review-pending-pr-reviews
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes PR review drafts that ingest content from external pull requests. This attack surface is mitigated by mandatory human review ('grill the findings') and explicit confirmation requirements before any data is transmitted to GitHub. The instructions specify extracting content using clear structural boundaries (e.g., from '## Review draft' to '## Prior reviews'), reducing the risk of accidental instruction following.\n- [COMMAND_EXECUTION]: Employs 'gh', 'git', and 'ls' for repository and file management. It uses 'cat' with single-quoted heredocs ('REVIEW_EOF') to prepare review bodies, which effectively prevents shell expansion and command injection. Operations are constrained by 'allowed-tools' to specific, safe directory paths (~/.nx-pr-reviews) and temporary files.
Audit Metadata