quick-fea

Warn

Audited by Gen Agent Trust Hub on Nov 7, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [DYNAMIC_EXECUTION]: The script scripts/fea.py uses the eval() function to evaluate mathematical expressions passed via the --fix and --load command-line arguments. While intended for coordinate-based facet selection (e.g., x[0] < 1e-3), this allows for the execution of arbitrary Python code because the environment is not restricted and includes the numpy module. An attacker who can influence the arguments passed to the skill can achieve arbitrary code execution.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external STEP files and user-defined selection logic, which creates an attack surface for indirect prompt injection.
  • Ingestion points: User-provided STEP files and the strings used for the --fix and --load arguments in scripts/fea.py.
  • Boundary markers: None; the strings are evaluated directly by the Python interpreter.
  • Capability inventory: The script performs file system writes (generating JSON results and PNG plots) and utilizes complex libraries for meshing and numerical analysis.
  • Sanitization: No sanitization or sandboxing is applied to the expressions before they are passed to eval().
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user or agent to install several Python packages including gmsh, scikit-fem, and matplotlib, which are standard tools for scientific computing and finite element analysis.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 7, 2026, 06:02 AM