What problem does it solve?
Operating the repository's headless gator review agent requires careful preflight checks, gateway validation, scoped launch commands, and log-based supervision; this Skill encodes that entire operator workflow so gator launches are safe, validated, and correctly scoped.
Core Features & Use Cases
- Preflight Validation: Verifies repository root, host tools (openshell, gh, jq, ruby), GitHub and Codex authentication, gateway health, and existing gator sandboxes before launching.
- Scoped Supervised Launches: Starts gator watchers for PRs, issues, or linked pairs using
scripts/agents/run.sh with --watch --background, strict input normalization, and DNS-1123 sandbox naming.
- Monitoring and Restarting: Interprets log sentinels (waiting, blocked, transient_failure, terminal_failure, complete), inspects active sandboxes, and restarts stale or wedged gators after payload version bumps.
- Use Case: A maintainer asks to run gator on PR #1234; the Skill validates the gateway and credentials, launches a supervised sandbox named
gator-pr-1234-supervised, reads the background log, and reports the sandbox status and latest sentinel.
Quick Start
Launch a supervised gator agent to review and monitor pull request number 1234 on my registered OpenShell gateway.