Help us improve
Share bugs, ideas, or general feedback.
From claude-video-vision
Analyzes video files or YouTube URLs: extracts frames/audio, detects scenes/silence, transcribes, and provides summaries or answers questions.
npx claudepluginhub jordanrendric/claude-video-vision --plugin claude-video-visionHow this command is triggered — by the user, by Claude, or both
Slash command
/claude-video-vision:watch-video path/to/video.mp4 or YouTube URL [optional prompt or question about the video]The summary Claude sees in its command listing — used to decide when to auto-load this command
# Watch Video Parse the user's input to extract: 1. **Video path** — the file path (required) - YouTube URLs are also supported; pass the URL directly as `path` 2. **Prompt** — any question or instruction about the video (optional) 3. **Flags** — `--fps <number>`, `--resolution <number>` (optional) Then follow this workflow **in order — do NOT skip step 2**: 1. Call `video_info` on the path/URL to verify it's a valid video and get duration. For YouTube URLs, the MCP server downloads the video with `yt-dlp` and uses YouTube subtitles/auto-captions before falling back to the confi...
/watchDownloads, transcribes, and analyzes video content from URLs or local files. Extracts frames and captions to answer questions about the video.
/videoAnalyzes a YouTube video, local file, or directory of videos, extracting title, key points, timestamps, topics, and sentiment with optional local caching.
/async-analyzeStarts an asynchronous video analysis task from a URL, local file, asset ID, or base64 data. Also supports status, list, and delete subcommands.
/analyze-footageAnalyze existing video files using TwelveLabs AI. Understand what's in your footage — find specific scenes, detect objects and speakers, get timestamps for the best clips.
/claude-watchWatches a video via URL or path and produces study notes, optionally focused on a specified topic or question.
Share bugs, ideas, or general feedback.
Parse the user's input to extract:
path--fps <number>, --resolution <number> (optional)Then follow this workflow in order — do NOT skip step 2:
Call video_info on the path/URL to verify it's a valid video and get duration.
For YouTube URLs, the MCP server downloads the video with yt-dlp and uses
YouTube subtitles/auto-captions before falling back to the configured audio backend.
REQUIRED for videos > 30s: Call video_analyze BEFORE video_watch. This is NOT optional.
Use filters: scene_changes: true, silence: true, transcription: true at minimum.
Add other filters based on the user's question (motion, blur, exposure, loudness, etc.).
The analysis tells you WHERE to look — use it to plan smart frame extraction.
Call video_watch:
fps: "auto" without view_sample — full coverage to avoid missing brief moments.segments with variable FPS based on analysis data. Use view_sample to limit initial frames.If the user asks for more detail on a specific moment, use video_detail to drill in with higher FPS/resolution on a 3-5 second window. Use view_sample: 3 first, then request specific timestamps.
If the user provided a prompt/question, answer it based on the video content.
If no prompt was provided, give a comprehensive summary of what happens in the video.
If video_watch fails with a setup error, call video_setup first, then retry.