MCP Media Server

What is MCP Media Server?

MCP Media Server is a custom MCP server built with Python that enables AI assistants like Claude to interact with media files, perform video processing, and integrate with Supabase and Pinecone for data storage and vector search. It provides YouTube downloading, FFmpeg processing, and RESTful API endpoints for media workflows.

How to use MCP Media Server?

Install manually (Python 3.10+, FFmpeg, dependencies) or via Docker. Run with python main.py for stdio transport, python main.py --transport sse for SSE, or python main.py --api to enable the REST API. Optionally use --transport sse --api for both. For production, deploy with Docker Compose and configure a systemd service. Integrate with Claude Desktop by adding an MCP server entry in its settings JSON.

Key features of MCP Media Server

• YouTube video downloading via yt-dlp
• Video processing with FFmpeg
• Supabase integration for metadata storage
• Pinecone integration for vector search
• MCP server for AI assistant (Claude) integration
• RESTful API gateway for HTTP access
• Webhook, batch processing, and caching support
• User authentication and API key management

Use cases of MCP Media Server

• Download YouTube videos and process them for analysis
• Perform semantic video search using vector embeddings
• Integrate media processing into AI assistant workflows (e.g., Claude)
• Run automated media pipelines with scheduled tasks and webhooks
• Deploy a production-grade media server with Docker and monitoring

FAQ from MCP Media Server

What are the prerequisites to run MCP Media Server?

Python 3.10+, FFmpeg, a Supabase account, a Pinecone account, and an OpenAI API key (for embeddings). Docker and Docker Compose are needed for containerized deployment.

How do I start the MCP server?

Run python main.py from the project root. For SSE transport use python main.py --transport sse. To enable the API server add --api or combine both with --transport sse --api.

How can I integrate MCP Media Server with Claude Desktop?

In Claude Desktop's settings, add a new MCP server with the command pointing to your Python executable and arguments pointing to main.py. The server uses stdio transport by default.

What transport modes are available?

Two transport modes are available: stdio (default, for direct Claude Desktop integration) and SSE (Server-Sent Events, for remote HTTP-based connections).

Where are downloaded and processed files stored?

Files are stored locally in the downloads/, processed/, thumbnails/, and cache/ directories within the project folder. Logs go to logs/.