youtube-research

Pass

Audited by Gen Agent Trust Hub on May 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Executes the yt-dlp utility via a Python subprocess to download video records from YouTube URLs.
  • [COMMAND_EXECUTION]: Uses the ffmpeg utility for media generation and Node.js for running internal workspace management scripts.
  • [EXTERNAL_DOWNLOADS]: Downloads YouTube metadata, comments, and video files through the PostPlus Cloud collection service and external URLs.
  • [PROMPT_INJECTION]: Processes untrusted data in the form of YouTube comments, creating an indirect prompt injection surface when this data is displayed to the user or agent.
  • Ingestion points: Audience comments gathered from YouTube videos via the youtube-comments collection key.
  • Boundary markers: Explicit instructions are provided to the agent to clarify the nature of public comment data to the user.
  • Capability inventory: Subprocess execution for media tools, file system writes for asset management, and network communication via Node.js HTTP modules.
  • Sanitization: The local workspace dashboard implements HTML entity escaping for data rendered in the UI.
Audit Metadata
Risk Level
SAFE
Analyzed
May 17, 2026, 03:47 AM
Security Audit — agent-trust-hub — youtube-research