youtube-transcript

Fail

Audited by Gen Agent Trust Hub on Aug 23, 2026

Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses sudo apt install to acquire administrative privileges for installing software on Linux systems. It also executes local shell commands for video processing, transcription (yt-dlp, whisper), and file management.
  • [EXTERNAL_DOWNLOADS]: The skill installs external dependencies from well-known package managers including Homebrew, Apt, and PyPI. It also directs users to the official GitHub repository for manual installation instructions.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from external YouTube sources.
    • Ingestion points: Fetches video titles and transcript content from external URLs provided by the user (SKILL.md).
    • Boundary markers: No explicit delimiters or instructions are used to prevent the agent from following directions embedded in the transcripts.
    • Capability inventory: The skill possesses Bash and Write capabilities, allowing it to execute shell commands and modify the filesystem based on external data.
    • Sanitization: Employs basic sanitization for filenames (removing slashes and special characters) and deduplication for transcript text, but does not fully sanitize the natural language content.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Aug 23, 2026, 04:00 AM
Security Audit — agent-trust-hub — youtube-transcript