x-youtube-analyzer

Fail

Audited by Gen Agent Trust Hub on Feb 21, 2026

Risk Level: HIGHCOMMAND_EXECUTIONCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION] (HIGH): The skill explicitly instructs agents to use required_permissions: ["all"] to bypass sandbox restrictions, granting unrestricted network and system access.\n- [CREDENTIALS_UNSAFE] (HIGH): Accesses sensitive credentials stored in ~/.config/google/profiles.json. Setup instructions encourage users to store plaintext API keys in a predictable location accessible by the agent.\n- [COMMAND_EXECUTION] (MEDIUM): Executes scripts/youtube_analyzer.py via poetry run. The script's source is not provided, making its file and network operations unverifiable.\n- [EXTERNAL_DOWNLOADS] (LOW): Uses yt-dlp to fetch content from YouTube, introducing untrusted external data into the environment.\n- [PROMPT_INJECTION] (LOW): Vulnerable to indirect prompt injection from YouTube video metadata or content, which lacks explicit sanitization or boundary markers in the ingestion process.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 21, 2026, 04:13 AM