Flaky ci forensics
Skill oceanfsdfsvfdsvs/flaky-ci-forensics/.claude/skills/flaky-ci-forensics
Turn flaky CI failures into triage reports|把不稳定 CI 失败变成排查报告
npx -y skills add oceanfsdfsvfdsvs/flaky-ci-forensics --skill flaky-ci-forensicsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Analyze flaky CI/test failures from JUnit XML, test history, and CI logs. Use when a team needs to distinguish true regressions from flakes, estimate rerun cost, choose quarantine versus fix-now actions, or produce a prioritized triage plan.
SKILL.md
1.6 KB, as published. Nobody here has run it
Flaky CI Forensics
This is a Claude Code install mirror for flaky-ci-forensics. Keep the canonical implementation in flaky-ci-forensics/.
Trigger
Use when the user provides CI logs, JUnit XML, test history, or asks whether failing tests are flaky, should be rerun, should be quarantined, or need a fix-now plan.
Workflow
- Preserve the exact evidence: test names, failure messages, commit, retry count, runner metadata, and whether the same commit passed on retry.
- If file artifacts exist, run the canonical analyzer from the repository root:
python3 flaky-ci-forensics/scripts/flaky_ci_forensics.py \
--junit flaky-ci-forensics/scripts/fixtures/junit.xml \
--log flaky-ci-forensics/scripts/fixtures/ci.log \
--history flaky-ci-forensics/scripts/fixtures/history.csv
- Classify failures into the canonical buckets from
flaky-ci-forensics/references/flaky-test-rules.md. - Return the standard report: CI decision, failure cluster, cost and frequency, likely root cause, minimal fix plan, instrumentation, and guardrails.
Guardrails
- Do not label deterministic assertion mismatches as flaky without retry-pass or environment evidence.
- Do not quarantine auth, billing, payment, migration, security, deletion, or compliance coverage without explicit approval.
- Quarantine recommendations need owner, expiry, and tracking issue.