physics-solver
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFECREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill documentation includes a hardcoded local path '/Users/zhangmingda/clawd/.venv/bin/activate' which exposes a specific local system username ('zhangmingda').
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied physics problems without defining boundary markers or input validation protocols. This creates a surface where malicious instructions could be embedded in problem descriptions to influence agent behavior during code generation or calculation. 1. Ingestion points: User-provided physics problem descriptions (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: Execution of numerical integration and symbolic math via SciPy and SymPy (SKILL.md). 4. Sanitization: Absent.
Audit Metadata