| name | test-hardening |
| description | Use when a test suite is flaky, shallow, or easy to bypass and you need stronger coverage around critical behavior, failure modes, and regression boundaries. |
Test Hardening
Strengthen trust in the system by improving the tests that matter most, not by maximizing test count.
Selective Reading Rule
| File | Purpose | Read When |
|---|
verify.md | decide whether the new tests truly prove behavior | after authoring or tightening tests |
Focus Areas
- flaky tests and nondeterministic setup
- critical business flows with weak assertions
- edge cases around errors, retries, empty states, and bad inputs
- integration points that are mocked too aggressively
- missing regression coverage for recent incidents
Workflow
- Identify the highest-risk behaviors that lack reliable proof.
- Remove avoidable flakiness from setup, timing, and shared state.
- Add tests at the lowest level that still proves the behavior.
- Tighten assertions so tests fail for the right reason.
- Verify the suite is still maintainable and fast enough to run often.
Deliverable
Summarize:
- what was previously unproven
- what is now covered
- any remaining blind spots
- follow-up tests worth adding later
Preferred Pairings
- Workflow:
.agents/workflows/test.md or .agents/workflows/verify.md
- Primary subagent:
test_writer
- Common supporting skills:
testing-patterns, tdd-workflow, webapp-testing
- Pair with
debugger when the expected behavior is still not well understood
Avoid
- snapshot-heavy tests with weak intent
- duplicating implementation details across tests
- adding slow end-to-end coverage when a targeted integration test would prove the same behavior
