The Agent Skills Directory

[SAFE]: The skill implements numerical stability checks using local Python scripts that utilize standard mathematical libraries and avoid unsafe operations such like eval() or exec().
[SAFE]: All scripts include comprehensive input validation to ensure physical parameters like grid spacing and time steps are finite and positive, while spatial dimensions are restricted to valid physical ranges (1, 2, or 3).
[SAFE]: Data ingestion for matrix files is secured by a 500 MB file size limit and the explicit disabling of pickle loading in NumPy (allow_pickle=False), which prevents potential arbitrary code execution from malicious data files.
[SAFE]: Resource exhaustion is prevented through hard limits on computational complexity, including a 100,000 per-side limit on matrix dimensions and a 10,000-entry cap for eigenvalues and stencil coefficients.
[SAFE]: The skill contains no network operations, hardcoded credentials, sensitive file access, or persistence mechanisms.

numerical-stability