Video Analyzer MCP Server
Transcripts, frames, OCR, and metadata extraction for YouTube, social media, and local files.
Data last scanned today · Reviewed today
Overview
Video Analyzer MCP Server processes multimedia content by generating transcripts, extracting key frames, and gathering video metadata. It supports diverse sources including YouTube, Instagram, TikTok, Loom, X, Vimeo, and direct file paths. By leveraging tools like ffmpeg-static, it converts raw video input into structured data that AI agents can query and analyze. Designed for flexibility, the server functions as a standalone MCP component or as the engine for more complex knowledge management pipelines. It allows users to perform per-video analysis or build expansive searchable databases from archived social media accounts.
Our verdict
With a security score of 38/100, this tool is categorized as high risk and should be used with caution in sensitive environments. The assessment is primarily due to the server operating without a read-only mode, meaning every installation is granted full write access, and the lack of an editorially reviewed authentication method. It is best suited for local, isolated development environments where users can manage their own video assets and platform integrations.
- Supports a wide range of platforms via yt-dlp integration
- Provides both transcripts and visual frame data
- Minimal supply chain with only seven direct dependencies
- Released under the permissive MIT license
- Lacks a read-only mode for restricted file access
- Community-maintained project without official vendor backing
- Authentication mechanisms have not undergone professional review
Setup
{
"mcpServers": {
"mcp-video-analyzer": {
"args": [
"-y",
"mcp-video-analyzer"
],
"command": "npx"
}
}
}Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.
- No read-only mode — running this grants full read/write access.
- No built-in authentication — anyone with access to this config can invoke every tool it exposes.
Tools
| Tool | Description | Risk |
|---|---|---|
| analyze_video | Full analysis: transcript + key frames + OCR + timeline + metadata | low |
| analyze_videos | Batch version, one structured result per source (resumable) | low |
| get_transcript | Transcript only (native captions or Whisper fallback) | low |
| get_metadata | Metadata + comments + chapters, no download | low |
| get_frames | Key frames only (scene-change or dense 1 fps) | low |
| analyze_moment | Deep-dive on a time range (burst frames + transcript + OCR) | low |
| get_frame_at | Single frame at a timestamp | low |
| get_frame_burst | N frames across a narrow window (motion/animation) | low |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›What software do I need installed on my machine to use this server?
Node.js 18+ is required to run the server via npx. While not strictly required for all functions, installing yt-dlp is highly recommended to support URLs from platforms like YouTube, Instagram, and TikTok.
›Can I use the video analyzer without an MCP-compliant client?
Yes, you can use the CLI directly in any terminal by running the npx command, or use the tool as an Agent Skill in environments like Cursor or Copilot.
›What happens if I do not have yt-dlp or Chrome installed?
Local files and direct video URLs will still be processed for frames using the bundled ffmpeg-static. However, platform-specific URLs like YouTube will result in a warning, and transcript, metadata, and comment extraction will remain functional.
›How can I reduce the processing time or data extraction depth?
You can use the --detail brief flag to skip frame extraction or use the --fields flag to limit the JSON output to specific data points like metadata or transcripts.
Badge
Maintain this server? Add the live badge to your README.