| name | verify-pipelines |
| description | Monitor GitHub Actions and Jenkins pipelines for the current branch/commit and confirm both are green before closing a work item. Trigger on any request to 'check pipelines', 'verify CI', 'watch the build', 'is CI green', 'check before closing', or 'monitor pipelines'. |
Verify Pipelines Before Closing
Run this skill after pushing changes. It watches both GitHub Actions and Jenkins until they settle, then gives a clear pass/fail verdict. Do not consider a work item done until both are green.
Inputs
Determine these from context before starting:
- COMMIT — the SHA of the head commit on the current branch (
git rev-parse HEAD)
- BRANCH — the current branch name (
git branch --show-current)
- REPO —
flying-dice/dcs-dropzone
- JENKINS_JOB —
dcs-dropzone
Step-by-step
1. Resolve commit and branch
git rev-parse HEAD
git branch --show-current
2. Find the GitHub Actions run for this commit
gh run list --repo flying-dice/dcs-dropzone --branch <BRANCH> --limit 5 --json databaseId,status,conclusion,headSha,displayTitle,createdAt
Find the run whose headSha matches COMMIT. If it isn't listed yet, wait 10 seconds and retry (the push may still be propagating).
3. Watch GitHub Actions until complete
gh run watch <RUN_ID> --repo flying-dice/dcs-dropzone
This blocks until the run finishes. If it exits before finishing (e.g. already done), follow up with:
gh run view <RUN_ID> --repo flying-dice/dcs-dropzone
Record the conclusion (success / failure / cancelled).
4. Find the Jenkins build for this commit
Use the mcp__jenkins__getJob tool to get the latest build:
mcp__jenkins__getJob(jobFullName: "dcs-dropzone", tree: "lastBuild[number,result,building,timestamp]")
Check if the latest build was triggered after the push timestamp. If building: true, poll every 15 seconds until building: false:
mcp__jenkins__getJob(jobFullName: "dcs-dropzone", tree: "lastBuild[number,result,building]")
If the build number hasn't changed from what existed before the push (i.e. Jenkins hasn't picked up the SCM change yet), wait 20 seconds and retry — Jenkins polls on a schedule.
Record the result (SUCCESS / FAILURE / UNSTABLE).
5. On any failure — pull the tail log and diagnose
GitHub Actions failure:
gh run view --job=<JOB_ID> --repo flying-dice/dcs-dropzone --log 2>&1 | tail -100
Jenkins failure:
mcp__jenkins__getBuildLog(jobFullName: "dcs-dropzone", buildNumber: <N>, limit: -80)
Read the output. Identify the root cause. Fix it, push, and re-run this skill from step 1.
6. Verdict
Both must be green to close:
| GitHub Actions | Jenkins | Verdict |
|---|
| success | SUCCESS | ✅ Both green — work item can be closed |
| success | UNSTABLE | ⚠️ Jenkins unstable — check JUnit failures |
| failure / * | * | ❌ GitHub Actions failed — do not close |
| * | FAILURE | ❌ Jenkins failed — do not close |
Polling guidance
- Check every 15 seconds when polling Jenkins manually.
- Use
gh run watch for GitHub Actions (it handles polling internally).
- Timeout after 15 minutes — if either pipeline is still running, report the current state and ask the user how to proceed.
Rules
- Never declare a work item done on a green local test run alone — CI must confirm it.
- If GitHub Actions and Jenkins disagree (one green, one red), investigate both before deciding.
- A Jenkins result of
UNSTABLE is a failure for the purposes of closing work — treat it the same as FAILURE.
- Always report the build URLs so the user can inspect the full results themselves:
- GitHub Actions:
https://github.com/flying-dice/dcs-dropzone/actions/runs/<RUN_ID>
- Jenkins:
https://falcon.beluga-sirius.ts.net/job/dcs-dropzone/<BUILD_NUMBER>/
