matlab-code-reviewer
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted MATLAB scripts and method plans from the workspace. While this presents an attack surface for indirect injection, the risk is mitigated by a rigid workflow, specific structural requirements, and a mandatory verification gate requiring five concrete pass items.
- [DATA_EXPOSURE]: File access is restricted to the local workspace directory for reading code and writing review reports. The skill does not perform network requests, hardcode credentials, or access sensitive system paths.
- [COMMAND_EXECUTION]: The skill performs static and structural analysis of code. It does not implement direct shell execution or dynamic command construction based on untrusted input.
- [DYNAMIC_EXECUTION]: The skill makes minimal fixes to MATLAB scripts involving surgical syntax or path corrections. It does not generate or execute arbitrary code at runtime or use unsafe deserialization.
Audit Metadata