AI Agents Tutorial

This repository contains a step-by-step tutorial that guides you through building increasingly sophisticated AI agent systems. Each level builds upon the previous one, demonstrating core concepts in AI agent development.

Tutorial Levels

The tutorial is structured in progressive levels:

1. Basic LLM: Use only an AI and nothing else
2. Agent with Context: Adds short-term memory to maintain conversation relevance
3. Agent with Context and System Prompt: Uses a persistent system prompt to guide behavior and tone
4. Agent with Context and Read-Only Tools: Agent can query live crypto prices and view historical trends
5. Agent with Context, Write Access, and Wallet Integration: Agent can execute trades and interact with blockchain via wallet

Companion Guide

For a guided walkthrough of using this demo and understanding each level, check out the companion guide: AI Agents: From 0 to 1

Getting Started

Development

• The frontend is built with Remix
• The backend uses Koa and implements the Model Context Protocol
• Each level demonstrates a different aspect of AI agent architecture

For more information on the Remix framework, visit remix.run.

Prerequisites

• Node.js v22.13.1 (this repository has only been tested with this version)
  • You can use nvm to install and manage Node.js versions:

    nvm install 22.13.1
    nvm use 22.13.1

Setup

1. Clone this repository
2. Install dependencies:

   npm ci

3. Set up environment variables:
   • Copy .env.example to .env
   • Fill in the required API keys and wallet information:
     • OpenAI API key OR Gemini API key for having access to an LLM model
     • Alchemy API key (for interacting with the blockchain)
     • Wallet private key (use the keygen script in scripts/keygen.ts to generate)

To get the wallet private key run

npm run keygen

Running the Application

Start both the Remix frontend and the MCP server:

npm run dev:all

• Frontend will be available at: http://localhost:5173
• MCP server will run in the background

Testing the MCP Server

You can test the Model Context Protocol server using the MCP Inspector:

This will open the MCP Inspector interface where you can interact with your MCP server and debug requests/responses.

For more information on the Model Context Protocol, visit modelcontextprotocol.io.