do-adversarial-review
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists of markdown instructions for code review and contains no executable code, shell commands, or scripts.
- [SAFE]: No network operations, external downloads, or third-party dependencies are identified.
- [SAFE]: The skill enforces a read-only policy, directing the agent to propose fixes without modifying files or accessing data outside the defined scope.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input in the form of code diffs. 1. Ingestion points: Code diffs and source files. 2. Boundary markers: The instructions do not specify delimiters to separate untrusted code from the agent instructions. 3. Capability inventory: The skill is restricted to read-only analysis and lacks permissions for command execution, file writes, or network access. 4. Sanitization: No sanitization is present, but the lack of active capabilities mitigates the impact of potential injection.
Audit Metadata