# open-mcp-client

> MCP Client

This repository contains a LangGraph agent service and a CopilotKit frontend application. The agent connects to MCP servers and utilizes their tools, while the frontend provides UI and state synchronization.

## Overview

- **Category:** AI
- **Language:** TypeScript
- **Stars:** 1646
- **Forks:** 135
- **Owner:** CopilotKit
- **GitHub:** https://github.com/CopilotKit/open-mcp-client
- **Homepage:** https://open-mcp-client.vercel.app/
- **Created:** 2025-03-08T19:22:42+00:00
- **Updated:** 2025-03-29T00:03:03+00:00
- **Source:** https://model-context-protocol.com/clients/langgraph-mcp-agent-copilotkit-frontend

## Setup

## Setup

1.  **Set Up Environment Variables:** Create a `.env` file at the root and add `LANGSMITH_API_KEY` and `OPENAI_API_KEY`.
2.  Create another `.env` file inside the `agent` folder and add `OPENAI_API_KEY` and `LANGSMITH_API_KEY`.
3.  **Set Up Poetry:** Install Poetry using `pip install poetry`.
4.  Verify the installation using `poetry --version`.
5.  **Development:** Run the frontend and agent separately using `pnpm run dev-frontend` and `pnpm run dev-agent`.
6.  Alternatively, run both services together with `pnpm run dev`.
7.  Open [http://localhost:3000](http://localhost:3000) in your browser.

## Tools

## Available Tools

1.  LangGraph agent (connects to MCP servers and calls their tools).
2.  Frontend application (uses CopilotKit for UI and state synchronization).