review-team
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides a purely instructional framework for code analysis and does not contain executable scripts, network requests, or access to sensitive credentials.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code targets (e.g., patches, PR diffs), creating a potential surface for indirect prompt injection. 1) Ingestion points: 'Review target' in SKILL.md. 2) Boundary markers: None specified. 3) Capability inventory: Instructional only; no direct tool access or network capabilities. 4) Sanitization: None. Mitigation: Managed by multi-agent consensus and evidence-based severity gates.
Audit Metadata