AI Agent Demo Project

This project demonstrates a simple AI agent capable of reading, listing, and renaming files in a local directory using pydantic-ai. It supports Google Gemini, OpenAI, and DeepSeek models.

Features

• Multi-Model Support: Works with Google Gemini, OpenAI, and DeepSeek.
• File Operations: Can listing, reading, and renaming files in a safe test directory.
• Interactive CLI: Simple command-line interface for natural language interaction.

Prerequisites

• Python 3.10 or higher.
• An API key for your chosen AI provider (Google, OpenAI, or DeepSeek).

Setup

1. Clone the project: Navigate to the project directory.

2. Create and Activate Environment: You can choose between a local environment or a named global environment.

   Option A: Local Environment (Recommended) Creates a hidden .venv directory in the project root. Best for isolation and IDE integration.

   mamba env create -p ./.venv -f environment.yml
   conda activate ./.venv

   Option B: Named Environment Creates a named Conda environment (pydantic-ai-agent).

   mamba env create -f environment.yml
   conda activate pydantic-ai-agent

   Option C: Pip (Standard Python) Create a virtual environment manually and install dependencies.

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate
   pip install pydantic-ai python-dotenv openai google-generativeai

3. Configure Environment: Create a .env file in the project root to select your provider and set your API keys.

   Google Gemini (Default)

   AI_PROVIDER=google
   GOOGLE_API_KEY=your_gemini_key

   OpenAI

   AI_PROVIDER=openai
   OPENAI_API_KEY=your_openai_key
   # Optional:
   # OPENAI_MODEL_NAME=gpt-4o

   DeepSeek

   AI_PROVIDER=deepseek
   DEEPSEEK_API_KEY=your_deepseek_key

Usage

1. Run the Agent:

   python main.py

2. Interact:

   • The agent operates safely within the ./test directory.
   • Type commands in natural language, for example:
     • "List all files in the test folder"
     • "Create 2D strcuture and SMILES of water molecule."

Project Structure

• main.py: Entry point of the application. Handles configuration and runs the agent loop.
• tools.py: Defines the file operation tools (read, list, rename) used by the agent.
• environment.yml: Defines the project dependencies.
• test/: Directory where file operations are performed.