FactoryWager Wiki

Repository review (JIT)

Upstream playbook: repository-review. Transform only — do not clone the full playbook.

Follow one representative job from request → outcome. Ask each question; open one owner.

1. How does the agent classify the task?
   → AGENTS.md · docs/AGENTS.md
2. Which root guidance routes to domain context?
   → docs/harness/README.md · bun run harness:status
3. Can it find the existing owner of the concept?
   → lib/docs/repo-docs.ts · brand map · WIRE_BOUNDARY
4. Can it observe the behavior without a human relay?
   → bun run help · named CLI / harness:status ratchets
5. One domain model / one source of truth?
   → brands · wire eslint · path-bun · bun-env · UNIFIED
   Ratchet → bun run check:path-bun · bun run check:bun-env · brand staged gate
6. Does proof exercise the claim at the right boundary?
   → PROOF.md · FRESH-RERUN.md · lib/harness/proof.ts · bun run proof:install
7. Review, CI, conflicts, delivery?
   → husky / pre-commit-harness · .github/workflows/ · AUTHORITY lanes
   Ratchet → bun run ci:harness:fast
8. Which ops need human judgment / new authority?
   → AUTHORITY.md
9. Durable improvement for the next run?
   → FEEDBACK.md · bun run harness:lesson

Stop conditions and credentials: AUTHORITY. “Done” without matching evidence: PROOF.

This site is open source. Improve this page.