🔎 OpenAI Agent Server - MCP Bridge

!OpenAI API !NestJS !TypeScript

A powerful NestJS backend that bridges OpenAI's Agent framework with MCP (Model Context Protocol) servers, enabling AI assistants to execute real-world actions through natural language requests.

🌟 Key Features

• AI-Powered Chat Interface: Seamless integration with OpenAI's latest models
• Function Calling: Intelligent mapping of natural language to executable functions
• MCP Tool Execution: Bridge between AI models and MCP-compatible tools
• Task Management API: Complete CRUD operations for task management
• Stateless Architecture: Designed for scalability and reliability
• Developer-Friendly: Comprehensive documentation and type safety

🛠️ Tech Stack

• NestJS framework for robust backend architecture
• TypeScript for type-safe development
• OpenAI API with advanced function calling capabilities
• Axios for efficient HTTP requests
• MCP Protocol compatibility for tool execution

🔗 How It Works

This server acts as an intelligent bridge between your frontend application and MCP-compatible tools:

1. User Request: The frontend sends a natural language request
2. AI Processing: OpenAI models interpret the request and identify required actions
3. Function Mapping: The server maps AI intentions to appropriate MCP tools
4. Tool Execution: MCP tools are executed with the necessary parameters
5. Response Generation: Results are processed and returned to the user

🚀 Getting Started

Prerequisites

• Node.js (v16 or later)
• npm or yarn
• MCP server running (see the mcp-server project)
• OpenAI API key

Installation

1. Clone the repository:

git clone https://github.com/yourusername/mcp-agent-server-example.git
cd agent-server-example

2. Install dependencies:

npm install

3. Configure the environment:

Create a .env file based on the provided .env.example:

PORT=4000
OPENAI_API_KEY=your_openai_api_key
MCP_SERVER_URL=http://localhost:3001

4. Start the development server:

npm run start:dev

The server will be available at http://localhost:4000.

💻 API Endpoints

Chat API

POST /chat

Enables natural language interaction with the AI assistant.

Request body:

{
  "message": "Create a task to finish the report by Friday",
  "history": [
    { "role": "user", "content": "Hello" },
    { "role": "assistant", "content": "Hi there! How can I help you today?" }
  ]
}

Task Management API

GET /tasks?status=[all|pending|completed]  # List tasks with optional filtering
POST /tasks                                # Create a new task
PATCH /tasks/:id/complete                 # Mark a task as completed

📂 Project Structure

• src/controllers/ - API endpoints and request handling
• src/services/ - Business logic and integration with OpenAI and MCP
• src/interfaces/ - TypeScript type definitions
• src/config/ - Application configuration

🔧 Building for Production

To build the application for production:

npm run build

The compiled files will be in the dist directory.

🔄 Integration with Frontend

This server pairs perfectly with the mcp-agent-client-example project, which provides a web interface for interacting with this backend.

📋 Future Enhancements

• Authentication and user management
• Support for multiple MCP servers
• Streaming responses for real-time interactions
• Advanced context management for improved AI understanding
• Performance metrics and monitoring

📄 License

MIT

---

Bridging the gap between natural language and executable actions 🤖⚒️