- Home
- Skills
- Code Quality & Review
- Flaky Test Triage
Works with the AI Agents you already use
Flaky Test Triage
Group flaky failures by cause, rank by real cost, and fix the suite with a quarantine that expires.
$12
Flaky Test Triage
Example session with this skill installed
About 8 percent of our CI runs fail without a code change. Engineers just hit re-run. Here are 3,000 test results with failure messages and shard info.
- Read your context and instructions
- Compiled the flaky test triage
Worked example — illustrative data.
Flaky test triage: web app suite
Verdict
8.2% of runs flaky, costing about 11 engineering hours a week. Top cause: fixed sleeps and shared fixture state. 14 tests quarantined with owners and dates.
Clusters
| Cluster | Tests | Failures/wk | Cause | Cost | Action |
|---|---|---|---|---|---|
| checkout.spec timeouts | 9 | 120 | fixed 500ms sleep | 4.1h | wait for condition |
| DB fixture collisions | 6 | 80 | shared rows across shards | 3.0h | unique data per test |
| date formatting | 4 | 40 | uses current date | 1.2h | inject clock |
| upload test | 2 | 30 | temp path collision | 0.9h | unique temp dir |
| race in cart service | 1 | 12 | real product race | 1.8h | fix product |
Fixes
Checkout: replace sleeps with condition waits and a 10s deadline. Fixtures: per-test unique keys and transactional rollback. Date: frozen clock. Upload: random temp dir per test. Cart: the one real bug, fixed in the service.
Verification
Each fix run 200 times in CI before the cluster is closed; a fix that fails once goes back to quarantine.
Quarantine policy
Separate non-gating job, ticket with owner, two-week cap, then fix or delete. Quarantine count tracked weekly.
Retry policy
Retries logged and alerted above 2%. A rising retry rate is a defect, not a comfort.
Metric to watch
Flake rate weekly; target under 1% within three sprints.
Connects securely to your tools. The creator never sees your data.
About this skill
Flaky test triage: cluster failures by cause, rank by real cost, and prescribe quarantine or deletion with a reason for each.
Use this skill when:
- CI is flaky
- tests fail intermittently
- decide which tests to quarantine
- a suite is red and slow
Who it's for: Teams with an unreliable CI suite.
Example request
CI fails 30% of the time. Tell us what to fix first.
What the example returns
- Six causes, 71% from two
- Quarantine list ordered by cost
- Two fixes, three deletions
In the package: five files — SKILL.md with the trigger and workflow, README.md, references/METHOD.md (the full method), references/CHECKLIST.md (the pass/fail gate) and examples/EXAMPLE.md (a worked run you can copy).
Works with Claude Code, Cursor, Codex CLI and any agent that reads SKILL.md. Plain markdown: no API key, no account, no network access.

Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Code Quality & Review.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 days ago