CatChat

A demo of using GenAIScript to create an interactive conversational application. Any GenAIScript configuration could be used. For the demo openai:gpt-4o-mini works best. If you want to use local models from ollama comment out the files: ['rag/**/*.*'] or they struggle.

🐱 Overview

Simulates a chat room where:

• A user inputs messages
• AI participants respond to the conversation
• An AI agent manages and summarizes the conversation
• A virtual cat interjects

🧩 Pieces

• Server: Express/Socket.IO server handling HTTP routes and real-time communication
• CatChat: Main state machine orchestrating the conversation inside a GenAIScript with a run loop
• Cat: AI agent that simulates cat behavior and is subscribed to state changes
• Participant: AI agents representing human participants in the chat
• Scribe: AI agent tracks and stores conversation history
• RAG: A folder containing interesting cat info is referenced by the AI agents

🚀 Setup

Prerequisites

• Node.js
• npm
• GenAIScript VS Code extension (if Debugging)

Steps

1. Clone the repository

2. Install dependencies

   npm install

3. Configure environment variables (copy .env.genaiscript.sample to .env.genaiscript. [

   OPENAI_API_KEY=your_openai_api_key
   

4. Open the folder in VS Code

5. Navigate to the cat-chat.genai.mts script and click the Run GenAIScript... or Debug GenAIScript... menu option, or run via the terminal npx genaiscript run cat-chat

6. A browser will launch and open http://localhost:3000