| name | test-runner |
| description | Run focused test suites, diagnose failures, and verify fixes. |
Test Runner
Use this skill when validating a change or investigating a failing test suite.
- Identify the project's test entry points from
package.json, Makefiles, CI config, or language conventions.
- Prefer the smallest relevant test command first instead of jumping straight to the full suite.
- Record the exact command used so results can be reproduced.
- Capture failing test names, stack traces, and the first meaningful error line.
- Separate product failures from environment failures such as missing services, ports, or credentials.
- Check whether the failure is deterministic before labeling it flaky.
- If a test fails after a code change, inspect whether the behavior or the expectation is wrong.
- Fix root causes before updating snapshots, fixtures, or golden outputs.
- Re-run the focused test after each fix.
- Re-run the broader suite before declaring the change safe.
Useful diagnosis buckets:
- Assertion mismatch
- Setup or teardown leak
- Timing or race condition
- Mock drift from production behavior
- API contract change
- Environment or dependency failure
When reporting results:
- Include the command.
- Include pass/fail status.
- Call out flaky or skipped tests explicitly.
- Note any tests you could not run and why.
If the suite is too slow, narrow by file, test name, package, or affected subsystem first.