The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands dynamically generated by the math_router.py script. This creates a risk where the agent may execute arbitrary code or system commands if the router returns a malicious string.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by passing untrusted user-provided math requests directly into a command-line routing workflow.
Ingestion points: User input is ingested via the command-line argument to scripts/cc_math/math_router.py as shown in SKILL.md.
Boundary markers: No boundary markers or instructions to ignore embedded commands are present in the skill definition.
Capability inventory: The skill leverages uv run python to execute local scripts like math_router.py and sympy_compute.py with full shell access.
Sanitization: No sanitization or validation of the user input is described in the skill documentation, making the security of the operation dependent on the opaque implementation of the router script.

math-router