The Agent Skills Directory

[SAFE]: The skill consists entirely of educational documentation and code snippets for scientific computing using MATLAB and GNU Octave. All provided commands and patterns follow standard industry practices for these environments, such as using headless flags (-batch, --no-gui) for automation.
[INDIRECT_PROMPT_INJECTION]: The skill provides instructions for reading external files using functions like readtable and imread. While processing untrusted external data always presents an indirect prompt injection surface, the documentation does not include any patterns that actively exploit this surface or encourage unsafe handling of external inputs.
Ingestion points: Data import functions documented in references/data-import-export.md (e.g., readtable, readmatrix, load).
Boundary markers: The documentation does not provide specific boundary markers for untrusted data, which is typical for a general-purpose programming reference.
Capability inventory: The skill documents capabilities including file system writes, data visualization, and external process execution (via MATLAB's ! operator and Python integration).
Sanitization: Standard for a language reference, the material assumes the developer manages data sanitization logic within their scripts.

matlab