pr-resolution

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly fetches and parses untrusted, user-generated PR review content (e.g., via gh API calls in Phase 1 / references/discovery.md and bin/parse-coderabbit-review, and the shepherd's check-new-comments and parse-coderabbit-review steps) and the agent is instructed to read and act on those bot/human comments to decide fixes, launch agents, and perform commits/pushes—so third-party content can materially influence tool use and next actions.