Digital Twin MCP Server

AI-powered RAG system built with Next.js, Upstash Vector, and Groq LLM, implementing the Model Context Protocol (MCP).

🎯 Overview

This project implements a Digital Twin knowledge base using:

• Next.js 15 - React framework with server actions
• Upstash Vector - Managed vector database for embeddings
• Groq API - Fast LLM inference (llama-3.1-8b-instant)
• MCP Protocol - Model Context Protocol for AI tool integration

📁 Project Structure

src/
├── actions/           # Server actions (business logic)
│   ├── query.ts      # Query digital twin with RAG
│   ├── embed.ts      # Add documents to vector DB
│   ├── reindex.ts    # Bulk reindex from JSON
│   └── stats.ts      # Get index statistics
├── app/
│   ├── api/          # API routes (HTTP endpoints)
│   │   ├── query/
│   │   ├── embed/
│   │   ├── reindex/
│   │   ├── health/
│   │   └── test/
│   └── page.tsx      # Landing page
├── lib/
│   └── upstash.ts    # Upstash Vector client
└── mcp/
    └── server.ts     # MCP server implementation

🚀 Quick Start

1. Install Dependencies

npm install

2. Configure Environment

Create .env.local:

UPSTASH_VECTOR_REST_URL="your-upstash-url"
UPSTASH_VECTOR_REST_TOKEN="your-upstash-token"
GROQ_API_KEY="your-groq-api-key"
ADMIN_API_KEY="your-admin-key"

3. Prepare Data

Place your digitaltwin.json in the parent directory or create data/digitaltwin.json.

4. Run Development Server

npm run dev

Visit http://localhost:3000

📡 API Endpoints

See full documentation in the project README.

🤖 MCP Server

The MCP server provides AI tool integration with 4 main tools:

• query_digital_twin
• add_document
• reindex_knowledge_base
• get_index_stats

📚 Tech Stack

• Next.js 15.5.6, TypeScript 5
• Upstash Vector 1.2.2
• Groq SDK (llama-3.1-8b-instant)
• Model Context Protocol SDK

---

Built for Digital Twin Workshop Step 7