ci-triage

name	ci-triage
description	Investigate failing GitHub Actions runs and produce root-cause plus Beads follow-up.
disable-model-invocation	true

CI Triage

Use this skill when a branch or PR has failing CI and you need fast root-cause with actionable follow-up.

Inputs

• PR number or URL (preferred)
• Or branch name (e.g. feature/<feature-id>)

Workflow

1) Collect CI signal

gh pr checks <pr-number>
gh run list --branch <branch> --limit 20

2) Identify failing runs and jobs

gh run view <run-id>
gh run view <run-id> --log-failed

If logs are missing, query run metadata and check-suite details:

gh api repos/<owner>/<repo>/actions/runs/<run-id>
gh api repos/<owner>/<repo>/actions/runs/<run-id>/jobs

3) Classify root cause

Pick one primary class:

• Workflow config issue
• Test/regression failure
• Environment/dependency drift
• Permissions/secrets issue
• Flaky/infra transient

4) Propose smallest safe fix

• Point to exact workflow/job/step
• Provide command-level repro if possible
• Suggest one patch path and one fallback

5) Track in Beads

Create a follow-up item when CI is not green:

bd create --title="CI: <short root cause>" --type=bug --priority=1
bd dep add <new-id> <parent-feature-id>
scripts/beads_transport.sh export

Output Template

## CI Triage Result

- Branch/PR: ...
- Failing workflow(s): ...
- Root cause: ...
- Evidence: run/job/step links
- Proposed fix: ...
- Beads follow-up: <id>

Recovery

Symptom	Fix
Skill produces no output	Check working directory is project root with docs/workstreams/backlog/
"checkpoint not found"	Run sdp-orchestrate --feature <ID> to create initial checkpoint
"workstream files missing"	Run sdp-orchestrate --index to verify, then @feature to regenerate
Skill hangs / no progress	Check .sdp/log/events.jsonl for last event; use sdp reset --feature <ID> if stuck
Review loop exceeds 3 rounds	Use @review --override "reason", @review --partial, or @review --escalate