gemini-review

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text is ingested via GitHub PR review-thread comment bodies (e.g., bodyText/body from reviewThreads.nodes[].comments.nodes[] and pullRequest.comments.nodes[]) fetched at runtime by scripts/wait-pr-review.sh and scripts/pr-review-summary.sh using gh api graphql, and then included in the agent’s context/output for classification.