README.md

metadata

title: Lightweight LLM with MCP Integration
emoji: 🧠
colorFrom: green
colorTo: blue
sdk: docker
app_port: 7860
pinned: false
license: mit
tags:
  - llm
  - mcp-integrated
  - inference
  - fastapi
  - prestashop
short_description: Lightweight LLM with integrated PrestaShop MCP tools

Lightweight LLM with Integrated MCP

A lightweight Large Language Model inference endpoint with integrated MCP (Model Context Protocol) server for enhanced PrestaShop capabilities. Everything runs in a single container for maximum efficiency and simplicity.

🚀 Features

• Lightweight Architecture: Single container with minimal dependencies
• Hugging Face Compatible: Standard HF API format support
• Integrated MCP: Built-in PrestaShop MCP server (no external dependencies)
• Fast Inference: Optimized for quick response times with local MCP calls
• PrestaShop Ready: Direct integration with PrestaShop APIs
• Docker Single Container: Everything in one container for easy deployment

📡 API Endpoints

Main Inference Endpoint: POST /

Hugging Face compatible format

Request:

{
  "inputs": "What are the features of product 123?",
  "parameters": {
    "max_new_tokens": 150,
    "temperature": 0.7,
    "prestashop_api_url": "https://your-shop.com"
  }
}

Response:

[
  {
    "generated_text": "Product 123 is a premium item with advanced features..."
  }
]

Health Check: GET /health

Service health and configuration status

Model Info: GET /info

Model and configuration information

🔧 Configuration

Environment Variables

• PRESTASHOP_API_URL: Default PrestaShop API URL
• MODEL_NAME: Hugging Face model to use (default: lightweight model)
• MAX_NEW_TOKENS: Default max tokens (default: 256)

Integrated MCP

The LLM includes a built-in MCP server with PrestaShop tools:

• Direct Integration: No network overhead, fastest possible responses
• Auto-Detection: Automatically detects when to use PrestaShop tools
• Fallback: Works without PrestaShop (direct responses)
• Single Container: Everything runs together for simplicity

🛠️ Usage

Direct Inference

curl -X POST https://your-inference-endpoint.hf.space/ \
  -H "Content-Type: application/json" \
  -d '{
    "inputs": "Hello, how are you?",
    "parameters": {"max_new_tokens": 100}
  }'

Integrated MCP Inference

curl -X POST https://your-inference-endpoint.hf.space/ \
  -H "Content-Type: application/json" \
  -d '{
    "inputs": "Tell me about product 123",
    "parameters": {
      "max_new_tokens": 200,
      "prestashop_api_url": "https://your-shop.com"
    }
  }'

🏗️ Architecture

HTTP Request → Lightweight LLM → Integrated MCP Server → Enhanced Response

1. Request Processing: Analyze user input for PrestaShop tool requirements
2. Tool Decision: Determine if PrestaShop tools are needed
3. MCP Integration: Call integrated MCP server directly (no network overhead)
4. Response Generation: Generate final response with enriched PrestaShop data
5. HF Format: Return in standard Hugging Face format

🚀 Deployment

Hugging Face Spaces

1. Create new Space with Docker SDK
2. Upload all files from inference_point/ directory
3. Set app_port to 7860 (already configured in README.md header)
4. Configure PRESTASHOP_API_URL environment variable

Hugging Face Inference Endpoints

1. Use this Docker setup for custom inference endpoints
2. Configure scaling and instance types as needed
3. Set environment variables for PrestaShop integration

📊 Performance

• Startup Time: < 30 seconds
• Inference Speed: 1-3 seconds for direct responses
• MCP-Enhanced: 2-5 seconds (local integration, no network overhead)
• Memory Usage: Optimized for single container deployment
• Concurrent Requests: Supports multiple simultaneous requests

🔗 Integrated MCP Tools

This service includes built-in PrestaShop MCP tools:

• get_product_details: Get complete product information
• get_product_features: Get product specifications and features
• get_product_images: Get product images and media
• search_products: Search the product catalog

Keywords that trigger MCP usage:

• product, price, stock, inventory
• order, customer, database
• search, find, get, fetch, check
• prestashop, shop, cart, purchase

📝 License

MIT License - Feel free to use and modify for your projects.

---

⚡ Lightweight & Fast: Optimized for production inference workloads.