The Agent Skills Directory

[COMMAND_EXECUTION]: The unit test script scripts/test_lp_certificate.py utilizes subprocess.run to execute the check_lp_certificate.py utility. These calls are restricted to executing the skill's own local scripts using the current Python interpreter and hardcoded test data to verify CLI behavior.
[INDIRECT_PROMPT_INJECTION]: The skill provides a Python utility that processes user-supplied JSON data representing mathematical optimization problems, which presents a potential injection surface.
Ingestion points: The scripts/check_lp_certificate.py script reads data from files or standard input provided to the command-line interface.
Boundary markers: The script strictly enforces a JSON schema requiring specific keys (A, b, c, x, y) and rejects any unexpected or duplicate keys.
Capability inventory: The skill performs file reads (Path.read_text) and executes its own scripts via subprocess.run in a testing context. It does not perform network operations.
Sanitization: All input scalars are validated against a strict regular expression for decimal numbers and limited to 1,000 characters and a specific exponent magnitude. Data is converted to fractions.Fraction for exact arithmetic, preventing interpretation as code or instructions.

foundations-mathematical-optimization