gemini-yt-video-transcript
Pass
Audited by Gen Agent Trust Hub on Mar 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates the execution of a Python script (
scripts/youtube_transcript.py) which manages the transcription process by interfacing with external APIs and the local filesystem.\n- [EXTERNAL_DOWNLOADS]: The skill interacts with well-known and trusted external services to perform its primary function:\n - It retrieves video titles and metadata from YouTube's oEmbed API (
https://www.youtube.com/oembed).\n - It sends video data and prompts to the Google Gemini API (
https://generativelanguage.googleapis.com) for processing.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from YouTube videos and metadata, which represents a potential attack surface.\n - Ingestion points: Video titles and content are fetched from YouTube and passed to the LLM within
scripts/youtube_transcript.py.\n - Boundary markers: The prompt used for transcription includes structural constraints to guide the output format, although it does not use complex delimiters for the URL input.\n
- Capability inventory: The skill has the capability to write files to the local workspace (
out/directory) and perform network requests to Google and YouTube.\n - Sanitization: The script implements a
safe_slugfunction that sanitizes strings before using them as filenames, preventing path traversal attacks or the creation of malicious file names from untrusted video titles.
Audit Metadata