Shopify MCP Proxy & Mock Server (ShopifyMCPMockShop)

This project implements a Model Context Protocol (MCP) server that acts as an intelligent bridge and proxy to Shopify's Storefront API. It uniquely supports mock.shop for safe development and testing without requiring real store credentials.

The server communicates via Standard Input/Output (stdio), making it suitable for local integration with clients like Claude Desktop and Cursor.

Motivation

Generic AI developer tools and off-the-shelf MCP servers often promise much but can lack the transparency, safety mechanisms (like explicit ToolAnnotations), stability, or specific features needed for controlled development and testing, especially against rapidly evolving APIs like Shopify's.

This project was born from the need for a reliable, open-source Shopify MCP server that:

• Prioritizes Safety: Integrates deeply with mock.shop to provide a crucial sandbox for risk-free development and accurate testing without impacting live stores.
• Emphasizes Transparency & Control: Allows developers (and their AI partners) to understand exactly how interactions with the Shopify API occur, including which API versions are used and the potential impact of actions (via ToolAnnotations like readonly, destructive, etc.).
• Addresses Specific Needs: Fills the gap left by purely informational tools or closed-source commercial offerings by providing a safe execution engine tailored for Shopify development workflows.

The goal is to enable developers to confidently leverage AI assistance for Shopify tasks, knowing they have a tool built with intentional guardrails and clear operational context.

Setup & Running Locally

This server is a standalone Node.js script.

1. Clone the repository:

   git clone https://github.com/ramakay/ShopifyMockMCP.git
   cd ShopifyMCPMockShop
   

2. Install dependencies:

   npm install
   

3. Build the server:

   npm run build
   

   This compiles the TypeScript code into JavaScript in the dist directory (dist/server.js).

4. Configure Environment (Optional - for Real Store):

   • If you want to connect to a real Shopify store instead of mock.shop, set the following environment variables before running the server (e.g., via export or a .env file loaded by your client):
     • SHOPIFY_STORE: Your store domain (e.g., your-store.myshopify.com). Required for real store usage.
     • SHOPIFY_ACCESS_TOKEN: Your Storefront API access token. Required for real store usage.
     • SHOPIFY_VERSION: Optional specific Storefront API version (e.g., 2025-04). Defaults to latest stable.
   • Note: Using mock.shop (default when SHOPIFY_STORE is not set) requires no environment variables.
   • Note: Admin API functionality is currently disabled but planned for future updates. Configuration variables (USE_ADMIN_API, ADMIN_ACCESS_TOKEN, ADMIN_VERSION) are not currently used.

5. Running: The server is intended to be launched automatically by MCP clients (like Claude Desktop or Cursor) using the stdio configuration. You do not need to run npm run start manually. See the "Usage" section below.

Usage with Cursor or Claude Desktop

Configure your MCP client (Cursor, Claude Desktop) to launch this server via stdio. Add an entry like the following to the respective configuration file:

• Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json
• Cursor: ~/.cursor/mcp.json (or similar)

{
  "mcpServers": {
    // ... other servers ...
    "shopify-mcp": {
      "command": "node",
      "args": [
        "/full/path/to/ShopifyMCPMockShop/dist/server.js"
      ],
      "env": {
        // Only needed if connecting to a REAL store:
        // "SHOPIFY_STORE": "your-store.myshopify.com",
        // "SHOPIFY_ACCESS_TOKEN": "your-storefront-token",
        // "SHOPIFY_VERSION": "2025-04"
      },
      "disabled": false,
      "alwaysAllow": []
    }
    // ... other servers ...
  }
}

Important:

• Replace /full/path/to/ShopifyMCPMockShop with the absolute path to where you cloned the repository.
• Only include the env variables if you are connecting to a real store.
• Restart your client (Cursor/Claude Desktop) after modifying the configuration.

Available Tools

This MCP server currently provides tools for interacting with the Shopify Storefront API.

Tool Name	Description	API
getShopInfo	Fetches basic shop details (name, description, currency).	Storefront
getProductById	Fetches a specific product by ID. If includeImages is true, returns JSON details and the first image (75px) as a base64 ImageContentBlock.	Storefront
findProducts	Searches/filters products with pagination/sorting.	Storefront
getCollectionById	Fetches a specific collection by ID, optionally including products.	Storefront
findCollections	Searches/filters collections with pagination/sorting.	Storefront
cartCreate	Creates a new shopping cart.	Storefront
cartLinesAdd	Adds line items to an existing shopping cart.	Storefront
cartLinesUpdate	Updates line items (e.g., quantity) in an existing shopping cart.	Storefront
cartLinesRemove	Removes line items from an existing shopping cart.	Storefront
getCart	Fetches the details of an existing shopping cart by ID.	Storefront

(Note: Admin API tools like getCustomerById and createProduct are planned for future updates and are currently disabled.)

Inspecting Server Capabilities

To verify the server is built correctly and see the exact capabilities it advertises to MCP clients, you can use the inspect script defined in package.json:

# Ensure you have run 'npm install' and 'npm run build' first
# Run this from the project root directory
npm run inspect

This command executes the compiled server (node dist/server.js) and sends the standard mcp/capabilities request via its standard input. The server processes this request and prints its capabilities (including available tools, resource templates, and server information) as a JSON response to its standard output, which you will see in your terminal. This is useful for debugging and confirming the server's advertised functionality.

Known Issues

• Inline Image Rendering: Clients like Claude Desktop and Cursor currently render ImageContentBlocks (used by getProductById when includeImages is true) in a separate UI block below the main text response, rather than inline. Attempts to use Markdown or HTML image tags within text blocks also failed to render inline in these clients.

Future Plans

• Re-integrate and test Admin API tools (getCustomerById, createProduct, etc.).
• Add configuration options for Admin API credentials (USE_ADMIN_API, ADMIN_ACCESS_TOKEN, ADMIN_VERSION).
• Implement a testing framework.
• Explore resource handling if client support improves.

License

MIT License