| name | routr-test |
| description | Write and run tests (TDD, unit, E2E, Playwright). Use when: add tests, TDD, coverage, failing test suite, 'write tests for this'. Not for: browser smoke test (→ routr-qa); diagnosing why a test fails (→ routr-debug). |
routr-test
Goal: tests that prove behavior — right order, right level.
When to activate
- Adding test coverage, TDD, "write tests for this"
- Failing suite that needs new/updated tests (not root-cause debugging)
- Setting up or extending Playwright/E2E
Do not activate
- Click through the running app / smoke test →
routr-qa
- Diagnosing why an existing test fails →
routr-debug
- Fixing the code so a test passes and shipping the diff →
routr-ship
Iron law
Failing test before implementation when doing TDD — never write the pass first.
0. Bootstrap
See routr-catalog/references/skill-registry.md:
| Order | Skill | Tier |
|---|
| 1 | test-driven-development | required |
| 2 | tdd | recommended |
| 3 | webapp-testing | recommended |
| 4 | playwright-best-practices | optional |
| 5 | playwright-cli | optional |
Budget gate: test-driven-development + the test level's own docs for a routine test. Add Playwright rows only for E2E work.
If webapp-testing missing → read routr-depth-test OR references/fallback.md.
Failing tests debug → routr-debug + systematic-debugging.
1. Choose test level (medium freedom)
| Level | When |
|---|
| Unit | Pure logic, utils, mocked deps |
| Integration | API routes, DB, services |
| E2E | Critical user journeys |
2. TDD loop (low freedom)
- Failing test first
- Minimal pass
- Refactor green
3. Verify (low freedom)
Output format
## Test report
**Level:** unit / integration / E2E
**Coverage added:** files / behaviors
**Run command:** …
**Verified:** yes/no — failing-first confirmed, now green
Handoff
| Need | Router |
|---|
| Ship the change | routr-ship |
| Review test PR | routr-review |
| Test fails for unknown reason | routr-debug |
| Browser smoke test / preview check | routr-qa |
References
Anti-patterns
- Writing the implementation before a failing test in a TDD flow
- Testing implementation details instead of behavior
- Retrying a flaky test instead of fixing its root cause
- Loading Playwright rules for a pure-unit-test task
