Use this checklist before tagging a release.
-
python3 tools/sync_schemas_from_rfcs.pyproduces no uncommitted schema drift. -
python3 tools/validate.pypasses. -
python3 tools/diff_checker.py <before> <after> --strict --min-severity majorreviewed. - Tier A examples are present and valid.
- Synthetic dataset manifest and safety checklist reviewed.
- Benchmark task spec and scoring scripts are reproducible.
- Leaderboard run card template is up to date.
- RFC lifecycle status updates applied (
draft,implementation_required,stable, etc.). - Migration notes written for major schema changes.
-
CHANGELOG.mdupdated. - README and roadmap links verified.
- Local OSS runbook executed (or dry-run validated).
- API baseline runbook executed (or dry-run validated).
- Reference run artifacts and hashes published.