babysit-pr

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). Yes: the required runtime workflow runs scripts/pr-snapshot which fetches outsider-authored PR review threads and top-level comment/review bodies via gh pr view/GraphQL, then passes the resulting JSON (including untrusted free text) into the agent tick where it is used as LLM context for resolve-pr-feedback/debug decisions.