doc-generator
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- Indirect Prompt Injection (HIGH): The skill ingests untrusted docstrings and comments from Python files without sanitization. Evidence: 1. Ingestion points: Python source files (via 'parse Python files' instruction). 2. Boundary markers: Absent. 3. Capability inventory: Code execution via 'inspect' and file system access. 4. Sanitization: Absent.
- Remote Code Execution (HIGH): The instructions recommend using the 'inspect' module for runtime introspection. In Python, inspecting a module often requires importing it, which executes any top-level code. If the files being documented are untrusted, this allows for arbitrary code execution.
- Command Execution (MEDIUM): The skill performs file system operations (pathlib) and potentially executes code logic via introspection on user-controlled file paths.
Recommendations
- AI detected serious security threats
Audit Metadata