What problem does it solve? Managing a pull request end-to-end requires constant manual polling of CI status, review comments, and deployment signals, which fragments developer attention and delays merges. ## Core Features & Use Cases - Automated PR Watching: Polls GitHub PR state every 60 seconds via the GitHub CLI, emitting JSON snapshots with recommended actions for CI failures, new reviews, and merge readiness. - CI Failure Triage: Classifies failures as branch-related or flaky using documented heuristics, inspects failed logs, and reruns transient failures within a bounded retry budget. - Review Score Tracking: Extracts explicit reviewer scores (e.g., 4/5) from comments and treats sub-maximal scores as unfinished work before merge. - Post-Merge Deployment Monitoring: Watches GitHub deployment records and deployment-like workflow runs tied to the merge commit, alerting on failures or prompting production validation. - Use Case: After pushing a feature branch, run the watcher to automatically detect a flaky CI failure, retry it once, address a reviewer's inline comment, and confirm the Vercel deployment succeeded after merge. ## Quick Start Ask the AI to create a pull request for the current branch and babysit it through CI, review feedback, and post-merge deployment validation.