work-with-pr

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

• Third-party content exposure detected (medium risk: 0.30). In Phase 3, the workflow reads outsider-authored free text from the GitHub PR itself by running gh pr checks (and parsing/reading failed job logs) and by invoking the “review-work” skill that performs “context mining” and reviews the PR changes/diff on {BRANCH_NAME}.