matlab

Pass

Audited by Gen Agent Trust Hub on Jun 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides detailed instructions and patterns for executing MATLAB and GNU Octave code via command-line interfaces such as matlab -batch and octave. These are documented as standard practices for automation and CI/CD workflows.
  • [EXTERNAL_DOWNLOADS]: The documentation includes instructions for installing GNU Octave using well-known package managers such as Homebrew (brew install octave) and APT (sudo apt install octave), and provides links to official documentation for MathWorks and GNU Octave.
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it is designed to ingest and process data from various external file formats (CSV, Excel, MAT, Text) which could contain malicious instructions.
  • Ingestion points: Multiple file reading functions are documented in references/data-import-export.md (e.g., readtable, readmatrix, readcell, readlines, fileread, load).
  • Boundary markers: No explicit instructions for delimiters or 'ignore embedded instructions' warnings are provided for data ingestion.
  • Capability inventory: The skill allows shell execution via CLI patterns, dynamic Python code execution (pyrun), network access through Python integration (e.g., requests library), and file-write operations.
  • Sanitization: There are no documented sanitization or validation steps for content loaded from external files.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 14, 2026, 09:23 AM
Security Audit — agent-trust-hub — matlab