The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill executes code from a pre-compiled MATLAB P-file located at scripts/configure_for_codegen.p. This is a standard binary format for distributing MATLAB functions from this vendor and is treated as a safe vendor-owned resource.
[INDIRECT_PROMPT_INJECTION]: The skill processes Simulink models (.slx files) which are untrusted external data sources provided by the user.
Ingestion points: Reads model files and associated data dictionaries (.sldd) through the configure_for_codegen function called in SKILL.md.
Boundary markers: The skill instructions provide constraints for the agent's behavior but do not define specific delimiters or "ignore instructions" markers for the processed model data.
Capability inventory: The skill uses evaluate_matlab_code to execute MATLAB scripts and generates C/C++ code and reports to the filesystem.
Sanitization: No explicit sanitization or validation of model content is mentioned in the skill instructions.
[COMMAND_EXECUTION]: The skill interacts with the MATLAB environment via the evaluate_matlab_code tool. The agent is strictly instructed to use an atomic wrapper function, minimizing the risk of arbitrary command injection.

simulink-generate-embedded-code