mcp-client-app

# MCP Client Application A desktop application for interacting with AI models via the Model Context Protocol (MCP), providing a powerful interface for chat conversations with tool execution capabilities. ## What is MCP? The Model Context Protocol (MCP) is a standardized communication protocol that enables AI models to request execution of tools and functions with user confirmation. This creates a secure and transparent environment for AI assistants to interact with your system while maintaining user control. ## Features - 🤖 **Multi-model support**: Connect to Claude, GPT, and other LLMs through a unified interface - 🔧 **Tool execution**: Allow AI models to perform actions through a secure permission system - 🔄 **Real-time streaming**: Experience fluid conversations with streaming responses - 🔌 **Server management**: Register, configure and connect to MCP servers - 💾 **Conversation history**: Save and revisit your chat sessions - 🔐 **Secure vault**: Store sensitive information like API keys safely ## Why Use MCP Client? MCP Client demonstrates how to create applications that leverage the full potential of modern AI systems while maintaining appropriate security boundaries. It serves as both a practical tool and a learning resource for developers interested in building applications with AI capabilities. ## Getting Started ### Prerequisites - Node.js 18+ - npm package manager ### Installation ```bash # Clone the repository git clone git@github.com:RegiByte/mcp-client-app.git cd mcp-client-app # Install dependencies npm install # Start the development server npm dev ``` ### Building for Production ```bash # For Windows npm build:win # For macOS npm build:mac # For Linux npm build:linux ``` ## Using the Application ### Starting a Conversation 1. Open the application 2. Navigate to the conversations page 3. Click on "New conversation" 4. Start chatting! ### Working with Tools When an AI assistant needs to perform an action: 1. You'll see a tool execution request in the chat 2. Review the requested action and parameters 3. Approve or deny the request 4. See the results integrated back into the conversation ## Architecture Overview The MCP Client App is built with: - **Electron**: For cross-platform desktop capabilities - **React**: For the user interface - **TypeScript**: For type-safe code - **Vercel AI SDK**: For standardized AI model interactions - **Model Context Protocol**: For secure and standardized tool execution The application uses a driver pattern to support multiple LLM providers while maintaining a consistent interface. ## Learning Resources - [MCP Protocol Documentation](https://modelcontextprotocol.github.io/) - [Vercel AI SDK Guide](https://sdk.vercel.ai/docs) - [Building MCP-enabled Applications](https://yourwebsite.com/tutorials) - [Tool Development Guide](https://yourwebsite.com/tool-guide) ## Contributing We welcome contributions to the MCP Client App! Please see our [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines. ## License This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details. ## Acknowledgments - The Anthropic team for Claude and MCP - Vercel for the AI SDK - The Electron community for the application framework --- _Project maintained by [@RegiByte](https://github.com/RegiByte). For support, open an issue on the repository._