The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied MATLAB code and requirements to generate test cases, creating a surface for indirect prompt injection.
Ingestion points: The 'Workflow' section in SKILL.md explicitly instructs the agent to gather requirements and code from the user to test.
Boundary markers: No specific delimiters or instructions to ignore embedded commands within the processed code are provided in the instructions.
Capability inventory: The skill utilizes the run_matlab_test_file and evaluate_matlab_code tools (defined in manifest.yaml) to execute the generated MATLAB code.
Sanitization: No explicit sanitization or validation of the user-provided code or requirements is mentioned before they are incorporated into the test generation process.
[DYNAMIC_EXECUTION]: The skill generates MATLAB source files (.m files) based on user requests and then executes them to verify test success.
Evidence: The 'Workflow' and 'Conventions' sections in SKILL.md instruct the agent to write test files and then 'Run via run_matlab_test_file MCP tool to confirm tests pass.'

matlab-write-tests