Nest Llm Argent

🚀 Nest-LLM-AIagent

This project is a forwarding solution designed to address the challenges of integrating the MCP (Model Context Protocol) into existing business systems. It acts as an AI agent, enabling the configuration of multiple MCP servers and providing a unified HTTP service for large models.

🚀 Quick Start

This project aims to solve the integration problem of the MCP protocol in existing business systems. The MCP protocol, while providing a unified standard, has a natural inclination towards the CS architecture, which makes it difficult to integrate into existing business. This forwarding solution not only solves this problem but also allows the configuration of multiple MCP servers and provides a unified HTTP service for large models.

The following is the specific architecture diagram:

✨ Features

Easy Integration

It can quickly connect to existing Web services (such as those built on NestJS) through an adaptation layer.

Seamless Expansion

In the form of a private NPM package, it enables the rapid integration and unified management of MCP servers while maintaining high portability.

Convenient Deployment

Through a private NPM package, it enables centralized management and version control of MCP servers, seamlessly integrating them into the existing release process for easy deployment.

Wide Model Support

It supports almost all types of large model calls.

📚 Documentation

Interface Definitions

To meet the functional requirements of different scenarios, the following standardized interfaces are designed:

📦 Installation

It is recommended to deploy the MCP server in an integrated manner to improve flexibility and unified management. The following solutions are recommended:

Private NPM Package

Package the MCP server as an NPM package and distribute and manage it through the enterprise private NPM repository.

Installation and Configuration

Install the MCP server package directly via npm install and achieve rapid integration through a configuration file (such as mcp.config.json).

Example configuration file mcp.config.json:

{
  "mcpServers": {
    "server1": {
      "name": "example-server",
      "args": ["server.js"],
      "path": "./servers/server1/"
    }
  },
  "mcpClient": {
    "name": "mcp-client",
    "version": "1.0.0"
  }
}

🔧 Technical Details

Current Status

• Currently, the MCP server only implements the stdio of the SSE protocol.
• During requests, the context protocol and global variables can be shared among the interface layer, MCP client layer, and MCP server layer.
• Currently, only HTTP interfaces are supported.

Future Plans

• Future plans include encapsulating these interfaces into WebSockets to better support chat scenarios.
• After completion, there is an idea to write a common class based on the MCP native protocol, enabling extension programs (MCP servers) based on this class to be developed once and run anywhere.