renderer
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill recommends executing Blender in headless mode (
blender -b --python script.py) to generate assets from scripts stored in the repository. - [DYNAMIC_EXECUTION]: The architecture involves running Python scripts via the Blender CLI, which constitutes dynamic execution of local scripts. If the agent generates these scripts based on untrusted user input, it could lead to arbitrary code execution within the Blender environment.
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to install the
mathpackage from npm while providing a link to thepmndrs/mathGitHub repository. Themathpackage on npm is an unrelated, deprecated package; the correct package for the provided repository is@pmndrs/math. This naming discrepancy poses a dependency confusion risk. - [INDIRECT_PROMPT_INJECTION]: The renderer is designed to process external inputs such as "game observation, a query result, a document". This creates a surface where malicious instructions embedded in these external data sources could attempt to influence the agent's behavior during the rendering process or asset generation.
Audit Metadata