🤖 AI Customer Support Bot - MCP Server

A modern, extensible MCP server framework for building AI-powered customer support systems

Features • Quick Start • API Reference • Architecture • Contributing

🌟 Overview

A Model Context Protocol (MCP) compliant server framework built with modern Python. Designed for developers who want to create intelligent customer support systems without vendor lock-in. Clean architecture, battle-tested patterns, and ready for any AI provider.

graph TB
    Client[HTTP Client] --> API[API Server]
    API --> MW[Middleware Layer]
    MW --> SVC[Service Layer]
    SVC --> CTX[Context Manager]
    SVC --> AI[AI Integration]
    SVC --> DAL[Data Access Layer]
    DAL --> DB[(PostgreSQL)]

✨ Features

🏗️ Clean Architecture
Layered design with clear separation of concerns

📡 MCP Compliant
Full Model Context Protocol implementation

🔒 Production Ready
Auth, rate limiting, monitoring included

🚀 High Performance
Built on FastAPI with async support

🔌 AI Agnostic
Integrate any AI provider easily

📊 Health Monitoring
Comprehensive metrics and diagnostics

🛡️ Secure by Default
Token auth and input validation

📦 Batch Processing
Handle multiple queries efficiently

🚀 Quick Start

Prerequisites

Python 3.8+
PostgreSQL
Your favorite AI service (OpenAI, Anthropic, etc.)

Installation

# Clone and setup
git clone https://github.com/ChiragPatankar/AI-Customer-Support-Bot--MCP-Server.git
cd AI-Customer-Support-Bot--MCP-Server

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Setup environment
cp .env.example .env
# Edit .env with your configuration

Configuration

# .env file
DATABASE_URL=postgresql://user:password@localhost/customer_support_bot
SECRET_KEY=your-super-secret-key
RATE_LIMIT_REQUESTS=100
RATE_LIMIT_PERIOD=60

Run

# Setup database
createdb customer_support_bot

# Start server
python app.py
# 🚀 Server running at http://localhost:8000

📡 API Reference

Core Endpoints

Health Check

GET /mcp/health

Process Single Query

POST /mcp/process
Content-Type: application/json
X-MCP-Auth: your-token
X-MCP-Version: 1.0

{
  "query": "How do I reset my password?",
  "priority": "high"
}

Batch Processing

POST /mcp/batch
Content-Type: application/json
X-MCP-Auth: your-token

{
  "queries": [
    "How do I reset my password?",
    "What are your business hours?"
  ]
}

Response Format

Success Response

{
  "status": "success",
  "data": {
    "response": "Generated AI response",
    "confidence": 0.95,
    "processing_time": "120ms"
  },
  "meta": {
    "request_id": "req_123456",
    "timestamp": "2024-02-14T12:00:00Z"
  }
}

Error Response

{
  "code": "RATE_LIMIT_EXCEEDED",
  "message": "Rate limit exceeded",
  "details": {
    "retry_after": 60,
    "timestamp": "2024-02-14T12:00:00Z"
  }
}

🏗️ Architecture

Project Structure

📦 AI-Customer-Support-Bot--MCP-Server
├── 🚀 app.py              # FastAPI application
├── 🗄️  database.py         # Database configuration
├── 🛡️  middleware.py       # Auth & rate limiting
├── 📋 models.py          # ORM models
├── ⚙️  mcp_config.py      # MCP protocol config
├── 📄 requirements.txt   # Dependencies
└── 📝 .env.example      # Environment template

Layer Responsibilities

Layer	Purpose	Components
API	HTTP endpoints, validation	FastAPI routes, Pydantic models
Middleware	Auth, rate limiting, logging	Token validation, request throttling
Service	Business logic, AI integration	Context management, AI orchestration
Data	Persistence, models	PostgreSQL, SQLAlchemy ORM

🔌 Extending with AI Services

Add Your AI Provider

Install your AI SDK:

pip install openai  # or anthropic, cohere, etc.

Configure environment:

# Add to .env
AI_SERVICE_API_KEY=sk-your-api-key
AI_SERVICE_MODEL=gpt-4

Implement service integration:

# In service layer
class AIService:
    async def generate_response(self, query: str, context: dict) -> str:
        # Your AI integration here
        return ai_response

🔧 Development

Running Tests

pytest tests/

Code Quality

# Format code
black .

# Lint
flake8

# Type checking
mypy .

Docker Support

# Coming soon - Docker containerization

📊 Monitoring & Observability

Health Metrics

✅ Service uptime
🔗 Database connectivity
📈 Request rates
⏱️ Response times
💾 Memory usage

Logging

# Structured logging included
{
  "timestamp": "2024-02-14T12:00:00Z",
  "level": "INFO",
  "message": "Query processed",
  "request_id": "req_123456",
  "processing_time": 120
}

🔒 Security

Built-in Security Features

🔐 Token Authentication - Secure API access
🛡️ Rate Limiting - DoS protection
✅ Input Validation - SQL injection prevention
📝 Audit Logging - Request tracking
🔒 Environment Secrets - Secure config management

🚀 Deployment

Environment Setup

# Production environment variables
DATABASE_URL=postgresql://prod-user:password@prod-host/db
RATE_LIMIT_REQUESTS=1000
LOG_LEVEL=WARNING

Scaling Considerations

Use connection pooling for database
Implement Redis for rate limiting in multi-instance setups
Add load balancer for high availability
Monitor with Prometheus/Grafana

🤝 Contributing

We love contributions! Here's how to get started:

Development Setup

# Fork the repo, then:
git clone https://github.com/your-username/AI-Customer-Support-Bot--MCP-Server.git
cd AI-Customer-Support-Bot--MCP-Server

# Create feature branch
git checkout -b feature/amazing-feature

# Make your changes
# ...

# Test your changes
pytest

# Submit PR

Contribution Guidelines

📝 Write tests for new features
📚 Update documentation
🎨 Follow existing code style
✅ Ensure CI passes

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

Built with ❤️ by Chirag Patankar

⭐ Star this repo if you find it helpful! ⭐

Report Bug • Request Feature • Documentation