BuddyImagine

BuddyImagine is a Telegram bot that leverages the power of FAL AI for high-quality image generation and Cloudflare R2 for efficient storage.

Features

• 🤖 AI-Powered Generation: Uses an intelligent agent to understand your prompts.
• 🎨 Multiple Models: Supports various state-of-the-art models like Flux, SDXL, Recraft, and more.
• ☁️ Cloud Storage: Automatically saves all generated images to Cloudflare R2.
• 🔍 Model Discovery: Built-in fuzzy search to find the perfect model for your needs.

Commands

• /start - Initialize the bot and see the welcome message.
• /generate <prompt> - Generate an image. You can specify the model naturally in the prompt (e.g., "Generate a cyberpunk city using Flux").
• /models [query] - List available models or search for specific ones (e.g., /models flux).
• /help - Show the help message.

Setup

1. Clone the repository:

   git clone <repository_url>
   cd BuddyImagine

2. Install dependencies: This project uses uv for dependency management.

   uv sync

3. Configure Environment Variables: Create a .env file in the root directory with the following variables:

   BOT_TOKEN=your_telegram_bot_token
   OPENAI_API_KEY=your_openai_api_key  # For the agent logic
   FAL_KEY=your_fal_ai_key             # For image generation
   
   # Cloudflare R2 Configuration
   R2_ACCESS_KEY_ID=your_r2_access_key
   R2_SECRET_ACCESS_KEY=your_r2_secret_key
   R2_BUCKET_NAME=your_bucket_name
   R2_ENDPOINT_URL=your_r2_endpoint_url

4. Run the Bot:

   uv run main.py

Development

• Run Tests:

  uv run pytest

• Type Check:

  uv run mypy .

Architecture

• main.py: Entry point, sets up the bot and polling.

• handlers.py: Telegram message handlers (commands).

• agent.py: logic using pydantic-ai to process prompts and select tools.

• services/:

  • fal.py: Service for interacting with FAL AI API.
  • r2.py: Service for uploading/downloading files to Cloudflare R2.

  docker-compose down && docker-compose build --no-cache && docker-compose up -d