Nango MCP Client Example

This repository demonstrates how to connect any Large Language Model (LLM) provider to the Nango Model Context Protocol (MCP) server using Vercel's AI SDK. It provides a simple interactive CLI interface for sending messages to the LLM with tool-calling capabilities via the MCP protocol.

Overview

The project creates a conversation interface that:

• Connects to the Nango MCP server
• Enables LLM tool-calling capabilities via the MCP protocol
• Uses OpenAI models through Vercel's AI SDK
• Provides a streaming interface for real-time responses

Prerequisites

• Node.js (v18 or newer)
• npm or yarn
• OpenAI API key

Setup Instructions

1. Clone the repository:

   git clone https://github.com/yourusername/nango-mcp-client-example.git
   cd nango-mcp-client-example

2. Copy the environment variables example file:

   cp .env.example .env

3. Configure your .env file with the following:

   OPENAI_API_KEY=your_openai_api_key
   NANGO_SECRET_KEY_DEV=your_nango_secret_key
   CONNECTION_ID=your_connection_id
   INTEGRATION_ID=your_integration_id
   

4. Install dependencies:

   npm install

5. Start the application:

   npm start

Usage

Once running, you can interact with the CLI:

• Type messages to send to the LLM
• Watch real-time responses with tool-calling capabilities
• Type exit to end the session