code-review
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is subject to an indirect prompt injection surface due to its core function of processing untrusted data.
- Ingestion points: Untrusted source code files, pull request diffs, and repository changes processed during invocation as specified in
SKILL.md. - Boundary markers: Missing specific delimiters or isolation guidelines to separate instruction syntax from raw text/comments within reviewed files.
- Capability inventory: Code reading, repository searching, tree navigation, and running sub-commands like
/security-audit-staticand/performance-audit-static. - Sanitization: Absent; the engine parses source code directly without escaping or filtering comment blocks that could contain embedded instructions.
Audit Metadata