- Home
- Skills
- Testing & Debugging
- CI Failure Detective
Works with the AI Agents you already use
CI Failure Detective
Triage CI build logs to pinpoint first-cause errors and generate evidence-based diagnostic plans.
$8
CI Failure Detective
Example session with this skill installed
Triage this GitHub Actions failure. The build failed at the 'npm test' step. Recent changes were only in the auth-service logic. Here is the log excerpt: [pasted log with Vitest error]
- Read your context and instructions
- Compiled the ci failure detective
I have reviewed the provided log excerpt from the npm test step.
Triage summary
Failure class: Code/test failure
Likely cause: Logic regression in authentication middleware
Confidence: High
Regression established: Yes, correlates with reported auth-service changes.
First meaningful error
The error occurs at src/services/auth.test.ts:42.
AssertionError: expected 401 to be 403.
This is the first failure in the test suite; subsequent failures in user.test.ts appear to be cascading errors from the failed auth state.
Ranked hypotheses
| Rank | Hypothesis | Evidence for / missing | Confidence | Cheapest check |
|---|---|---|---|---|
| 1 | Incorrect status code in middleware | Log shows 401 returned instead of 403 | High | npm test auth.test.ts |
| 2 | Test mock mismatch | Missing: mock definition for RBAC service | Medium | Inspect auth.test.ts mocks |
| 3 | Environment variable missing | Missing: AUTH_SCOPE in CI env | Low | Check CI env tab |
Next steps
- Local reproduction: Run
npx vitest src/services/auth.test.tsto confirm the status code mismatch locally. - Code audit: Verify
src/services/auth.tsline 88 where the error handling logic was recently modified. - Provider check: Ensure the CI secret
JWT_SECRETis correctly scoped to the branch.
Unknowns
Full dependency lockfile was not provided; however, current evidence points strongly to a logic regression rather than a toolchain issue.
Next steps
- Run the suggested local test command.
- Inspect the middleware change for recent status code overrides.
Connects securely to your tools. The creator never sees your data.
About this skill
CI logs are noisy, often burying the root cause under hundreds of lines of downstream fallout and cleanup errors. Manually triaging these failures wastes developer hours on guesswork and chasing red herrings.
What it does
- Pinpoint first failures by isolating the earliest causal error from secondary crashes and job summaries.
- Classify failure types into categories like code regressions, dependency mismatches, resource limits, or network timeouts.
- Rank hypotheses using an evidence-based matrix that tracks confidence levels and supporting log data.
- Formulate diagnostic plans by recommending the cheapest, safest commands to verify a specific failure theory.
- Correlate changes by comparing failing logs against provided diffs or recent successful runs to establish regression attribution.
How it works
- Input logs and context by providing failure excerpts, job names, or recent commit summaries for the agent to analyze.
- Review the Triage Summary to quickly see the failure class, confidence level, and whether a code regression is likely.
- Execute the Next Steps following an ordered plan that separates local reproduction from provider-side checks.
Frameworks & tools
This tool works with logs from GitHub Actions, GitLab CI, CircleCI, Jenkins, and other major providers. It analyzes output from compilers, linters, test runners, and dependency managers like npm, pip, and cargo.
Why this beats prompting it yourself
Generic prompts often hallucinate root causes based on the final line of a log. This skill enforces a strict evidence boundary, preventing the agent from claiming a fix without proof or echoing sensitive secrets back into the chat.
Use cases
- Diagnose a failing CI job without scrolling through thousands of lines of verbose output.
- Distinguish between a genuine code bug and an intermittent flaky test or runner resource limit.
- Triage dependency and toolchain errors during environment setup or lockfile updates.
- Generate a safe reproduction plan for a teammate to follow when a build breaks.
Known limitations
Cannot claim a confirmed root cause without repository context or a baseline green run. Does not automatically modify CI configurations or trigger new deployments.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.
Requires log excerpts or context to function. Cannot claim root causes or flakiness without evidence from multiple runs or repository baselines. Does not perform actions in the CI provider.
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 14 days ago
Permissions
No special permissions declared or detected