An AI-powered product search and discovery agent

A NestJS-based semantic search engine that enables natural language product discovery. Designed as a plug-and-play microservice that can connect to any e-commerce backend through configurable adapters.

• AI-Powered Search — Natural language queries with automatic filter extraction (brand, category, price range)
• Semantic Vector Search — Powered by Qdrant for fast similarity matching
• Hybrid Filtering — Combines vector search with hard filters (inclusion & exclusion)
• Flexible Adapters — Built-in Vendure support + AI-based fallback for any data source
• Async Processing — Robust queue-based ingestion with BullMQ and Redis
• Production Ready — Docker-first deployment with health checks and auto-recovery

┌─────────────────┐     ┌──────────────────┐     ┌─────────────────┐
│   Your Store    │────▶│       MCIP       │────▶│     Qdrant      │
│   (Vendure/API) │     │    (NestJS)      │     │   (Vector DB)   │
└─────────────────┘     └──────────────────┘     └─────────────────┘
                                │
                                ▼
                        ┌──────────────┐
                        │    Redis     │
                        │   (Queue)    │
                        └──────────────┘

Data Flow:

1. Ingestion — Products are fetched from your store via REST/GraphQL
2. Normalization — Raw data is mapped to a unified schema (via adapters)
3. Vectorization — OpenAI generates embeddings for semantic search
4. Storage — Products + vectors are stored in Qdrant
5. Search — User queries are vectorized and matched against products

• Docker & Docker Compose
• OpenAI API Key

git clone https://github.com/your-org/mcip.git
cd mcip

Create a .env file:

# Required
OPENAI_API_KEY=sk-your-api-key

# Data Source
SOURCE_URL=https://demo.vendure.io/shop-api
STORE_PROVIDER=VENDURE
GRAPHQL_QUERY={products{items{id name slug description variants{id sku name price priceWithTax currencyCode stockLevel}facetValues{name facet{name}}collections{name slug}featuredAsset{preview source}}}}

# Security
ADMIN_API_KEY=your-secret-admin-key

# Optional
PORT=8080
STOREFRONT_URL=https://your-store.com

docker-compose up -d

The API will be available at http://localhost:8080.

curl -X POST http://localhost:8080/admin/sync \
  -H "x-admin-api-key: your-secret-admin-key"

# Simple search
curl "http://localhost:8080/search?q=gaming+laptop"

# Natural language with filters
curl "http://localhost:8080/search?q=nike+shoes+under+100"

GET /search?q={query}&take={limit}&skip={offset}

Search products using natural language. Automatically extracts filters from queries like "Nike shoes under $100 but not running shoes".

POST /admin/sync
Headers: x-admin-api-key: {ADMIN_API_KEY}

Trigger product synchronization from configured source.

POST /admin/recreate-indexes
Headers: x-admin-api-key: {ADMIN_API_KEY}

Rebuild Qdrant payload indexes.

GET /health

Returns service health status.

# Install dependencies
npm install

# Start dependencies (Redis + Qdrant)
docker-compose up redis qdrant -d

# Run in development mode
npm run start:dev

npm run build

npm run test
npm run test:e2e

• Runtime: Node.js 20 + NestJS 11
• Vector Database: Qdrant
• Queue: BullMQ + Redis
• AI: OpenAI (embeddings + structured output)
• Validation: Zod schemas

This project is licensed under the MIT License.