2d-generalist

Pass

Audited by Gen Agent Trust Hub on Nov 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Several Python scripts execute local system commands to process media or verify the environment.
  • scripts/check_render.py and scripts/compare_look.py use ffmpeg and ffprobe to analyze video files and compare frames for visual consistency.
  • scripts/check_vox_toolchain.py executes commands like blender --version and node --version to audit the local installation of creative software.
  • [EXTERNAL_DOWNLOADS]: The scripts/fetch_images.py script retrieves media assets from external sources.
  • It fetches images from Openverse (api.openverse.org) and Wikimedia Commons (commons.wikimedia.org), which are well-known and legitimate repositories for creative assets.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external media and geospatial data (GeoJSON/KML), which represents a standard ingestion surface for AI agents handling multi-modal content.
  • The skill includes explicit warnings that downloaded images should be treated as data and never executed.
  • The processing of external data is limited to visual and spatial parameters within the context of motion graphics production.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 17, 2026, 02:18 AM