youtube-content
Pass
Audited by Gen Agent Trust Hub on Aug 28, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses the
uvpackage manager to installyoutube-transcript-apifrom the official Python Package Index (PyPI). This is a well-known and standard library for accessing YouTube transcript data. - [COMMAND_EXECUTION]: The skill instructions define a workflow where the agent executes a local Python script (
scripts/fetch_transcript.py) usinguv runto fetch data. This is standard behavior for a tool-assisted skill. - [INDIRECT_PROMPT_INJECTION]: As the skill is designed to process external transcript content, it inherently possesses an attack surface for indirect prompt injection where a video transcript could contain instructions for the AI.
- Ingestion points: Video transcripts are fetched from YouTube via the
fetch_transcript.pyscript. - Boundary markers: The instructions do not specify explicit delimiters or guardrails to separate transcript content from system instructions.
- Capability inventory: The skill's operations are limited to text processing and execution of a specific local helper script; it does not include general-purpose network or file-writing tools.
- Sanitization: Transcript text is processed as retrieved from the API without additional sanitization layers.
Audit Metadata