Agent Zero - Gaia Integration

This fork of the Agent Zero project integrates the Gaia node for model inference, replacing the original OpenAI API. Additionally, it includes a minor adjustment for Docker SSH connection to enhance reliability under proxy conditions.

Adjustments

1. Gaia Node Integration:

   • The entire process now uses Gaia's node for model inference: https://llama.us.gaianet.network/v1.
   • This replaces the previous OpenAI integration.

2. Docker SSH Address:

   • The SSH address for connecting to Docker has been changed from localhost to 127.0.0.1.
   • This adjustment prevents connection timeouts when running under certain proxy conditions.

Installation and Usage

Prerequisites

• Python 3.x
• Docker installed and running on your system

Steps

1. Install Dependencies:

   pip install -r requirements.txt

2. Start Docker:

   • Windows:
     • If Docker Desktop is installed, simply open Docker Desktop, and it will start automatically.
   • macOS:
     • If Docker Desktop is installed, open it from the Applications folder or use Spotlight Search.
   • Linux:
     • Ensure Docker is installed, and start the service with the following command:

       sudo systemctl start docker

3. Project Environment Variables Configuration Guide:

This project allows you to customize certain environment variables to connect your AI models and choose your preferred search engine. If you don't configure these variables, the program will run with default values.

① Configuring Environment Variables

First, you need to create a .env file in the root directory of the project. You can do this by copying the example.env file and renaming it to .env:

For Linux or macOS:

cp example.env .env

For Windows (Command Prompt):

copy example.env .env

For Windows (PowerShell):

Copy-Item -Path example.env -Destination .env

② Editing the .env File Open the .env file in your preferred text editor and fill in the following variables if you want to use your own models: CHAT_MODEL_BASE_URL CHAT_MODEL_NAME CHAT_API_KEY EMBEDDING_MODEL_BASE_URL EMBEDDING_MODEL_NAME EMBEDDING_API_KEY

Optional Configuration for PERPLEXITY Search

If you want to use PERPLEXITY search, add your API key by filling in the API_KEY_PERPLEXITY variable If this variable is not set, the program will default to using DuckDuckGo combined with the chat model you configured.

3. Run the Project:

   python main.py

Notes

• The project runs without requiring any environment variables to be set.
• The performance and output should be nearly identical to the original version using OpenAI.