Warning

Do not try to use this in a production environment.

It does not have all the security and safety features that a production AI agent would have. It is for learning purposes only.

Prerequisites

• Python 3.x: Ensure you have a compatible version of Python installed (e.g., Python 3.9+).
• uv (or pip): The project uses uv for dependency management and running, which is a fast Python package installer and resolver. You can install it via pip:

  pip install uv

  Alternatively, you can use pip directly if you prefer, but uv is recommended by the project's structure.

Setup Instructions

1. Clone the Repository:

   git clone https://github.com/MN-nagy/AIagent.git
   cd AIagent

2. Create a Virtual Environment (Recommended with uv):

   uv venv
   uv pip install -r requirements.txt

   If using pip directly:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   pip install -r requirements.txt

3. Obtain a Google Gemini API Key:

   • Go to the Google AI Studio or Google Cloud Console.
   • Create a new API key.

4. Configure Environment Variables:

   • Create a file named .env in the root directory of the project (where main.py is located).
   • Add your Gemini API key to this file:

     GEMINI_API_KEY="YOUR_GEMINI_API_KEY_HERE"
     

     Replace "YOUR_GEMINI_API_KEY_HERE" with the actual API key you obtained.

Running the Agent

After completing the setup, you can run the agent using uv:

uv run python3 main.py "Your prompt here"

For verbose output to debug or see the agent's internal thought process (function calls and tool outputs):

uv run python3 main.py "Your prompt here" -v

---

Project Structure (for context)

AIagent/
├── .env
├── main.py
├── config.py
├── requirements.txt
├── pyproject.toml
├── functions/
│   ├── __init__.py
│   ├── call_func.py
│   ├── get_file_content.py
│   ├── get_files_info.py
│   ├── run_python_file.py
│   └── write_file.py
└── calculator/
    ├── __init__.py
    ├── main.py
    └── pkg/
        ├── __init__.py
        ├── calculator.py
        └── render.py

Note: Made this while following boot.dev course.