The Agent Skills Directory

[OBFUSCATION]: The skill includes a pre-compiled MATLAB P-code file located at scripts/AgentProgressMonitor.p. P-code is a standard MATLAB format that provides executable logic in an obfuscated binary form to protect intellectual property while allowing the script to run natively within the environment.
[DYNAMIC_EXECUTION]: The workflow involves adding the skill's script directory to the MATLAB path and instantiating the AgentProgressMonitor class. This mechanism dynamically loads and executes the logic contained in the pre-compiled helper file to provide real-time status updates during project artifact indexing.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and display data derived from external Simulink models and project artifacts, which constitutes an indirect injection surface.
Ingestion points: Metric data, component names, and diagnostic messages are ingested via the getMetrics and getArtifactIssues methods in SKILL.md and scripts/AgentProgressMonitor.p.
Boundary markers: The instructions for presenting results to the user do not include specific delimiters or boundary markers for the interpolated model metadata.
Capability inventory: The skill has the capability to execute metric algorithms via execute and write interactive reports to the file system using generateReport.
Sanitization: The reference code for result inspection in SKILL.md outputs component names and values directly from the metric.Result objects using fprintf without explicit sanitization or filtering of the model-provided strings.

simulink-collect-metrics