An automated utility for downloading videos, generating high-accuracy transcripts with Whisper, and extracting frames for visual analysis.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install xeonen-video-analyzer
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install xeonen-video-analyzer using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Video Watcher?
Video Watcher is a sophisticated media processing tool designed to bridge the gap between video content and textual analysis. By integrating industry-leading utilities such as yt-dlp, ffmpeg, and OpenAI Whisper, it allows developers to programmatically ingest video data from various platforms and convert it into a structured format. This skill is a vital component for anyone building automated research pipelines or documentation engines within the Openclaw Skills ecosystem, providing a reliable way to handle high-volume video processing tasks.
The core strength of Video Watcher lies in its multi-modal approach. It doesn't just download a file; it deconstructs the media into its constituent parts: raw video, isolated audio, timestamped subtitles, and periodic visual snapshots. This granular output makes it significantly easier for AI agents or human researchers to parse long-form content, identify key moments, and generate summaries without manually watching hours of footage.
Video Watcher Use Cases
- Performing due diligence (DD) on company presentations and technical product demos.
- Generating comprehensive lecture notes and study materials from educational videos.
- Transforming podcast recordings into searchable text archives and summaries.
- Creating technical documentation for software tutorials by capturing both steps and screenshots.
- Processing internal meeting recordings for automated action-item extraction.
How Video Watcher Works
- The user triggers the workflow by providing a video URL to the analysis script.
- yt-dlp fetches the media from the source URL and saves it to the local output directory.
- ffmpeg extracts the audio track from the video file to ensure optimal transcription quality.
- OpenAI Whisper processes the extracted audio to generate both plain text transcripts and SRT subtitle files.
- Simultaneously, ffmpeg captures frame screenshots at a defined interval (e.g., every 30 seconds) to provide visual context.
- The resulting assets are organized into a standardized folder structure, ready for further AI-assisted summarization or manual review.
Video Watcher Setup
To utilize this entry in the Openclaw Skills collection, you must first install the required binary dependencies on your system:
brew install yt-dlp ffmpeg openai-whisper
After installation, you can initiate a video analysis by running the provided shell script:
./scripts/analyze.sh "https://youtube.com/watch?v=example"
Video Watcher Data Schema & Taxonomy
All processed data is stored in the outputs/ directory with a standardized taxonomy for easy retrieval:
| Asset | File Name | Description |
|---|---|---|
| Video | video.mp4 |
The raw downloaded media file. |
| Audio | audio.mp3 |
The extracted audio stream used for processing. |
| Text | transcript.txt |
A plain text version of the speech-to-text output. |
| Subtitles | transcript.srt |
Time-coded subtitles for sync-heavy workflows. |
| Visuals | frames/ |
A sub-directory containing periodic JPG screenshots. |
Video Watcher Advanced Features
- Support for multiple Whisper model sizes (tiny, base, medium, large) to optimize for speed or accuracy.
- Adjustable frame-capture intervals via configuration to match the pace of different video types.
- Direct integration with AI agents for immediate summarization using the
clawdbot askcommand. - Flexible output directory management to support batch processing and organizational workflows.
- Compatibility with a wide range of video hosting platforms supported by the yt-dlp backend.
SKILL.md
Loading
Related Openclaw Skills
Arena System
The Arena System implements an adversarial self-improvement framework where AI agents engage in debate loops to minimize hallucinations and validate logic.
Video Watcher
A comprehensive automation tool for downloading videos, generating AI-powered transcripts, and capturing periodic screenshots for deep content analysis.
Video Watcher
Download videos, transcribe audio with Whisper, and extract frames for deep content analysis.
Arena System
An adversarial self-improvement framework that reduces AI hallucinations through agent versus anti-agent debate loops.
AIEOS (AI Entity Object Specification)
AIEOS is a standardization framework that provides a consistent data structure for defining how AI agents speak, react, and remember across different ecosystems.
claudemem
claudemem provides a persistent, local memory layer for AI agents to remember technical decisions, API specifications, and session context across multiple conversations.
Featured*
