Gemini Video Understanding

Documents

gemini-video-understanding

Gemini Video Understanding is an Agent Skill for documents published in the majiayu000/claude-skill-registry repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Analyze and understand videos using Google's Gemini API. Use when the user asks to analyze, understand, describe, summarize, transcribe, or extract information from videos. Supports local video files (MP4, MOV, WebM, etc.) and YouTube URLs. Can answer questions about video content, describe scenes, identify objects/people/actions, extract text/timestamps, and more. Use this skill when user provides a video file path or YouTube link and wants to understand its content.

View on GitHub ↗

QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

Open your project in Codex.
Copy and paste the prompt below.
Review files and risks before approving any action.

Prompt to paste

Inspect the Agent Skill at https://github.com/majiayu000/claude-skill-registry/blob/HEAD/skills/marketing/gemini-video-understanding/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/gemini-video-understanding, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Gemini Video Understanding

uv run ~/.claude/skills/gemini-video-understanding/scripts/understand_video.py \
  --source "path/to/video.mp4" \
  --prompt "Describe what happens in this video" \
  [--fast] \
  [--fps 1] \
  [--start 0] \
  [--end 60]

Source Types

Local files: --source "video.mp4" — MP4, MOV, WebM, AVI, MKV, etc.
YouTube: --source "https://youtube.com/watch?v=..." — Any YouTube URL

Options

Option	Description
`--fast`	Use faster flash model instead of pro
`--fps`	Frame rate sampling (default: 1 fps). Use 2+ for fast action
`--start`	Start time in seconds for clipping
`--end`	End time in seconds for clipping

Example Prompts

"Summarize this video in 3 bullet points"
"What text appears on screen? Include timestamps"
"Describe the main actions and events with timestamps"
"Identify all people and what they're doing"

API Key

Uses GEMINI_API_KEY env var, or pass --api-key KEY.

Updating Models

Edit ~/.claude/skills/gemini-video-understanding/config.json:

{
  "default_model": "gemini-3-pro-preview",
  "fast_model": "gemini-3-flash-preview"
}