audit-merges

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

• Third-party content exposure detected (medium risk: 0.30). At runtime, the skill enumerates and deep-reads merged PRs by using GitHub GraphQL search to fetch PullRequest nodes including reviewThreads and (after identifying PRs) pulls the PR details/diffs for the landed mergeCommit, so outsider-authored free text in PR bodies, review comments, and thread text is ingested.