factory-babysit-pr
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from pull request feedback which could potentially influence agent behavior.
- Ingestion points: The skill reads a live snapshot of pull requests including 'review threads', 'current reviews', and 'unresolved threads' (SKILL.md).
- Boundary markers: The instructions include manual verification steps requiring confirmation of authority and ownership, explicitly stating that labels or green checks are not sufficient authorization (SKILL.md).
- Capability inventory: The agent is authorized to perform high-impact actions including modifying code ('Fix'), publishing to branches ('Publish'), and final repository actions ('Approve' and 'Merge') (SKILL.md).
- Sanitization: The instructions do not define specific sanitization or filtering mechanisms for data ingested from PR comments or review threads.
Audit Metadata