Merge Pr

When the user asks to merge a pull request (e.g., "merge PR ", "merge this PR", or monitor its merge):

1. Enqueue for Merge: Run gh pr merge <pr_number> --auto --squash to enable auto-merge or add the pull request to the merge queue.
   • Force / Admin Merge: CRITICAL: Passing --admin to bypass the merge queue or required checks requires explicit user permission or consent. Only pass --admin (e.g., gh pr merge <pr_number> --admin --squash) if the user has explicitly requested or approved a force/admin merge. Never invoke --admin autonomously.
2. Invoke a Background Shepherd: Launch a background subagent with the role Merge PR Shepherd to continuously watch the PR until it merges.
3. Leverage Existing CI Skills:
   • Have the subagent use the monitor-ci-results skill to watch for CI check failures and generate analysis reports.
   • Have the subagent use the buildkite-retry-job skill (retry_buildkite_jobs.py <pr_number>) to automatically retry any transient network flakes (e.g., HTTP 504 gateway timeouts, downloader errors).
   • Soft-Failing Jobs: Experimental Buildkite jobs (e.g. *rolling* Bazel) are non-blocking soft failures; do not treat them as merge blockers.
   • When the PR is queued, actively discover the merge queue branch via gh api repos/:owner/:repo/branches --jq '.[].name | select(test("gh-readonly-queue/.*/pr-<pr_number>-"))' and monitor commit statuses/Buildkite builds running on that temporary branch.
4. Queue Shepherding: Periodically check gh pr view <pr_number> --json state,autoMergeRequest,mergeStateStatus,mergeable. While state is "OPEN", ensure auto-merge is enabled / queued by running gh pr merge <pr_number> --auto --squash. If autoMergeRequest is null (e.g., ejected from the merge queue due to a CI flake in the temporary queue branch), re-enqueue it for merge by running gh pr merge <pr_number> --auto --squash once checks are retried or green.
5. Completion Notification: Once state becomes "MERGED", send a high-priority message back to the parent conversation.