The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill documents advanced SymPy features including autowrap, ufuncify, and codegen. These tools are designed to generate source code in C or Fortran and compile it at runtime to optimize mathematical calculations, which provides a high degree of dynamic execution capability.
[DYNAMIC_EXECUTION]: A pattern in references/code-generation-printing.md demonstrates the use of pickle.load for restoring symbolic expressions from local storage. Deserialization of data using the pickle module is inherently risky if the source files are not verified.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data through mathematical parsers like parse_expr, parse_latex, and parse_mathematica.
Ingestion points: The 'Interactive Computation' section in references/code-generation-printing.md shows a pattern for taking user input via the input() function and passing it directly to parse_expr.
Boundary markers: The provided code examples do not implement delimiters or 'ignore' instructions to separate the mathematical expression from potential injected text.
Capability inventory: The skill possesses extensive system capabilities, including the ability to write to the file system (e.g., open('output.py', 'w')) and generate executable Python, C, and Fortran code.
Sanitization: The skill documentation includes a specific textual warning stating that user input must be validated and sanitized to avoid code injection, though the simple code snippets provided for demonstration do not implement these checks.

sympy