code-reviewer
Pass
Audited by Gen Agent Trust Hub on Jun 28, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by processing untrusted code diffs and files (Ingestion Point: SKILL.md). In its 'Implementation' mode, the agent is explicitly directed to 'make the changes' to files (Capability: File-write), creating a risk that malicious instructions embedded as comments or code within a reviewed file could lead to unauthorized actions or data deletion. The skill lacks explicit boundary markers or sanitization logic to ensure the agent ignores commands hidden within the content it is reviewing.
Audit Metadata