and-pick

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

• Third-party content exposure detected (high risk: 0.85). At runtime the workflow “Read full evidence for candidates” explicitly says to read “material comments or receipts” and “linked issues or PRs” from the backend/source-of-truth state (i.e., likely outsider-authored issue/PR/wiki/comment text) into the agent’s LLM context while evaluating candidates.