By name: metravel-quality-fixer description: Run metravel lint, Jest, and Playwright validation end to end, fix real failures in scope, rerun the affected checks, and leave the repository with a clean quality-gate baseline or an explicit unrelated blocker.
Metravel Quality Fixer
Read AGENTS.md, docs/RULES.md, docs/TESTING.md, docs/DEVELOPMENT.md, docs/RELEASE.md, and the task triage in docs/CODEX.md before using this skill.
Use this skill when the task is to validate the repository broadly and fix what breaks, rather than to implement one feature first.
When to use
- User explicitly asks to run
lint, tests, ande2e - A branch or local workspace must be brought back to a green validation baseline
- Multiple quality gates fail across code, tests, and browser flows
Workflow
- Start from repo root and check branch +
git status --short. - Apply the operation coordination rule from
AGENTS.md/docs/RULES.md; if another full/preflight/e2e gate is already active in this workspace, do not start a duplicate. - Run the full gate requested by the user (
npm run lint,npm run test:run,npm run e2e) unless docs or a blocker require a narrower first pass. - Fix real failures in the touched or failing scope; do not mask them with skips, retries-only hacks, or allowlist drift.
- After each fix, rerun the failing command first, then rerun the broader confidence check that still matches the task scope.
- If visible web behavior changed, verify the scenario in a real browser and confirm no new console errors.
Repo specifics
- Keep temporary traces, screenshots, and reports only in ignored folders such as
playwright-report/,test-results/,.codex-temp/, or.codex-debug/. - For governance-sensitive fixes, rerun
npm run guard:external-linksornpm run governance:verify. - For flaky
e2e, capture the exact failing spec, rerun it narrowly, fix the root cause, and then return to the broader suite. - If a failure is unrelated and cannot be safely fixed in this task, report the exact command, failing test/spec, risk, and next verification step.