mirror of https://github.com/garrytan/gstack.git
e7b5451431
Authors per-milestone acceptance criteria — each measurable and paired with the
check that validates it (the deliverable) — into the design doc, in a format
/autobuilder-loop parses. Closes a real gap: autobuilder-loop re-reads
per-milestone {next_milestone, acceptance_criteria} from the plan each iteration,
but nothing upstream authored that field.
Runs after /office-hours, before the plan reviews. Per-piece loop: draft criteria
-> default-on cross-model gap-check (Codex gpt-5.6-sol ultra + Grok grok-build,
honoring the existing codex_reviews/grok_reviews kill-switches) -> probe the user
for judgment calls -> pair each criterion with a repo-native validating check (or
an explicit manual-verification) -> write the block. Idempotent and resume-safe.
Wires the two skills together via <!-- status: pending|built-gate-pending|complete -->
milestone markers: autobuilder-loop step 1.0 selects the first pending milestone;
step e and the deferral valve write status back.
Registered in the router, coverage matrix, office-hours/plan-eng-review chaining,
README and CLAUDE.md.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
||
|---|---|---|
| .. | ||
| SKILL.md | ||
| SKILL.md.tmpl | ||