matlab-generate-code

Pass

Audited by Gen Agent Trust Hub on Aug 18, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill processes user-provided MATLAB source code for analysis and compilation, creating an indirect prompt injection surface.
  • Ingestion points: User-authored MATLAB functions and classes are ingested for screening and code generation (SKILL.md, references/generate-code.md).
  • Boundary markers: No specific boundary markers or "ignore instructions" warnings are utilized when processing user source code.
  • Capability inventory: The skill utilizes evaluate_matlab_code to invoke codegen, coder.screener, and performance profiling tools (references/generate-code.md, references/accelerate-mex.md).
  • Sanitization: The skill explicitly instructs users to run coder.screener programmatically to identify and fix unsupported calls or potentially dangerous language constructs before generation.
  • [COMMAND_EXECUTION]: The skill facilitates the execution of system compilers and build tools via the codegen command to produce C/C++ or CUDA binaries (references/generate-code.md). It also documents the use of coder.ceval to call external C/C++ functions directly from the MATLAB environment (references/write-c-interop.md). These capabilities are essential for the skill's primary purpose of code generation.
  • [EXTERNAL_DOWNLOADS]: The skill references a license document hosted on a domain belonging to the authoring organization (SKILL.md, manifest.yaml). This reference is informative and does not involve the execution of untrusted remote code.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 18, 2026, 05:10 PM
Security Audit — agent-trust-hub — matlab-generate-code