YouTube Tools

You can run the YouTube Tools MCP Server to enable AI assistants to search YouTube, fetch transcripts, and store video summaries locally. It’s designed for seamless integration with MCP-compatible clients, giving you fast access to searchable video data and accurate transcripts for analysis, research, or accessibility workflows.

How to use

This MCP server exposes capabilities to search YouTube videos, extract transcripts, and store or retrieve video summaries from a local database. You use an MCP-compatible client to trigger searches, request transcripts in different languages, and manage stored summaries. The server is configured locally and accessed through the MCP protocol, so you don’t interact with it through a traditional web interface. Instead, you run the server locally and point your client to it as an MCP endpoint.

How to install

Prerequisites: you need Node.js 18 or newer and a package manager (pnpm is recommended, but npm works too). You also need access to the YouTube Data API v3 key for search functionality and a RapidAPI subscription for transcripts.

Step-by-step commands you will run:

# 1. Clone the project and install dependencies
 git clone <repository-url>
 cd youtube-tools-mcp
 pnpm install

# 2. Set up API keys in a .env file (create this file in the project root)
 # YouTube Data API v3 key for search
 echo 'YOUTUBE_API_KEY=your_youtube_api_key_here' > .env
 # RapidAPI key for transcripts
 echo 'RAPIDAPI_KEY=your_rapidapi_key_here' >> .env
 # Optional: custom database path for video summaries
 echo 'DATABASE_PATH=./video_summaries.db' >> .env

# 3. Build the project
 pnpm run build

# 4. Test transcript extraction (optional local test script)
 node temp/test-rapidapi.js

# 5. Start the MCP server (see stdio config below for exact command)
 # The final runtime command is the stdio entry you use to launch the server

MCP setup for Claude Desktop (stdio configuration)

The MCP server runs as a local process started by your MCP client. You configure Claude Desktop to launch the server using the following command and environment variables. This starts the Node.js process that runs the built MCP server from the compiled index file.

{
  "mcpServers": {
    "youtube_tools": {
      "command": "node",
      "args": ["/absolute/path/to/youtube-tools-mcp/build/index.js"],
      "env": {
        "YOUTUBE_API_KEY": "your_youtube_api_key_here",
        "RAPIDAPI_KEY": "your_rapidapi_key_here",
        "DATABASE_PATH": "/absolute/path/to/youtube-tools-mcp/video_summaries.db"
      }
    }
  }
}

Build and run commands you’ll use

Build and run flow mirrors the steps above. After building, you start the MCP server via the runtime command shown in the stdio configuration to keep the server running locally. You then interact with the server through your MCP client by issuing requests to search YouTube, fetch transcripts, or store/retrieve video summaries.

Additional setup tips

If you plan to use transcripts in multiple languages, ensure the transcript API supports the target languages and that you specify the language code when requesting transcripts. Store and later fetch video summaries from the local database for quick access and offline usage. If you encounter issues with transcripts, check that the video is public and that transcripts exist for the requested language.

Troubleshooting

Common issues include missing transcripts, private or restricted videos, and invalid video IDs. The MCP server returns actionable error messages to guide you toward a resolution, such as trying a different video, specifying a different language, or confirming the video URL format.

Notes on data handling and security

Transcripts and video summaries are accessed via local storage. Protect your environment by keeping API keys secure and limiting access to the local server. If you share configurations, remove sensitive values or replace them with placeholders.

Available tools

search_youtube

Search for YouTube videos using the official YouTube Data API v3 with optional filters such as date, duration, and quality.

get_youtube_transcript

Extract complete transcripts from YouTube videos using RapidAPI, with support for specifying language codes.

storeVideoSummary

Store or update a video summary in the local libsql database for quick retrieval later.

fetchExistingVideoSummary

Fetch an existing video summary from the local database by video ID.