matlab-document-toolbox
Pass
Audited by Gen Agent Trust Hub on Aug 2, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill implements an automated workflow that analyzes local project files (.m files) to extract function metadata, signatures, and H1 lines. This creates an indirect prompt injection surface where untrusted data from the analyzed codebase could be interpolated into the documentation plan or final artifacts, potentially influencing agent behavior or generating misleading content.
- [COMMAND_EXECUTION]: The skill instructions involve the automated execution of MATLAB code for validation and publishing.
- In Step 6 and Step 10, the agent is directed to run generated MATLAB scripts (e.g., GettingStarted.m and example scripts) to ensure they execute without error.
- The documentation publishing guidelines in 'references/examples-conventions.md' specify the use of the MATLAB 'publish' command with 'evalCode' set to true, which executes the example code to generate rich documentation output.
Audit Metadata