Open MCP Client: Interact with MCP Servers in Your Apps via a Single URL 182

What is Open MCP Client?

Open MCP Client is the first Web-based MCP client project that enables interaction between an AI assistant and MCP servers through a single URL. The project uses a LangGraph agent built with LangChain to connect to MCP servers and invoke tools, while the front end is built with CopilotKit to deliver a smooth client-side experience. The project is currently open source.

Features and Highlights

• Minimal Integration: Open MCP Client offers an extremely simple integration approach—developers can embed MCP capabilities into their own applications through just one URL, with no need to build a communication layer from scratch.

• Multi-Server Collaboration: Supports simultaneous connection to multiple MCP servers—for example, one server for handling GitHub operations and another for managing Slack messages.

• Dynamic Tool Discovery: The client can dynamically discover and invoke the tools provided by servers, with no preset fixed logic required.

• Security and Flexibility: Supports both local and cloud deployment, with built-in access control mechanisms to meet enterprise-grade security requirements.

Use Cases

• Automated Development Workflows: For example, users can submit a request through Slack, and Open MCP Client will automatically invoke code generation tools and GitHub APIs to generate code and submit a Pull Request.

• Local Resource Access: Through the MCP protocol, the client can securely access local resources such as databases and file systems.

• AI Assistant Enhancement: Provides AI assistants with extended capabilities, such as calling external tools via MCP servers.

Technical Details

• Protocol Version Negotiation: Ensures compatibility with MCP servers.

• Capability Negotiation: Dynamically determines available features.

• Message Transport: Supports JSON-RPC communication.

• Tool Discovery and Execution: The client can discover and invoke the tools provided by servers.

Installation and Usage Guide

1. Clone the Project

git clone https://github.com/CopilotKit/open-mcp-client.git
cd open-mcp-client

2. Set Environment Variables

(1) Create a .env file in the project root directory:

LANGSMITH_API_KEY=your LangSmith key

(2) Create a .env file inside the /agent folder:

OPENAI_API_KEY=your OpenAI key
LANGSMITH_API_KEY=your LangSmith key

3. Install Dependencies

pnpm install

4. Start the Project

(1) Start the front end:

pnpm run dev-frontend

(2) Start the agent:

pnpm run dev-agent

Or simply run:

pnpm run dev

Open MCP Client is one of the core components of the MCP protocol. Through standardized interfaces and dynamic tool discovery capabilities, it greatly simplifies the interaction between AI assistants and external tools.

GitHub Repository:https://github.com/CopilotKit/open-mcp-client

Project DEMO:https://open-mcp-client.vercel.app/