Model Context Protocol (MCP)

Overview

MCP is a modular framework for managing, executing, and monitoring AI model contexts, including LLM prompts, Jupyter notebooks, and Python scripts. It provides a FastAPI backend and a Streamlit dashboard frontend.

Features

• Register and manage different types of MCPs (LLM prompts, notebooks, scripts)
• Execute MCPs and view results in a web UI
• Monitor server health and statistics
• Extensible for new MCP types

Setup

Prerequisites

• Python 3.9+
• (Recommended) Create and activate a virtual environment

Install dependencies

pip install -r requirements.txt

Environment Variables

• Set MCP_API_KEY for API authentication (optional, defaults provided)
• For LLMs, set ANTHROPIC_API_KEY if using Claude

Start the backend

uvicorn mcp.api.main:app --reload

Start the frontend

streamlit run mcp/ui/app.py

Usage

• Access the dashboard at http://localhost:8501
• Create, manage, and test MCPs from the UI
• Monitor health and stats from the sidebar

Adding New MCPs

• Implement a new MCP class in mcp/core/
• Register it in the backend
• Add UI support in mcp/ui/app.py

Running Tests

pytest

Project Structure

• mcp/api/ - FastAPI backend
• mcp/ui/ - Streamlit frontend
• mcp/core/ - Core MCP types and logic
• tests/ - Test suite

License

MIT

API Documentation

Once the server is running, you can access:

• API documentation: http://localhost:8000/docs
• Prometheus metrics: http://localhost:8000/metrics
• Health check: http://localhost:8000/health
• Statistics: http://localhost:8000/stats

Security

• API key authentication is required for all endpoints
• Rate limiting is enabled by default
• CORS is configured to allow only specific origins
• All sensitive configuration is managed through environment variables

Monitoring

The server includes:

• Prometheus metrics for request counts, latencies, and server executions
• Structured JSON logging
• Health check endpoint
• Server statistics endpoint

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

Additional Dependencies for Notebook and LLM Integration

This project now requires the following additional Python packages:

• pandas
• numpy
• matplotlib
• papermill
• nbformat
• jupyter
• anthropic

Install all dependencies with:

pip install -r requirements.txt

Using the Notebook MCP to Call an LLM (Claude)

The example notebook (mcp/notebooks/example.ipynb) demonstrates:

• Data analysis and plotting
• Calling the Claude LLM via the anthropic Python package

To use the LLM cell, ensure you have set your ANTHROPIC_API_KEY in your environment or .env file.

The notebook cell for LLM looks like this:

import os
import anthropic

api_key = os.getenv('ANTHROPIC_API_KEY')
if not api_key:
    raise ValueError('ANTHROPIC_API_KEY not set in environment!')

client = anthropic.Anthropic(api_key=api_key)
response = client.messages.create(
    model='claude-3-sonnet-20240229',
    max_tokens=256,
    temperature=0.7,
    messages=[
        {'role': 'user', 'content': 'Tell me a joke about data science.'}
    ]
)
print('Claude says:', response.content[0].text)