pr-autopilot

Purpose and capabilities are internally aligned for a PR-autopilot skill, and external tool trust is mostly sound because it uses official GitHub tooling. The main risk is not malware but scope: it gives an AI agent autonomous write/exec and public-action authority over a repo while consuming untrusted PR content and potentially executing PR-controlled build scripts. Overall this is a coherent but high-risk automation skill that should be treated as dangerous in operation, especially on untrusted branches or fork PRs.