Youtube

The YouTube MCP Server exposes a set of tools to search YouTube videos, fetch transcripts, and perform semantic search over stored transcripts. It runs via the uv package manager and expects a Python-based server.py script. Before you start, ensure you have Python 3.8+ and a Google API key for embeddings, then install and run the server as described below. Once running, you can connect any MCP-compatible client (such as Claude) and discover the available tools automatically for use in prompts and workflows.

Available tools include: 1) search-youtube to find YouTube videos by a query, returning a list of results with basic metadata; 2) get-transcript to retrieve the transcript of a given video URL; 3) store-video-info to save video metadata and transcripts into a vector database for later semantic search; 4) search-transcripts to perform semantic search over stored transcripts and return relevant snippets or matches. These tools enable iterative, context-aware retrieval and analysis of video content without relying on the official YouTube API for search.

To use these tools from a client, start the server using the uv-based command, then connect your MCP client. The client will discover the tools automatically and you can invoke them with natural-language prompts such as “Search for Python tutorial videos” or “Get the transcript of this video: [video_url]”.

Prerequisites:

• Python 3.8+
• uv package manager
• A Google API key for embeddings

Installation steps:

1. Clone the repository:

git clone <repository-url>
cd <repository-directory>

2. Create and activate a virtual environment using uv: On Windows:

uv venv
.\venv\Scripts\activate

On Unix/MacOS:

source .venv/bin/activate

3. Install dependencies with uv:

uv pip install -r requirements.txt

4. Create a .env file with your Google API key:

GOOGLE_API_KEY=your_api_key_here

5. Run the server (direct method) to verify it starts:

uv run server.py

6. If you prefer Claude.app configuration, add the following to your MCP config (as described in the README) and ensure your environment variable GOOGLE_API_KEY is set:

"mcpServers": {
  "youtube": {
    "command": "uv",
    "args": ["run", "server.py"],
    "env": {
      "GOOGLE_API_KEY": "your_api_key_here"
    }
  }
}

Notes and tips:

• Ensure GOOGLE_API_KEY is correctly set in the environment where the server runs; the tools rely on embeddings and may fail if the key is missing.
• The vector database used by store-video-info should be available and writable; check any configuration for storage backend if you customize the setup.
• When using Claude.app or other clients, tools are discovered automatically once the server is running and reachable.
• If you encounter connectivity or library errors, verify that Python 3.8+ is in use and that dependencies listed in requirements.txt are installed successfully with uv.
• For Windows paths in Claude.config, you can point directly to the Python executable and server.py as shown in the README example.