code-review-clone

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

• Third-party content exposure detected (medium risk: 0.30). In references/workflow.md step 3/4, the trainer runs scripts/collect-review-activity.py which downloads the authenticated reviewer’s GitHub PR review comment bodies from repos/{args.repo}/pulls/comments and then reads those comment chunk files into the model-building loop.