Skills
Navigate
skills/mengto/skills/url-to-motion-design/Gen Agent Trust Hub

url-to-motion-design

Pass

Audited by Gen Agent Trust Hub on Nov 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources (X posts, YouTube metadata) via scripts/fetch-source.mjs which can influence subsequent agent actions.
  • Ingestion points: External metadata and text are stored in <run>/source/source.json and read by the agent.
  • Boundary markers: No explicit delimiters or instructions are used to separate ingested content from system instructions.
  • Capability inventory: High-privilege actions including shell execution (ffmpeg, magick), local file writes, and canvas manipulation via MCP tools.
  • Sanitization: There is no sanitization of text retrieved from external URLs before processing.- [DYNAMIC_EXECUTION]: The rendering server dynamically loads code from paths specified at runtime.
  • Evidence: lab/server.mjs uses await import to load several JavaScript modules (sceneRuntime.js, scene.js, etc.) from a computed repository path (REPO).- [COMMAND_EXECUTION]: The skill relies on several local CLI tools for media processing.
  • Evidence: Subprocess calls are made to ffmpeg, ffprobe, magick (ImageMagick), and yt-dlp across multiple scripts.- [EXTERNAL_DOWNLOADS]: The skill fetches video files and metadata from the internet.
  • Evidence: scripts/fetch-source.mjs downloads media from api.fxtwitter.com and other user-provided URLs using fetch and yt-dlp.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 5, 2026, 08:57 PM