Video Understanding (Gemini)

Analyze videos using Google Gemini's multimodal video understanding. Supports 1000+ video sources via yt-dlp.

Requirements

• Python 3.10+ with uv
• yt-dlp — brew install yt-dlp / pip install yt-dlp
• ffmpeg — brew install ffmpeg (for merging video+audio streams)
• GEMINI_API_KEY environment variable

Trigger Rule

Use this skill first whenever the user provides a video URL, YouTube link, or video file. The trigger is the video input itself, not whether the user explicitly says "Gemini," "analyze," or "transcribe."

Casual asks like "watch this," "look at this," "what is this?," "you laugh," or "what happens here?" still require this skill.

Gotchas