A modern, web-based chat application that implements the Model Context Protocol (MCP) for seamless integration between Large Language Models and external tools.

MCP Open Client is a NiceGUI-based chat application that serves as a bridge between AI models and external tools through the Model Context Protocol (MCP). Think of it as "USB-C for AI" - a universal interface that allows any compatible LLM to securely interact with external data sources, tools, and services.

• 🤖 Multi-LLM Support: Compatible with Claude, OpenAI, and other OpenAI-compatible APIs
• 🔧 MCP Integration: Connect to MCP servers for enhanced functionality
• 💬 Modern Chat Interface: Clean, responsive web UI built with NiceGUI
• 📚 Conversation Management: Save, load, and organize your chat history
• ⚙️ Easy Configuration: Web-based settings management
• 🌐 Local & Remote: Works with local models and cloud APIs
• 🔒 Secure: All configurations stored locally

┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│   Web Client    │◄──►│   MCP Client    │◄──►│  MCP Servers    │
│   (NiceGUI)     │    │  (Protocol)     │    │  (FastMCP)      │
└─────────────────┘    └─────────────────┘    └─────────────────┘
         │                       │                       │
         ▼                       ▼                       ▼
    Chat Interface         API Integration         External Tools
    • Conversations        • OpenAI API           • File Systems
    • History             • Claude API            • Databases
    • Settings            • Local Models          • Web Services

pip install mcp-open-client
mcp-open-client

git clone https://github.com/alejoair/mcp-open-client.git
cd mcp-open-client
pip install -e .
mcp-open-client

git clone https://github.com/alejoair/mcp-open-client.git
cd mcp-open-client
pip install -r requirements.txt
python -m mcp_open_client.main

1. Launch the application:

   mcp-open-client

2. Open your browser to http://localhost:8080

3. Configure your API settings:

   • Go to Configuration → API Settings
   • Add your API key and model preferences
   • Choose from OpenAI, Claude, or local models

4. Set up MCP servers (optional):

   • Go to Configuration → MCP Servers
   • Add servers for enhanced functionality

5. Start chatting!

Configure your preferred AI model in the web interface:

• API Key: Your OpenAI/Claude API key
• Base URL: API endpoint (supports local models like LM Studio)
• Model: Choose your preferred model
• System Prompt: Customize the assistant's behavior

Connect external tools and services:

{
  "mcpServers": {
    "mcp-requests": {
      "disabled": false,
      "command": "uvx",
      "args": ["mcp-requests"],
      "transport": "stdio"
    },
    "mcp-code-editor": {
      "disabled": false,
      "command": "uvx",
      "args": ["mcp-code-editor"]
    }
  }
}

The client works with any MCP-compliant server. Popular options include:

• mcp-requests: HTTP request capabilities
• mcp-code-editor: File system operations
• mcp-database: Database connectivity
• Custom servers: Build your own with FastMCP

mcp_open_client/
├── main.py                    # Application entry point
├── api_client.py             # LLM API communication
├── mcp_client.py             # MCP protocol handler
├── config_utils.py           # Configuration management
├── settings/                 # Default configurations
│   ├── user-settings.json    # API settings
│   └── mcp-config.json      # MCP server config
└── ui/                      # User interface components
    ├── home.py              # Home page
    ├── chat_interface.py    # Chat UI
    ├── configure.py         # Settings UI
    ├── mcp_servers.py       # MCP management
    └── chat_handlers.py     # Chat logic

python -m pytest tests/

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

• Real-time messaging with streaming responses
• Conversation history with search and organization
• Message formatting with syntax highlighting
• Tool integration via MCP protocol

• Web-based settings - no need to edit config files
• Multiple API providers - OpenAI, Anthropic, local models
• MCP server management - enable/disable servers on the fly
• Theme customization - dark/light mode support

• Protocol compliance - works with any MCP server
• Dynamic server loading - add servers without restart
• Tool discovery - automatic detection of available tools
• Error handling - graceful degradation when servers are unavailable

• Python 3.7+
• Modern web browser
• Internet connection (for cloud APIs) or local model setup

• nicegui - Web UI framework
• openai - API client library
• fastmcp - MCP protocol implementation
• websockets - WebSocket support
• requests - HTTP client
• jsonschema - Configuration validation

Connection Problems:

• Check your API key and base URL
• Verify network connectivity
• Ensure MCP servers are running

UI Issues:

• Clear browser cache
• Try a different browser
• Check browser console for errors

MCP Server Issues:

• Verify server configuration
• Check server logs
• Test servers independently

• 📖 Check the documentation
• 🐛 Report bugs
• 💬 Ask questions

This project is licensed under the MIT License - see the LICENSE file for details.

• Built with NiceGUI for the web interface
• Uses FastMCP for MCP protocol implementation
• Inspired by the Model Context Protocol specification

• Homepage: https://github.com/alejoair/mcp-open-client
• Documentation: Wiki
• Bug Tracker: Issues
• MCP Specification: modelcontextprotocol.io

Made with ❤️ by alejoair