| name | ci-failure-triage |
| display_name | CI Failure Triage |
| description | Diagnoses CI/CD failures from logs and workflow files, identifies likely root cause, and recommends minimal fixes or retry guidance. |
| category | devops |
| version | 1.0.0 |
| tags | ["ci","github-actions","debugging","deployment","automation"] |
| providers | ["claude-code","codex","copilot","openclaw","hermes-agent"] |
| author | agent-skills contributors |
CI Failure Triage
Use this skill when a CI or deployment job fails and the user provides logs, workflow files, or job metadata.
Inputs
- CI logs, failing job and step names, workflow YAML, recent commits, dependency changes, and environment details
Steps
- Locate the first failing step and first meaningful error.
- Classify the failure: test regression, build error, dependency resolution, cache issue, credentials/permissions, environment drift, rate limit, flaky test, or deployment failure.
- Compare the failing step to workflow configuration and recent changes.
- Identify whether retrying is reasonable or whether a code/config fix is required.
- Recommend the smallest fix and any workflow hardening that prevents recurrence.
- Provide commands or workflow rerun steps when appropriate.
Output
Return:
- Failure classification
- Root cause hypothesis with evidence
- Minimal fix plan
- Retry guidance
- Preventive hardening suggestions
Guidelines
- Do not focus on the last error if earlier setup failed.
- Be explicit when logs are insufficient.
- Separate likely root cause from secondary noise.
