MCP Server with LLM Integration and Model Context Protocol (MCP) Tools Management

This Flask web application enables users to:

• Approach LLM models (OpenAI and Anthropic) via a simple web interface
• Input prompts and receive responses
• Start and stop MCP servers (Model Context Protocol) for GitHub and Brave Search from the UI
• Automatically enrich your prompt with context provided by these tools before sending it to the LLM

Overview: What is the MCP Server?

The MCP server acts as a bridge between AI applications like Claude Desktop, Continue, Cursor, and other LLM clients that support the Model Context Protocol (MCP). It integrates directly into the application stack, providing enhanced functionality by leveraging external data sources and tools. This server allows for seamless context gathering, enrichment of prompts before they reach the LLM, and ultimately, improved response accuracy.

🔧 Core Features & MCP Capabilities

The core features of this MCP server include:

• LLM Model Selection: Support for selecting between OpenAI GPT-3.5 and Anthropic Claude 2 models.
• Prompt Input: A user-friendly interface for inputting prompts or questions.
• MCP Tools Management: Control over the start and stop of external tools, including Brave Search and GitHub integration.
• Context Enrichment: Automatically enhances your prompt with relevant context provided by active MCP tools before sending it to the LLM.
• Full Prompt Display: Shows the complete prompt, including any added context for clarity.

⚙️ MCP Architecture & Protocol Implementation

The architecture of this server integrates seamlessly with the Model Context Protocol (MCP). It comprises a web application built using Flask to provide an intuitive user interface and an underlying API for interacting with external tools. The protocol flow is designed to ensure secure and efficient communication between the AI application, the MCP server, and the respective data sources or tools. This architecture supports a wide range of use cases, ensuring that AI applications can leverage diverse resources while maintaining high security standards.

🚀 Getting Started with Installation

Prerequisites

• Python 3.7 or higher

Steps to Install

1. Clone the repository

   git clone https://github.com/Fbeunder/MCP_FLASK.git
   cd MCP_FLASK
   

2. Set up a virtual environment (recommended)

   # Windows
   python -m venv venv
   venv\Scripts\activate
   
   # macOS/Linux
   python3 -m venv venv
   source venv/bin/activate
   

3. Install the required packages

   # Recommended: install all dependencies via requirements.txt
   pip install -r requirements.txt
   
   # Alternative: install specific packages if needed
   pip install flask requests
   
   # For OpenAI model access
   pip install openai
   
   # For Anthropic model access
   pip install anthropic
   
   # For .env file support
   pip install python-dotenv
   

4. Configure API keys

   • You can configure API keys via environment variables or a .env file.

   Option 1: Create a .env file

   # Windows
   copy .env.example .env
   
   # macOS/Linux
   cp .env.example .env
   

   Edit the .env file and fill in your API keys.

   Option 2: Set environment variables directly

   export OPENAI_API_KEY="your-openai-api-key"
   export ANTHROPIC_API_KEY="your-anthropic-api-key"
   export BRAVE_API_KEY="your-brave-api-key"
   export GITHUB_TOKEN="your-github-token"
   

   On Windows:

   set OPENAI_API_KEY=your-openai-api-key
   set ANTHROPIC_API_KEY=your-anthropic-api-key
   set BRAVE_API_KEY=your-brave-api-key
   set GITHUB_TOKEN=your-github-token
   

💡 Key Use Cases in AI Workflows

1. Code Debugging and Optimization:

   • Scenario: A developer using Claude Desktop needs to debug a piece of code. By enriching the prompt with context from GitHub, the response can provide more specific details and suggestions.

2. Data-Specific Research Queries:

   • Scenario: An AI researcher wants to ask detailed questions about a specific dataset. Using the MCP server, they can get richer information directly integrated into their query.

🔌 Integration with MCP Clients

The server is compatible with the following MCP clients:

• Claude Desktop: Fully supported
• Continue: Fully supported
• Cursor: Tools only (no prompt enrichment)

MCP Client	Resources	Tools	Prompts
Claude Desktop	✅	✅	✅
Continue	✅	✅	✅
Cursor	❌	✅	❌

📊 Performance & Compatibility Matrix

This server is optimized for performance and compatibility across various AI applications. The protocol implementation ensures that data flows securely and efficiently, minimizing latency and maximizing accuracy.

🛠️ Advanced Configuration & Security

Advanced configuration includes setting up environment variables, managing dependencies, and ensuring secure handling of API keys.

Example Configuration Code

{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "API_KEY": "your-github-token"
      }
    },
    "braveSearch": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-bravesearch"],
      "env": {
        "API_KEY": "your-brave-api-key"
      }
    }
  }
}

❓ Frequently Asked Questions (FAQ)

1. How do I set up API keys for the MCP server?
   • You can configure API keys via environment variables or a .env file.
2. Do you support all LLM clients?
   • Partially, we currently support Claude Desktop and Continue, while Cursor only provides tools without prompt enrichment.
3. Can I integrate my custom data sources with the server?
   • Yes, the server is highly customizable; you can add your own data sources following the protocol guidelines.
4. Is there any impact on performance when using the MCP server?
   • Minimal impact. The server is designed to optimize communication and reduce latency.
5. How do I test connection with an MCP client?
   • Use tools provided by the MCP client or our documentation for testing.

👨‍💻 Development & Contribution Guidelines

Contributions are welcome! If you would like to contribute, please follow these guidelines:

• Fork the repository
• Make changes and improvements
• Ensure all tests pass
• Submit a pull request

🌐 MCP Ecosystem & Resources

For more information on the Model Context Protocol (MCP) ecosystem, see the official documentation and resources.

MCP Protocol Flow Diagram:

graph TD
    A[AI Application] -->|MCP Client| B[MCP Protocol]
    B --> C[MCP Server]
    C --> D[Data Source/Tool]
    style A fill:#e1f5fe
    style C fill:#f3e5f5
    style D fill:#e8f5e8

MCP Client Compatibility Matrix:

MCP Client	Resources	Tools	Prompts
Claude Desktop	✅	✅	✅
Continue	✅	✅	✅
Cursor	❌	✅	❌

This detailed documentation positions the MCP server as a critical component for developers building AI applications and integrating MCP clients into their workflows. It focuses on enhancing AI performance through context enrichment, ensuring optimal integration with various tools and resources.