The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes Simulink model structures and block parameters provided in the user's workspace. This creates an ingestion surface for untrusted model content. The risk is minimized by the skill's operational guardrails, which mandate a backup and require the user to explicitly approve each individual configuration change or block replacement.\n
Ingestion points: model_read and model_query_params calls in SKILL.md extract model metadata.\n
Boundary markers: The skill does not explicitly use specific delimiters for interpolated model data.\n
Capability inventory: The skill can modify models using model_edit and execute scripts via evaluate_matlab_code.\n
Sanitization: No specific sanitization of block parameters or configuration strings is performed before processing.\n- [COMMAND_EXECUTION]: The skill performs legitimate model lifecycle operations through evaluate_matlab_code, including backing up files with copyfile, cleaning up with delete, and configuring models via model_edit. All execution is directly related to the skill's stated purpose of model compatibility analysis.\n- [DYNAMIC_EXECUTION]: The skill uses two protected vendor scripts, raccel_deep_analysis.p and raccel_verify_build.p. These are P-code files provided by MathWorks for advanced model introspection. This is a standard distribution method for protected intellectual property within the MATLAB environment and is considered safe for vendor-provided content.

simulink-check-rapid-accelerator-compatibility