67e2ff7b25
* test(#2855): add failing-first regression test for cross-workstream archive leak Covers findPhaseInternal/getArchivedPhaseDirs in src/phase-locator.cts resolving a pending workstream phase to an unrelated workstream's (or flat-mode's) archived phase because the archive fallback hardcodes the project-root .planning/milestones/ tree. Fails against the current implementation; the fix lands in a follow-up commit. * fix(#2855): scope phase-locator archived-milestone fallback to the active workstream findPhaseInternal and getArchivedPhaseDirs in src/phase-locator.cts hardcoded the project-root .planning/milestones/ tree when falling back to search archived phases, ignoring GSD_WORKSTREAM. A pending phase in one workstream whose own phases/ directory didn't exist yet would silently resolve to a same-numbered phase archived under an unrelated workstream's (or flat-mode's) history, complete with stale plan/summary counts and an archived status. Route the archive fallback through planningDir(cwd) instead — the same workstream-aware helper the active-phase search (three lines above) and the archive-write path (archivePhaseDirectories in milestone.cts) already use. Flat/non-workstream projects are unaffected: planningDir(cwd) with no GSD_WORKSTREAM resolves to the same root .planning path as before. Also switch the reported relBase/basePath from a hardcoded '.planning/milestones/...' literal to path.relative(cwd, archivePath), so the paths returned to callers stay consistent with wherever the archive actually resolved to (root or workstream-scoped). * chore(#2855): add changeset for phase-locator workstream archive fix * fix(#2855): normalize getArchivedPhaseDirs basePath to posix separators Orthogonal code-review finding: findPhaseInternal's relBase/directory field was explicitly toPosixPath-normalized, but getArchivedPhaseDirs's basePath used a bare path.relative() call, leaving it native-separator on Windows — an inconsistency between two sibling "relative path from cwd" report fields introduced by the same #2855 fix. Wrap basePath in toPosixPath to match, and update the two existing assertions that compared basePath against path.join output (which would break on Windows now that the field is guaranteed posix) to compare against forward-slash literals instead, matching how the sibling `directory` field is already asserted elsewhere in this suite. * refactor(#2855): share archive-directory resolution between findPhaseInternal and getArchivedPhaseDirs Orthogonal code-review finding: the two functions carried independent copies of the same resolve-milestonesDir-then-enumerate-archive-dirs logic — the exact shape that let the original #2855 bug (hardcoded root path) exist in one copy while the workstream-aware active-phase search sat three lines above it. Extract listArchiveVersionDirs(cwd) as the single seam both functions now consume, so a future change to how the archive tree is located only needs to happen once. Byte-for-behaviour preserved: readSubdirectories and searchPhaseInDir already self-contain their own try/catch and never throw, so moving the iteration outside the old inline try block changes nothing observable (verified via manual repro scripts covering leak prevention, positive resolution, flat-mode parity, and multi-milestone reverse-sort ordering). * test(#2855): demonstrate ROADMAP.md presence does not affect the archive-leak guard Orthogonal code-review (spec axis) finding: issue #2855's AC1 states the guard must hold "regardless of whether workstream A's roadmap already lists the phase and when it doesn't yet" — an explicit two-value dimension that had no direct test coverage; it was only inferable by reading findPhaseInternal's source and confirming it never touches ROADMAP.md. Add a parametrized test creating the workstream's ROADMAP.md with and without a matching Phase heading, asserting the archive-leak guard resolves identically (null) either way. * chore(#2855): backfill changeset PR number to 3008 --------- Co-authored-by: sim <sim@local>
GSD Core
Git. Ship. Done.
English · Português · 简体中文 · 日本語 · 한국어
A light-weight meta-prompting, context engineering, and spec-driven development system for Claude Code, OpenCode, Antigravity CLI, Kimi CLI, Kilo, Codex, Copilot, Cursor, Windsurf, and more.
What is GSD Core
GSD Core is a context-engineering and spec-driven development framework that drives AI coding agents (Claude Code, Codex, Antigravity CLI, Kimi CLI, Copilot, Cursor, and more) through a disciplined phase loop. It solves context rot — the quality degradation that accumulates as an AI fills its context window — by running all heavy research, planning, and execution work in fresh-context subagents while keeping your main session lean.
How it works
Each milestone repeats the same five-step loop, one phase at a time:
- Discuss — capture implementation decisions before anything is planned
- Plan — research, decompose, and verify the plan fits a fresh context window
- Execute — run plans in parallel waves; each executor starts with a clean 200k-token context
- Verify — walk through what was built; diagnose and fix before declaring done
- Ship — create the PR, archive the phase, repeat for the next one
Quickstart
npx @opengsd/gsd-core@latest
The installer prompts for your runtime (Claude Code, OpenCode, Antigravity CLI, Kimi CLI, Kilo, Codex, Copilot, Cursor, Windsurf, and more) and whether to install globally or locally. The installer is required for cross-runtime compatibility — do not copy files from agents/ or commands/ directly.
On another runtime or without Node.js? See Install on your runtime.
Once installed, start a new project or onboard an existing repo:
/gsd-new-project # greenfield project
/gsd-onboard # existing codebase
New here? Follow Your first project for a guided walkthrough from install to first shipped phase, or Onboarding an existing codebase for brownfield setup.
Documentation
What's new in 1.7.0 → docs/whats-new-1.7.0.md
Tutorials — learning by doing:
How-to guides — task-focused recipes:
Reference — authoritative facts:
Explanation — concepts and design decisions:
Full index: docs/README.md. Other languages: 日本語 · 한국어 · Português · 简体中文.
Why it works
Most AI-coding setups fail at scale because context bloat silently degrades output quality, there is no shared memory between sessions, and nothing verifies that code actually works. GSD Core solves all three: heavy work runs in fresh subagents, structured artifacts like STATE.md and CONTEXT.md survive session boundaries, and the verify step walks through what was built and generates fix plans before a phase is declared done. See docs/explanation/context-engineering.md for the full reasoning.
Troubleshooting? See docs/how-to/recover-and-troubleshoot.md.
Community
| Project | Platform |
|---|---|
| gsd-opencode | Original OpenCode port |
| Discord | Community support |
Star History
License
MIT License. See LICENSE for details.
Claude Code is powerful. GSD Core makes it reliable.