Generic Chatbot - Gemini API

A modular, configuration-driven chatbot backend built with Flask and Google Gemini API. Designed for hackathon adaptability with pluggable prompts and data contexts.

🎯 Features

• Modular Architecture: Separated concerns (config, service, brain, API)
• Dynamic Prompt Injection: Change system prompts at runtime without redeployment
• Conversation Memory: Maintains session-based conversation history
• Flexible Configuration: YAML + Environment variables
• RESTful API: Clean JSON endpoints for integration
• Error Handling: Robust handling of API errors and rate limits

📁 Project Structure

chatbot/
├── app/
│   ├── __init__.py          # Package initializer
│   ├── config.py            # Configuration management
│   ├── service.py           # Gemini API wrapper
│   ├── brain.py             # Conversation & context logic
│   └── main.py              # Flask application
├── config.yaml              # Default prompts & settings
├── .env.example             # Environment variables template
├── requirements.txt         # Python dependencies
├── run.py                   # Application entry point
└── README.md                # This file

🚀 Quick Start

1. Prerequisites

• Python 3.10+
• Google Gemini API Key (Get one here)

2. Installation

# Clone or navigate to project directory
cd chatbot

# Install dependencies
pip install -r requirements.txt

# Copy environment template
copy .env.example .env

# Edit .env and add your API key
notepad .env

3. Configuration

Edit .env file:

GEMINI_API_KEY=your_actual_api_key_here
FLASK_PORT=5000

Edit config.yaml to customize default prompts:

system_prompt: |
  You are a helpful AI assistant...

4. Run the Server

Note: All code is currently commented out. Uncomment the code in all files to run.

# Run the chatbot server
python run.py

Server will start at http://localhost:5000

📡 API Endpoints

1. Health Check

GET /health

2. Chat

POST /chat
Content-Type: application/json

{
  "message": "Hello, how are you?",
  "session_id": "user123",
  "include_history": true
}

Response:

{
  "success": true,
  "response": "I'm doing well, thank you!",
  "session_id": "user123",
  "model": "gemini-pro",
  "usage": {
    "prompt_tokens": 45,
    "completion_tokens": 12
  }
}

3. Update Configuration

POST /config
Content-Type: application/json

{
  "system_prompt": "You are a pirate assistant. Speak like a pirate!",
  "temperature": 0.9
}

4. Clear Session

DELETE /session/user123

5. List Sessions

GET /sessions

🔧 Extending for Hackathons

Adding Custom Data Context

1. In brain.py, modify build_prompt():

# Add context injection
if session.context_data:
    prompt_parts.append(f"CONTEXT DATA:\n{session.context_data}")

2. Set context via API:

session.set_context('user_profile', {'name': 'John', 'preferences': 'tech'})

Changing Prompts Dynamically

# During hackathon, update prompt based on problem statement
curl -X POST http://localhost:5000/config \
  -H "Content-Type: application/json" \
  -d '{"system_prompt": "You are a legal advisor assistant..."}'

Supporting Different Data Formats

1. Add data loader in brain.py:

def load_data_context(self, data_file: str):
    with open(data_file) as f:
        data = json.load(f)
    return data

2. Use in prompt building

🧪 Testing

# Test with curl
curl -X POST http://localhost:5000/chat \
  -H "Content-Type: application/json" \
  -d '{"message": "What can you do?"}'

🛡️ Error Handling

The system handles:

• Invalid API keys
• Rate limiting (automatic retry logic can be added)
• Malformed requests
• Network errors

📝 Configuration Reference

Environment Variables (.env)

Variable	Description	Default
GEMINI_API_KEY	Google Gemini API key	Required
FLASK_PORT	Server port	5000
GEMINI_MODEL	Model name	gemini-pro
TEMPERATURE	Model temperature	0.7

YAML Config (config.yaml)

• system_prompt: Default AI persona and instructions
• model: Model parameters (temperature, tokens, etc.)
• conversation: History settings

🎓 Hackathon Workflow

1. Setup (5 min): Install, configure API key
2. Test (5 min): Verify basic chat works
3. Problem Statement Released:
   • Update system_prompt via /config endpoint
   • Add domain data to context
   • Test with sample queries
4. Demo: Show API calls and responses

📦 Deployment

For production/demo:

pip install gunicorn
gunicorn -w 4 -b 0.0.0.0:5000 app.main:app

🤝 Contributing

This is a hackathon template. Feel free to modify any module to fit your use case.

📄 License

MIT License - Use freely for hackathons and projects.

---

Built with ❤️ for hackathon speed and flexibility