rethinking-scientific-modeling-physically
Pass
Audited by Gen Agent Trust Hub on Jun 27, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill processes user-supplied building specifications (dimensions, stories, seismic intensity) which are interpolated into generated Python code. This represents an indirect prompt injection surface. The skill mitigates this risk by requiring the agent to 'mentally trace' the code for syntax and physical accuracy (Step 9) and verify fundamental structural periods against empirical formulas.
- [EXTERNAL_DOWNLOADS]: Includes a reference link to a GitHub repository (github.com/Jovanqing/AutoBM) for the AutoBM framework methodology. This is provided for documentation and informational purposes and does not trigger automated downloads or remote code execution at runtime.
Audit Metadata