The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data in the form of MATLAB source code and static analysis diagnostics from tools like check_matlab_code (as described in SKILL.md). This creates a surface for indirect prompt injection where malicious instructions could be embedded in the code being analyzed.
Ingestion points: MATLAB source code files and JSON/text diagnostics from the check_matlab_code MCP tool (referenced in SKILL.md Conventions).
Boundary markers: None identified in the instructions for separating code content from agent instructions.
Capability inventory: The agent is authorized to use evaluate_matlab_code and check_matlab_code (listed in manifest.yaml) and has the capability to write or modify file contents as part of the modernization process.
Sanitization: No specific sanitization or filtering of the processed code content is mentioned to prevent instruction leakage.
[DYNAMIC_EXECUTION]: The skill provides instructions on replacing eval with structured alternatives that still involve dynamic runtime execution. Specifically, in references/core-functions-guidance.md, it recommends using feval and str2func to execute functions based on string names, including scenarios where the name originates from user input. While safer than eval, these patterns still involve dynamic loading and execution from computed paths, which can be misused if the input is not strictly validated.

matlab-modernize-code