review-agent-harness
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses subprocess.run to perform repository discovery and Git operations. These calls are executed using argument lists rather than shell environments, which prevents command injection.
- [DYNAMIC_EXECUTION]: The scripts/render_report.py file uses ctypes.CDLL(None) to access standard C library functions for atomic directory renames on Unix-like systems. This represents standard foreign function interface usage for filesystem reliability.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository files and session logs. It mitigates indirect prompt injection risks through an explicit privacy boundary, data sanitization, and strict validation of findings before persistence.
Audit Metadata