nickolashkraus/babysit-pr

You are monitoring a pull request, checking CI, and deploying when approved. Follow every step in order.

Step 1: Parse arguments

Parse $ARGUMENTS for:

1. PR identifier (required): A GitHub PR URL or number. If a URL, extract the owner/repo and PR number. If a bare number, detect the repo from the current git remote.
2. Deploy script (optional): Path to a deploy script. If not provided, skip deployment steps and just report CI status.

If no PR identifier is provided, stop and tell the user:

Usage: /babysit-pr <pr-url-or-number> [deploy-script]

Step 2: Check CI status

Fetch all checks on the PR:

gh pr checks <pr-number> --repo <owner/repo>

Classify every check as pass, fail, or pending.

• If any checks are pending, report which checks are still running and stop. The next loop iteration will re-check.
• If any checks have failed, go to Step 3.
• If all checks pass, go to Step 4.

Step 3: Summarize failures

For each failed check, report:

• Check name
• Failure summary (from the check output or log if available)

Do not attempt to fix failures. Just report them and stop. The next loop iteration will re-check after the developer pushes a fix.

Step 4: All checks pass

Present the deployment plan:

1. Show the latest commit (SHA, message, author).
2. Show the PR title and branch.
3. If a deploy script was provided, show the deploy command that will be run.

Then ask: "All CI checks pass. Deploy? [Waiting for approval]"

Do not proceed until the user explicitly approves (e.g., "yes", "approved", "deploy", "ship it").

Step 5: Deploy

If no deploy script was provided, tell the user all checks pass and stop.

If a deploy script was provided:

1. Verify the script exists at the given path. If not, tell the user and stop.
2. Run the deploy script. If it has an interactive confirmation prompt, pipe echo "y" into it.
3. Report the result: success or failure with relevant output.

After a successful deploy, stop monitoring. The loop can be cancelled.