Agentic Monorepo Backend

This is a Python monorepo for an AI agentic backend workflow. The backend provides a basic scaffolding built with FastAPI, mimicking an agent-based architecture with routing capabilities based on generated model decisions.

Architecture Structure

agentic-monorepo/
├── backend/
│   ├── server/          # FastAPI server code (main.py)
│   ├── agent/           # Gemini agent logic (gemini_agent.py)
│   ├── router/          # Model router (model_router.py)
│   ├── adapters/        # Model adapters (veo, nanobanana, lyria)
│   ├── schemas/         # Pydantic models (agent_output.py, prompt_request.py)
│   └── utils/           # Logging, helpers, config (logger.py, config.py)

✨ Features

• Modular monorepo architecture
• Agent-based decision making
• Dynamic routing to multiple AI providers
• Pluggable adapter design
• Async FastAPI backend
• Clean separation of concerns

Running the Application Locally

1. Install dependencies:

   pip install -r requirements.txt

2. Set up environment variables in .env:

   • GEMINI_API_KEY: Your Google Gemini API key
   • REDIS_URL: Your Redis Cloud connection URL (e.g., redis://username:password@host:port)

3. Run the FastAPI application from the project root directory (agentic-monorepo):

   python -m backend.server.main

   Or use uvicorn:

   uvicorn backend.server.main:app --reload

4. Test the POST /prompt endpoint:

   curl -X POST "http://localhost:8000/prompt" \
        -H "Content-Type: application/json" \
        -d '{"user_prompt": "I want to generate a video."}'

5. Get conversation history:

   curl "http://localhost:8000/conversations/{session_id}"