Company Logo
MCIP

Menu

Get started

Quick Start Guide

This guide provides the fastest path to get a complete MCIP stack running locally using the recommended Docker Compose setup.

1. Prerequisites

Before you begin, ensure you have the following installed and configured:

  • Node.js (v18 or higher)
  • Docker and Docker Compose
  • OpenAI API Key: You will need an active account and API key.
  • Pinecone Account: You will need an account to create a vector index.

2. Initial Environment Setup

First, set up your project directory and configuration file.

  1. Create Project Directory: Make a new directory for your project.
  2. Clone Repository: Clone the MCIP repository into your new directory.
  3. Create .env File: Find the configuration template in the repository and create a .env file from it. This file will hold your secret keys and environment-specific settings.

3. Configure External Services

Now, populate your .env file with the necessary credentials.

ServiceConfiguration Steps
OpenAI1. Generate an API key from your OpenAI dashboard.
2. Add the key to your .env file. MCIP defaults to the text-embedding-3-small model.
Pinecone1. Create a new index in your Pinecone account.
2. Configure the index for 512 dimensions and use the cosine similarity metric.
3. Add your Pinecone index name and environment to your .env file.
RedisNo action needed. For local development, the Docker Compose setup "handles everything automatically".

4. Connect Your E-commerce Platform

MCIP provides deep integration with Vendure. To get started quickly, connect your Vendure instance: 1.Find your Vendure GraphQL endpoint and authentication credentials. 2. Add this information to your .env configuration file. Note: If you are using a different platform, you will need to build an adapter, which is a more advanced process.

5. Run the Stack

With your configuration complete, you can launch the entire MCIP stack. The Docker Compose configuration "sets up the entire stack with a single command".

Run the following command in your project's root directory:

docker-compose up

This command ensures all components (the MCIP server, Redis, etc.) are correctly configured and networked together.

6. Make Your First Search Request

Once the stack is running, you can connect and send your first query.

  1. Establish Connection: Connect to your local MCIP server using the MCP protocol over HTTP or WebSocket. The server will automatically create a session and return a session identifier.
  2. Send Search Query: Send a search_product request. Start with a simple query like "laptop" to verify the connection.
  3. Verify Response: You should receive a response containing:
  • An array of products with normalized fields (id, name, price, etc.).
  • A relevance_score for each product.
  • metadata about the search, such as the search method used (e.g., "RAG semantic search").
  1. Try Semantic Search: Now, test the system's power with a natural language query like "equipment for starting a podcast on a budget". You should see relevant results like microphones and audio interfaces, even if they don't contain the exact keywords.