- •Collect failure output (CI logs, local run output, stack traces).
- •Cluster failures into groups by root signature.
- •Categorize each group:
- •Environment/setup
- •Dependency/version drift
- •Flaky/timing
- •Deterministic code defect
- •Test expectation mismatch
- •Produce an ordered plan:
- •Fix blockers first (env/setup)
- •Then highest-impact deterministic failures
- •Then flaky stabilization
- •For the first group, recommend the exact next skill:
- •
debug-looporflaky-test-hunter
- •
failure-triage
快速梳理大量故障,制定高效的应对方案。
SKILL.md
--- frontmattername: failure-triage description: Triage many failures quickly and produce an efficient attack plan.