| name | quality-gates |
| description | This skill should be used after implementation changes or after any fix that could affect correctness. Runs and interprets automated quality gates in the canonical order. |
Quality Gates
Use this skill during Quality Gates.
Reference:
- CLAUDE.md (workflow contract)
- .claude/workflow-state.json
Canonical order
- type check
- lint
- tests
- verification handoff
Process
- Discover which checks actually exist in the repository.
- Run each applicable gate in order.
- Mark missing gates as
not-applicable with a reason.
- If a gate fails:
- capture the exact failure
- identify the likely root cause
- return to Implementation or Planning as needed
- rerun the entire automated gate sequence after the fix
- Update workflow state with the latest gate results.
- Report weak automation that comes from hard-to-test code shape as a real design issue when it affects confidence.
Output checklist
- checks run
- status per gate
- failure evidence, if any
- next phase decision
- testability debt or coverage-shape concerns, if any
Do not
- skip straight to tests because earlier gates looked safe
- claim success from partial checks
- deliver before verification is complete
