agentleFS
woodbearywoodbeary/later-dog/.claude/skills/monitor-and-ship/SKILL.md

monitor-and-ship

Skill2 starsChanged 29 days agoApache-2.0View on GitHub

Take a published PR through its checks, current-commit verification and independent review to ready-to-merge, repair what fails, and hand the merge to the person.

The file

Files
SKILL.md

Monitor and ship

Use once a cloud job has a published draft PR. Stay with it until the PR is ready for the person or blocked on something only they can fix. Shipping here ends with a verified, reviewed draft PR handed over; merging stays with the person. Use the later.dog MCP server. Discover tools by the names below; client-specific prefixes may vary.

Procedure

  1. Look before acting. inspect_cloud_job for the job, laterdog_workspace for everything else in flight. Do not start a verification, review or repair that already exists, and reuse the same requestKey when you retry a call.
  2. Make sure behaviour gets checked. The repository needs a recipe: behavioural CI check names or isolated commands taken from its feature map. Lint and type checks alone prove no behaviour. Setting the recipe (configure_verification_recipe) needs the admin connection; a dog asks its person to enter the checks in Cloud jobs → Connect repository.
  3. Verify the current commit with verify_cloud_job. It records the PR head, its check results and the configured isolated commands. No checks, a missing dependency, a moved head or an unavailable environment come back blocked, and blocked is not passed. Read the evidence, not only the verdict.
  4. Review the same commit with the independent-review skill.
  5. Repair what fails. Investigate first, then call correct_cloud_job with a brief that carries the original requirements and the actual failure (the failing check, the log lines, the review finding). The repair is a new run that updates the same PR. It voids earlier verification and review, so start again at step 3 for the new head.
  6. Hand it over when verification and review both passed at the PR's current commit: the PR, the commit, the evidence and anything it does not cover. Do not call merge_cloud_job. The person merges, on GitHub or with "Merge under repository policy" in Cloud jobs where they have granted it.
  7. Pick up after interruptions from inspect_cloud_job. The supervisor's receipts are durable, so a disconnect or restart is no reason to repeat calls. Count a merge only once the job shows merged.

Report

Which PRs are ready for the person, which are blocked and on what, which repairs are running, and the limits of the evidence.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.