🤖 AgentAI

AgentAI is a simple CLI agent that uses Google AI gemini-2.0-flash-001 model to perform various file and code operations through natural language commands.

✨ Features

• 📁 List files and directories
• 📖 Read file contents
• 🐍 Execute Python files with optional arguments
• ✍️ Write or overwrite files
• 🗣️ Natural language interaction via CLI

🎯 Motivation

AgentAI CLI was coded in semi-guided project from the Boot.dev course. Special thanks to the Boot.dev team for the excellent guided projects and learning materials!

🔧 Installation

Prerequisites

• Python 3.10+
• uv package manager

Setup Steps

1. Clone the repository:

   git clone https://github.com/Jazzman94/agentai.git
   cd agentai

2. Install dependencies:

   uv sync

   If you don't have uv installed, install it with: pip install uv

3. Set up API key:

   • Create your API key in Google AI Studio
   • Create a .env file in the root directory:

   GEMINI_API_KEY=your_api_key_here

4. Test installation:

   python main.py "Hello world!" --verbose

⚡ Usage

Basic Commands

# Basic interaction
python main.py "List all Python files in the current directory"

# Verbose mode for detailed output
python main.py "Read the contents of lorem.txt" --verbose

# File operations
python main.py "Create a new file called test.py with a hello world function"

# Code execution
python main.py "Run the calculator.py file with argument 5"

Configuration

The agent's behavior is controlled by the prompts.system_prompt variable, which defines the CLI agent's purpose and capabilities.

Important configuration parameters in config.py:

MAX_CHARS = 10000      # Maximum characters per message
WORKING_DIR = "./calculator"  # Working directory for operations
MAX_ITERS = 20         # Maximum iterations in single request

Feel free to adjust these values based on your needs:

• Change WORKING_DIR to point to your desired working directory
• Modify MAX_CHARS to allow longer/shorter messages
• Adjust MAX_ITERS for more complex operations

🚀 Advanced Usage

Don't hesitate to experiment! Here's how to test the agent's full potential:

Example: Building a REST API

1. Create a new working directory:

   mkdir example

2. Update configurateion

   # In config.py
   WORKING_DIR = "./example"

3. Run a complex task

• Run for example following:

python main.py "Create a simple REST API server using FastAPI framework with simple GET, POST, PUT, DELETE endpoints. Write down a manual how to run it and test it properly. Provide me a list of needed packages to install."

📁 Project Structure

agentai/
├── main.py           # Entry point
├── config.py         # Configuration settings
├── prompts.py        # System prompts
├── .env              # API keys (create this)
├── pyproject.toml    # Dependencies
└── README.md

🙏 Acknowledgments

• Boot.dev for the excellent AI agent course
• Google AI team for the Gemini API

⭐ If you find this project helpful, please give it a star on GitHub!