review-loop
Pass
Audited by Gen Agent Trust Hub on Jul 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The workflow involves running existing project tests to verify code integrity after applying suggested fixes. The commands are identified from local configuration files like
CLAUDE.md. - [PROMPT_INJECTION]: The skill processes external AI review output, which represents a surface for indirect prompt injection. This is mitigated by mandatory user approval of all changes before application, maintaining a human-in-the-loop safety boundary.
Audit Metadata