Model Context Protocol Server

What is Model Context Protocol Server?

A containerized Model Context Protocol (MCP) server that enables communication between LLMs and tools through a standardized protocol. It provides both REST API and WebSocket interfaces for tool execution and context management, targeting developers building LLM-powered applications.

How to use Model Context Protocol Server?

Deploy the server using Docker Compose: clone the repository, create tools, context, and data directories, configure a .env file, then run docker-compose up --build. Tools are placed as Python scripts in the tools directory, and the server exposes REST endpoints on port 3000 and a WebSocket at ws://localhost:3000 for real‑time communication.

Key features of Model Context Protocol Server

• Dual runtime environment (Node.js + Python)
• WebSocket support for real‑time updates
• REST API endpoints for tool management and execution
• Context persistence and management via JSON files
• Tool execution isolation inside the container
• Easy deployment with Docker

Use cases of Model Context Protocol Server

• Allow LLMs to invoke Python‑based tools through a standardized API
• Maintain conversation or session context across multiple interactions
• Enable real‑time tool execution and monitoring via WebSocket
• Provide a consistent, containerized interface for integrating tools with any MCP‑compatible LLM

FAQ from Model Context Protocol Server

What are the prerequisites to run Model Context Protocol Server?

You need Docker and Docker Compose installed on your system.

How do I add a new tool to the server?

Place a Python script in the tools directory. The script should accept JSON parameters via sys.argv[1] and return JSON output printed to stdout.

How is context managed in Model Context Protocol Server?

Context files are stored as JSON in the context directory, each with a unique ID. They can be retrieved via the GET /api/context/:id REST endpoint or through WebSocket messages.

How do I stop the server?

Run docker-compose down in the project directory.

What security measures are implemented?

Tool execution is isolated within the container, input validation is performed on all API endpoints, and context access can be restricted based on implementation needs.