversion (#1156)
OpenCode does not recognize 'model: inherit' as a valid model identifier —
it throws ProviderModelNotFoundError when spawning any GSD subagent.

Changes:
- Remove 'model: inherit' injection for OpenCode agents in install.js
- Strip 'model:' field entirely during OpenCode frontmatter conversion
  (OpenCode uses its configured default model when no model is specified)
- Update tests to verify model: inherit is NOT added

This fixes all 15 GSD agent definitions and the gsd-set-profile command
for OpenCode users.

Fixes #1156
2026-03-18 12:23:41 -04:00
Tom Boucher
aa9cb7bcb6 feat: add Codex hooks support for SessionStart (#1020) (#1138)
Codex v0.114.0 added experimental hooks with SessionStart and Stop events.
GSD's Codex installer now configures hooks in config.toml:

- Enables codex_hooks feature flag in [features] section
- Adds SessionStart hook for GSD update checking (gsd-update-check.js)
- Graceful fallback if config.toml write fails

Hook format follows Codex's TOML convention:
  [[hooks]]
  event = "SessionStart"
  command = "node /path/to/gsd-update-check.js"

PostToolUse hooks (context monitor) are not yet added since Codex only
supports SessionStart and Stop events currently.

Fixes #1020
2026-03-18 09:58:15 -06:00
Tom Boucher
e7198f419f fix: hook version tracking, stale hook detection, stdin timeout, and session-report command (#1153, #1157, #1161, #1162) (#1163)
* fix: hook version tracking, stale hook detection, and stdin timeout increase

- Add gsd-hook-version header to all hook files for version tracking (#1153)
- Install.js now stamps current version into hooks during installation
- gsd-check-update.js detects stale hooks by comparing version headers
- gsd-statusline.js shows warning when stale hooks are detected
- Increase context monitor stdin timeout from 3s to 10s (#1162)
- Set +x permission on hook files during installation (#1162)

Fixes #1153, #1162, #1161

* feat: add /gsd:session-report command for post-session summary generation

Adds a new command that generates SESSION_REPORT.md with:
- Work performed summary (phases touched, commits, files changed)
- Key outcomes and decisions made
- Active blockers and open items
- Estimated resource usage metrics

Reports are written to .planning/reports/ with date-stamped filenames.

Closes #1157

* test: update expected skill count from 39 to 40 for new session-report command
2026-03-18 09:57:20 -06:00
Matteo Michele Bianchini
8c1b224474 fix: use ~/ instead of resolved absolute paths in global install
On Windows, install.js resolves $HOME to the absolute path
(e.g. C:/Users/matte/.claude/) in all workflow .md files.
This breaks when ~/.claude is mounted into a Docker container
where the path doesn't exist — Node interprets the Windows path
as relative to CWD, producing paths like:
/workspace/project/C:/Users/matte/.claude/get-shit-done/bin/gsd-tools.cjs

For global installs, replace os.homedir() with ~ in pathPrefix
so that paths like ~/.claude/get-shit-done/bin/gsd-tools.cjs
work correctly across all environments.

Local installs keep using resolved absolute paths since they
may be outside $HOME.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-18 09:57:33 +00:00
mitchelladam
92e5c04e00 feat: add Cursor CLI runtime support
Add Cursor as a fifth supported runtime alongside Claude Code, OpenCode,
Gemini, and Codex. Cursor uses skills (~/.cursor/skills/gsd-*/SKILL.md)
like Codex, with tool name mappings (Bash→Shell, Edit→StrReplace),
subagent type conversion, and full Claude-to-Cursor content adaptation
including project conventions (CLAUDE.md→.cursor/rules/, .claude/skills/
→.cursor/skills/) and brand references.

Includes install, uninstall, interactive prompt, help text, manifest
tracking, and .cjs/.js utility script conversion.

Made-with: Cursor
2026-03-17 21:18:08 +00:00
CI
5a7d56e6c5 fix(codex): include required agent metadata in TOML 2026-03-17 20:24:35 +08:00
Medhat Galal
10294128c9 Fix duplicate Codex GSD agent blocks without marker 2026-03-17 08:10:59 -04:00
Elliot Drel
63af9af0f4 fix: track hook files in manifest for local patch detection (#769)
Hook files (gsd-statusline.js, gsd-check-update.js, gsd-context-monitor.js)
are installed during updates but were never included in gsd-file-manifest.json.
This means saveLocalPatches() could not detect user modifications to hooks,
causing them to be silently overwritten on update with no backup.

Add hooks/gsd-*.js to writeManifest() so the existing local patch detection
system automatically backs up modified hooks to gsd-local-patches/ before
overwriting, matching the behavior already in place for workflows, commands,
and agents.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-16 23:49:34 -04:00
Berkay Karaman
a5f5d50f14 feat: add Antigravity runtime support (#1058)
* feat: add Antigravity runtime support

Add full installation support for the Antigravity AI agent, bringing
get-shit-done capabilities to the new runtime alongside Claude Code,
OpenCode, Gemini, Codex, and Copilot.

- New runtime installation capability in bin/install.js
- Commands natively copied to the unified skills directory
- New test integration suite: tests/antigravity-install.test.cjs
- Refactored copy utility to accommodate Antigravity syntax
- Documentation added into README.md

Co-authored-by: Antigravity <noreply@google.com>

* fix: add missing processAttribution call in copyCommandsAsAntigravitySkills

Antigravity SKILL.md files were written without commit attribution metadata,
inconsistent with the Copilot equivalent (copyCommandsAsCopilotSkills) which
calls processAttribution on each skill's content before writing it.

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

* fix: update Copilot install test assertions for 3 new UI agents

* docs: update CHANGELOG for Antigravity runtime support

---------

Co-authored-by: Antigravity <noreply@google.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-03-15 19:31:01 -06:00
TÂCHES
5d703954c9 fix: use absolute paths for gsd-tools.cjs in all install types (#820) (#1062)
Local installs wrote $HOME/.claude/get-shit-done/bin/gsd-tools.cjs into
workflow files, which breaks when GSD is installed outside $HOME (e.g.
external drives, symlinked projects) and when spawned subagents have an
empty $HOME environment variable.

- pathPrefix now always resolves to an absolute path via path.resolve()
- All $HOME/.claude/ replacements use the absolute prefix directly
- Codex installer uses absolute path for get-shit-done prefix
- Removed unused toHomePrefix() function

Tested: 535/535 existing tests pass, verified local install produces
correct absolute paths, verified global install unchanged, verified
empty $HOME scenario resolves correctly.

Closes #820

Made-with: Cursor

Co-authored-by: ralberts3 <ralberts3@gatech.edu>
2026-03-15 11:44:25 -06:00
flongstaff
386fc0f40c fix: handle agent frontmatter correctly in OpenCode conversion (#981)
convertClaudeToOpencodeFrontmatter() was designed for commands but is
also called for agents. For agents it incorrectly strips name: (needed
by OpenCode agents), keeps color:/skills:/tools: (should strip), and
doesn't add model: inherit / mode: subagent (required by OpenCode).

Add isAgent option to convertClaudeToOpencodeFrontmatter() so agent
installs get correct frontmatter: name preserved, Claude-only fields
stripped, model/mode injected. Command conversion unchanged (default).

Includes 14 test cases covering agent and command conversion paths.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-15 11:42:12 -06:00
TÂCHES
637a3e720c fix: respect existing opencode.jsonc config files (#1056)
The installer hardcoded `opencode.json` in all OpenCode config paths,
creating a duplicate file when users already had `opencode.jsonc`.

Add `resolveOpencodeConfigPath()` helper that prefers `.jsonc` when it
exists, and use it in all three OpenCode config touchpoints: attribution
check, permission configuration, and uninstall cleanup.

Closes #1053

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-15 11:39:53 -06:00
TÂCHES
e5b389cdb1 fix: handle EPERM/EACCES in scanForLeakedPaths on Windows (#1055)
Wrap readdirSync and readFileSync calls in try/catch to silently skip
directories and files with restricted ACLs (e.g. Chrome/Gemini certificate
stores on Windows) instead of crashing the installer.

Closes #964

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-15 11:39:47 -06:00
TÂCHES
b4781449f8 fix: remove deprecated Codex config keys causing UI instability (#1051)
* fix: remove deprecated Codex config keys causing UI instability (closes #1037)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: update codex config tests to match simplified config structure

Tests asserted the old config structure ([features] section, multi_agent,
default_mode_request_user_input, [agents] table with max_threads/max_depth)
that was deliberately removed. Tests now verify the new behavior: config
block contains only the GSD marker and per-agent [agents.gsd-*] sections.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-14 21:30:09 -06:00
TÂCHES
278414a0a7 fix: detect WSL + Windows Node.js mismatch and warn user (closes #1021) (#1049)
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-14 21:24:08 -06:00
TÂCHES
8313cd27b2 fix: add uninstall mode indicator to banner output (closes #1024) (#1046)
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-14 21:15:57 -06:00
TÂCHES
893cee85d7 fix: use tilde paths in templates to prevent PII leak in .planning/ files (closes #987) (#1047)
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-14 21:15:09 -06:00
erre
c71c15c76e feat: add Copilot CLI runtime support and gsd-autonomous skill (#911)
* gsd: Installed

* docs: complete project research

Research for adding GitHub Copilot CLI as 5th runtime to installer.

Files:
- STACK.md: Zero new deps, Copilot reads from .github/, tool name mapping
- FEATURES.md: 18 table stakes, 4 differentiators, 6 anti-features
- ARCHITECTURE.md: Codex-parallel pattern, 5 new functions, 12 existing changes
- PITFALLS.md: 10 pitfalls with prevention strategies and phase mapping
- SUMMARY.md: Synthesized findings, 4-phase roadmap suggestion

* docs(01): create phase plan for core installer plumbing

* feat(01-01): add Copilot as 5th runtime across all install.js locations

- Add --copilot flag parsing and selectedRuntimes integration
- Add 'copilot' to --all array (5 runtimes)
- getDirName('copilot') returns '.github' (local path)
- getGlobalDir('copilot') returns ~/.copilot with COPILOT_CONFIG_DIR override
- getConfigDirFromHome handles copilot for both local/global
- Banner and help text updated to include Copilot
- promptRuntime: Copilot as option 5, All renumbered to option 6
- install(): isCopilot variable, runtimeLabel, skip hooks (Codex pattern)
- install(): Copilot early return before hooks/settings configuration
- finishInstall(): Copilot program name and /gsd-new-project command
- uninstall(): Copilot runtime label and isCopilot variable
- GSD_TEST_MODE exports: getDirName, getGlobalDir, getConfigDirFromHome

* test(01-01): add Copilot plumbing unit tests

- 19 tests covering getDirName, getGlobalDir, getConfigDirFromHome
- getGlobalDir: default path, explicit dir, COPILOT_CONFIG_DIR env var, priority
- Source code integration checks for CLI-01 through CLI-06
- Verifies --both flag unchanged, hooks skipped, prompt options correct
- All 481 tests pass (19 new + 462 existing, no regressions)

* docs(01-01): complete core installer plumbing plan

- Mark Phase 1 and Plan 01-01 as complete in ROADMAP.md
- All 6 requirements (CLI-01 through CLI-06) fulfilled

* gsd: planning

* docs(02): create phase 2 content conversion engine plans

* feat(02-01): add Copilot tool mapping constant and conversion functions

- Add claudeToCopilotTools constant (13 Claude→Copilot tool mappings)
- Add convertCopilotToolName() with mcp__context7__ wildcard handling
- Add convertClaudeToCopilotContent() for CONV-06 (4 path patterns) + CONV-07 (gsd:→gsd-)
- Add convertClaudeCommandToCopilotSkill() for skill frontmatter transformation
- Add convertClaudeAgentToCopilotAgent() with tool dedup and JSON array format
- Export all new functions + constant via GSD_TEST_MODE

* feat(02-01): wire Copilot conversion into install() flow

- Add copyCommandsAsCopilotSkills() for folder-per-skill structure
- Add isCopilot branch in install() skill copy section
- Add isCopilot branch in agent loop with .agent.md rename
- Skip generic path replacement for Copilot (converter handles it)
- Add isCopilot branch in copyWithPathReplacement for .md files
- Add .cjs/.js content transformation for CONV-06/CONV-07
- Export copyCommandsAsCopilotSkills via GSD_TEST_MODE
- CONV-09 not generated (discarded), CONV-10 confirmed working

* docs(02-01): complete content conversion engine plan

- Create 02-01-SUMMARY.md with execution results
- Update STATE.md with Phase 2 position and decisions
- Mark CONV-01 through CONV-10 requirements complete

* test(02-02): add unit tests for Copilot conversion functions

- 16 tests for convertCopilotToolName (all 12 direct mappings, mcp prefix, wildcard, unknown fallback, constant size)
- 8 tests for convertClaudeToCopilotContent (4 path patterns, gsd: conversion, mixed content, no double-replace, passthrough)
- 7 tests for convertClaudeCommandToCopilotSkill (all fields, missing optional fields, CONV-06/07, no frontmatter, agent field)
- 7 tests for convertClaudeAgentToCopilotAgent (dedup, JSON array, field preservation, mcp tools, no tools, CONV-06/07, no frontmatter)

* test(02-02): add integration tests for Copilot skill copy and agent conversion

- copyCommandsAsCopilotSkills produces 31 skill folders with SKILL.md files
- Skill content verified: comma-separated allowed-tools, no YAML multiline, CONV-06/07 applied
- Old skill directories cleaned up on re-run
- gsd-executor agent: 6 tools → 4 after dedup (Write+Edit→edit, Grep+Glob→search)
- gsd-phase-researcher: mcp__context7__* wildcard → io.github.upstash/context7/*
- All 11 agents convert without error, all have frontmatter and tools
- Engine .md and .cjs files: no ~/.claude/ or gsd: references after conversion
- Full suite: 527 tests pass, zero regressions

* docs(02-02): complete Copilot conversion test suite plan

- SUMMARY: 46 new tests covering all conversion functions
- STATE: Phase 02 complete, 3/3 plans done
- ROADMAP: Phase 02 marked complete

* docs(03): research phase domain

* docs(03-instructions-lifecycle): create phase plan

* feat(03-01): add copilot-instructions template and merge/strip functions

- Create get-shit-done/templates/copilot-instructions.md with 5 GSD instructions
- Add GSD_COPILOT_INSTRUCTIONS_MARKER and GSD_COPILOT_INSTRUCTIONS_CLOSE_MARKER constants
- Add mergeCopilotInstructions() with 3-case merge (create, replace, append)
- Add stripGsdFromCopilotInstructions() with null-return for GSD-only content

* feat(03-01): wire install, fix uninstall/manifest/patches for Copilot

- Wire mergeCopilotInstructions() into install() before Copilot early return
- Add else-if isCopilot uninstall branch: remove skills/gsd-*/ + clean instructions
- Fix writeManifest() to hash Copilot skills: (isCodex || isCopilot)
- Fix reportLocalPatches() to show /gsd-reapply-patches for Copilot
- Export new functions and constants in GSD_TEST_MODE
- All 527 existing tests pass with zero regressions

* docs(03-01): complete instructions lifecycle plan

- Create 03-01-SUMMARY.md with execution results
- Update STATE.md: Phase 3 Plan 1 position, decisions, session
- Update ROADMAP.md: Phase 03 progress (1/2 plans)
- Mark INST-01, INST-02, LIFE-01, LIFE-02, LIFE-03 complete

* test(03-02): add unit tests for mergeCopilotInstructions and stripGsdFromCopilotInstructions

- 10 new tests: 5 merge cases + 5 strip cases
- Tests cover create/replace/append merge scenarios
- Tests cover null-return, content preservation, no-markers passthrough
- Added beforeEach/afterEach imports for temp dir lifecycle
- Exported writeManifest and reportLocalPatches via GSD_TEST_MODE for Task 2

* test(03-02): add integration tests for uninstall, manifest, and patches Copilot fixes

- 3 uninstall tests: gsd-* skill identification, instructions cleanup, GSD-only deletion
- writeManifest hashes Copilot skills in manifest JSON (proves isCopilot fix)
- reportLocalPatches uses /gsd-reapply-patches for Copilot (dash format)
- reportLocalPatches uses /gsd:reapply-patches for Claude (no regression)
- Full suite: 543 tests pass, 0 failures

* docs(03-02): complete instructions lifecycle tests plan

- SUMMARY.md with 16 new tests documented
- STATE.md updated: Phase 3 complete, 5/5 plans done
- ROADMAP.md updated: Phase 03 marked complete

* docs(04): capture phase context

* docs(04): research phase domain

* docs(04): create phase plan — E2E integration tests for Copilot install/uninstall

* test(04-01): add E2E Copilot full install verification tests

- 9 tests: skills count/structure, agents count/names, instructions markers
- Manifest structure, categories, SHA256 integrity verification
- Engine directory completeness (bin, references, templates, workflows, CHANGELOG, VERSION)
- Uses execFileSync in isolated /tmp dirs with GSD_TEST_MODE stripped from env

* test(04-01): add E2E Copilot uninstall verification tests

- 6 tests: engine removal, instructions removal, GSD skills/agents cleanup
- Preserves non-GSD custom skills and agents after uninstall
- Standalone lifecycle tests for preservation (install → add custom → uninstall → verify)
- Full suite: 558 tests passing, 0 failures

* docs(04-01): complete E2E Copilot install/uninstall integration tests plan

- SUMMARY.md: 15 E2E tests, SHA256 integrity, 558 total tests passing
- STATE.md: Phase 4 complete, 6/6 plans done
- ROADMAP.md: Phase 4 marked complete
- REQUIREMENTS.md: QUAL-01 complete, QUAL-02 out of scope

* fix: use .github paths for Copilot --local instead of ~/.copilot

convertClaudeToCopilotContent() was hardcoded to always map ~/.claude/
and $HOME/.claude/ to ~/.copilot/ and $HOME/.copilot/ regardless of
install mode. For --local installs these should map to .github/ (repo-
relative, no ./ prefix) since Copilot resolves @file references from
the repo root.

Local mode:  ~/.claude/ → .github/  |  $HOME/.claude/ → .github/
Global mode: ~/.claude/ → ~/.copilot/  |  $HOME/.claude/ → $HOME/.copilot/

Added isGlobal parameter to convertClaudeToCopilotContent,
convertClaudeCommandToCopilotSkill, convertClaudeAgentToCopilotAgent,
copyCommandsAsCopilotSkills, and copyWithPathReplacement. All call
sites in install() now pass isGlobal through.

Tests updated to cover both local (default) and global modes.
565 tests passing, 0 failures.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix: use double quotes for argument-hint in Copilot skills

The converter was hardcoding single quotes around argument-hint values
in skill frontmatter. This breaks YAML parsing when the value itself
contains single quotes (e.g., "e.g., 'v1.1 Notifications'").

Now uses yamlQuote() (JSON.stringify) which produces double-quoted
strings with proper escaping.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* chore: complete v1.23 milestone — Copilot CLI Support

Archive milestone artifacts, retrospective, and update project docs.

- Archive: v1.23-ROADMAP.md, v1.23-REQUIREMENTS.md, v1.23-MILESTONE-AUDIT.md
- Create: MILESTONES.md, RETROSPECTIVE.md
- Evolve: PROJECT.md (validated reqs, key decisions, shipped context)
- Reorganize: ROADMAP.md (collapsed v1.23, progress table)
- Update: STATE.md (status: completed)
- Delete: REQUIREMENTS.md (archived, fresh for next milestone)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* chore: archive phase directories from v1.23 milestone

* chore: Clean gsd tracking

* fix: update test counts for new upstream commands and agents

Upstream added validate-phase command (32 skills) and nyquist-auditor agent (12 agents).

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* chore: Remove copilot instructions

* chore: Improve loop

* gsd: installation

* docs: start milestone v1.24 Autonomous Skill

* docs: internal research for autonomous skill

* docs: define milestone v1.24 requirements

* docs: create milestone v1.24 roadmap (4 phases)

* docs: phase 5 context — skill scaffolding decisions

* docs(5): research phase domain

* docs(05): create phase plan — 2 plans in 2 waves

* docs(phase-5): add validation strategy

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* test(05-01): add failing tests for colon-outside-bold regex format

- Test get-phase with **Goal**: format (colon outside bold)
- Test analyze with **Goal**: and **Depends on**: formats
- Test mixed colon-inside and colon-outside bold formats
- All 3 new tests fail confirming the regex bug

* fix(05-01): fix regex for goal/depends_on extraction in roadmap.cjs

- Fix 3 regex patterns to support both **Goal:** and **Goal**: formats
- Pattern: /\*\*Goal(?::\*\*|\*\*:)/ handles colon inside or outside bold
- Fix in both source (get-shit-done/) and runtime (.github/) copies
- All 28 tests pass including 4 new colon-outside-bold tests
- Live verification: all 4 phases return non-null goals from real ROADMAP.md

* feat(05-01): create gsd:autonomous command file

- name: gsd:autonomous with argument-hint: [--from N]
- Sections: objective, execution_context, context, process
- References workflows/autonomous.md and references/ui-brand.md
- Follows exact pattern of new-milestone.md (42 lines)

* docs(05-01): complete roadmap regex fix + autonomous command plan

* feat(05-02): create autonomous workflow with phase discovery and Skill() execution

- Initialize step with milestone-op bootstrap and --from N flag parsing
- Phase discovery via roadmap analyze with incomplete filtering and sort
- Execute step uses Skill() flat calls for discuss/plan/execute (not Task())
- Progress banner: GSD ► AUTONOMOUS ▸ Phase N/T format with bar
- Iterate step re-reads ROADMAP.md after each phase for dynamic phase detection
- Handle blocker step with retry/skip/stop user options

* test(05-02): add autonomous skill generation tests and fix skill count

- Test autonomous.md converts to gsd-autonomous Copilot skill with correct frontmatter
- Test CONV-07 converts gsd: to gsd- in autonomous command body content
- Update skill count from 32 to 33 (autonomous.md added in plan 01)
- All 645 tests pass across full suite

* docs(05-02): complete autonomous workflow plan

* docs: phase 5 complete — update roadmap and state

* docs: phase 6 context — smart discuss decisions

* docs(06): research smart discuss phase domain

* docs(06): create phase plan

* feat(06-01): replace Skill(discuss-phase) with inline smart discuss

- Add <step name="smart_discuss"> with 5 sub-steps: load prior context, scout codebase, analyze phase with infrastructure detection, present proposals per area in tables, write CONTEXT.md
- Rewire execute_phase step 3a: check has_context before/after, reference smart_discuss inline
- Remove Skill(gsd:discuss-phase) call entirely
- Preserve Skill(gsd:plan-phase) and Skill(gsd:execute-phase) calls unchanged
- Update success criteria to mention smart discuss
- Grey area proposals use table format with recommended/alternative columns
- AskUserQuestion offers Accept all, Change QN, Discuss deeper per area
- Infrastructure phases auto-detected and skip to minimal CONTEXT.md
- CONTEXT.md output uses identical XML-wrapped sections as discuss-phase.md

* docs(06-01): complete smart discuss inline logic plan

* docs(07): phase execution chain context — flag strategy, validation routing, error recovery

* docs(07): research phase execution chain domain

* docs(07): create phase plan

* feat(07-01): wire phase execution chain with verification routing

- Add --no-transition flag to execute-phase Skill() call in step 3c
- Replace step 3d transition with VERIFICATION.md-based routing
- Route on passed/human_needed/gaps_found with appropriate user prompts
- Add gap closure cycle with 1-retry limit to prevent infinite loops
- Route execute-phase failures (no VERIFICATION.md) to handle_blocker
- Update success_criteria with all new verification behaviors

* docs(07-01): complete phase execution chain plan

* docs(08): multi-phase orchestration & lifecycle context

* docs(08): research phase domain

* docs(08): create phase plan

* feat(08-01): add lifecycle step, fix progress bar, document smart_discuss

- Add lifecycle step (audit→complete→cleanup) after all phases complete
- Fix progress bar N/T to use phase number/total milestone phases
- Add smart_discuss CTRL-03 compliance documentation note
- Rewire iterate step to route to lifecycle instead of manual banner
- Renumber handle_blocker from step 5 to step 6
- Add 10 lifecycle-related items to success criteria
- File grows from 630 to 743 lines, 6 to 7 named steps

* docs(08-01): complete multi-phase orchestration & lifecycle plan

* docs: v1.24 milestone audit — passed (18/18 requirements)

* chore: complete v1.24 milestone — Autonomous Skill

* chore: archive phase directories from v1.24 milestone

* gsd: clean

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
2026-03-14 17:41:19 -06:00
Frank
0f38e3467e fix: strip unsupported Gemini agent skills frontmatter (#971) 2026-03-12 11:36:43 -06:00
Artspark
bc3d6db1c0 fix: emit valid codex agent TOML (#1008) 2026-03-12 11:36:04 -06:00
Lex Christopherson
e2b6179ba7 fix(install): replace $HOME/.claude paths for non-Claude runtimes
The installer only replaced ~/.claude/ (tilde form) when rewriting paths
for OpenCode, Gemini, and Codex installs. Source files also use
$HOME/.claude/ in bash code blocks (since ~ doesn't expand inside
double-quoted strings), leaving ~175 unreplaced references that break
gsd-tools.cjs invocations on non-Claude runtimes.

Adds $HOME/.claude/ replacement to all 6 path-rewriting code paths,
a toHomePrefix() utility to keep $HOME as a portable shell variable,
and a post-install scan that warns if any .claude references leak
through.

Closes #905

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-03 10:12:59 -06:00
Tibsfox
8b8d1074b8 fix(milestone): deduplicate phase filter and handle empty MILESTONES.md
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>
2026-03-03 08:01:54 -06:00
Tibsfox
630a705bdc fix(gemini): use AfterTool instead of PostToolUse for Gemini CLI hooks
Gemini CLI uses AfterTool as the post-tool hook event name, not
PostToolUse (which is Claude Code's event name). The installer was
registering the context monitor under PostToolUse for all runtimes,
causing Gemini to print "Invalid hook event name" warnings on every
run and silently disabling the context monitor.

Changes:
- install.js: use runtime-aware event name (AfterTool for Gemini,
  PostToolUse for others) when registering context monitor hook
- install.js: uninstall cleans up both PostToolUse and AfterTool
  entries for backward compatibility with existing installs
- gsd-context-monitor.js: runtime-aware hookEventName in output
- docs/context-monitor.md: document both event names with Gemini
  example

Closes #750

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-28 05:38:11 -08:00
Lex Christopherson
1455931f79 feat(codex): add request_user_input mapping, multi-agent config, and agent role generation
Expand Codex adapter with AskUserQuestion → request_user_input parameter
mapping (including multiSelect workaround and Execute mode fallback) and
Task() → spawn_agent mapping (parallel fan-out, result parsing).

Add convertClaudeAgentToCodexAgent() that generates <codex_agent_role>
headers with role/tools/purpose and cleans agent frontmatter.

Generate config.toml with [features] (multi_agent, request_user_input)
and [agents.gsd-*] role sections pointing to per-agent .toml configs
with sandbox_mode (workspace-write/read-only) and developer_instructions.

Config merge handles 3 cases: new file, existing with GSD marker
(truncate + re-append), existing without marker (inject features +
append agents). Uninstall strips all GSD content including injected
feature keys while preserving user settings.

Closes #779

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-27 15:29:35 -06:00
dungan
dba401fe54 fix: statusline migration regex too broad, clobbers third-party statuslines (#670)
The #330 migration that renames `statusline.js` → `gsd-statusline.js`
uses `.includes('statusline.js')` which matches ANY file containing
that substring. For example, a user's custom `ted-statusline.js` gets
silently rewritten to `ted-gsd-statusline.js` (which doesn't exist).

This happens inside `cleanupOrphanedHooks()` which runs before the
interactive "Keep existing / Replace" prompt, so even choosing "Keep
existing" doesn't prevent the damage.

Fix: narrow the regex to only match the specific old GSD path pattern
`hooks/statusline.js` (or `hooks\statusline.js` on Windows).

Co-authored-by: ddungan <sckim@mococo.co.kr>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-27 13:33:53 -06:00
Lex Christopherson
5cb2e740fd feat: improve onboarding UX with /gsd:new-project as default command 2026-02-25 07:20:14 -06:00
Lex Christopherson
e9e11580a4 merge: resolve conflicts with main for Codex skills-first support
- bin/install.js: integrate isCommand parameter (Gemini TOML fix) with Codex branch
- package.json: keep v1.20.6 with Codex description
- CHANGELOG.md: merge Codex entries into Unreleased above existing releases
2026-02-23 09:46:48 -06:00
vinicius-tersi
7542d364b4 feat: context window monitor hook with agent-side WARNING/CRITICAL alerts
Adds PostToolUse hook that reads context metrics from statusline bridge file and injects alerts into agent conversation when context is low.

Features:
- Two-tier alerts: WARNING (<=35% remaining) and CRITICAL (<=25%)
- Smart debounce: 5 tool uses between warnings, severity escalation bypasses
- Silent fail: never blocks tool execution
- Security: session_id sanitized to prevent path traversal

Ref #212
2026-02-20 14:40:08 -06:00