review
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions define a rigorous code review process that prioritizes identifying security vulnerabilities such as injection and secret exposure while emphasizing evidence-based findings.
- [PROMPT_INJECTION]: The skill instructions require the agent to ingest and follow guidance from project-local files (e.g.,
CLAUDE.md,REVIEW.md) and the source code being reviewed, which represents an indirect prompt injection surface. This is a standard architectural requirement for a code review tool and is considered safe in this context. - Ingestion points: Root project documentation (
CLAUDE.md,REVIEW.md) and the files under review as specified inSKILL.md. - Boundary markers: No explicit delimiters or instructions to ignore instructions within ingested data are defined.
- Capability inventory: The skill is focused on analytical text output and does not include tools for file modification, network access, or command execution.
- Sanitization: No explicit sanitization or escaping protocols are provided for the ingested content.
Audit Metadata