LangChain Multi-Agent System with LM Studio

This project implements a multi-agent system using LangChain and LM Studio, featuring three specialized agents: Research, Planning, and Execution.

Prerequisites

• Node.js (v16 or higher)
• LM Studio running locally
• npm or yarn

Setup

1. Install dependencies:

npm install

2. Configure LM Studio:

   • Download and install LM Studio
   • Start a local server in LM Studio (default port: 1234)
   • Load your preferred model in LM Studio

3. Configure environment variables:

   • The .env file should already be set up with default values
   • Modify if your LM Studio is running on a different port

Running the System

Start the application:

npm start

The system will run a test query about website performance improvement, demonstrating the interaction between the three agents:

1. Research Agent: Analyzes the problem and gathers information
2. Planning Agent: Creates a structured plan based on the research
3. Execution Agent: Provides concrete implementation steps

Customization

You can modify the agents' behavior by:

• Adjusting their system prompts in index.js
• Changing the temperature setting for different levels of creativity
• Adding more agents or modifying the chain of execution

Architecture

The system uses a sequential chain of agents:

1. Research → Planning → Execution

Each agent has a specific role and builds upon the work of the previous agent.