Shopify Dev MCP Server

This project implements a Model Context Protocol (MCP) server that interacts with Shopify Dev. This protocol supports various tools to interact with different Shopify APIs. At the moment the following APIs are supported:

• Admin GraphQL API
• Functions
• (Optional) Polaris Web Components

Setup

To run the Shopify MCP server using npx, use the following command:

npx -y @shopify/dev-mcp@latest

Usage with Cursor or Claude Desktop

Add the following configuration. For more information, read the Cursor MCP documentation or the Claude Desktop MCP guide.

{
  "mcpServers": {
    "shopify-dev-mcp": {
      "command": "npx",
      "args": ["-y", "@shopify/dev-mcp@latest"]
    }
  }
}

On Windows, you might need to use this alternative configuration:

{
  "mcpServers": {
    "shopify-dev-mcp": {
      "command": "cmd",
      "args": ["/k", "npx", "-y", "@shopify/dev-mcp@latest"]
    }
  }
}

Disable instrumentation

In order to better understand how to improve the MCP server, this package makes instrumentation calls. In order to disable them you can set the OPT_OUT_INSTRUMENTATION environment variable. In Cursor or Claude Desktop the configuration would look like this:

{
  "mcpServers": {
    "shopify-dev-mcp": {
      "command": "npx",
      "args": ["-y", "@shopify/dev-mcp@latest"],
      "env": {
        "OPT_OUT_INSTRUMENTATION": "true"
      }
    }
  }
}

Opt-in Polaris support (experimental)

If you want Cursor or Claude Desktop to surface Polaris Web Components documentation, include an env block with the POLARIS_UNIFIED flag in your MCP server configuration:

{
  "mcpServers": {
    "shopify-dev-mcp": {
      "command": "npx",
      "args": ["-y", "@shopify/dev-mcp@latest"],
      "env": {
        "POLARIS_UNIFIED": "true"
      }
    }
  }
}

Available tools

This MCP server provides the following tools:

Tool Name	Description
learn_shopify_api	Start here first - Teaches the LLM about supported Shopify APIs and how to use this MCP server's tools to generate valid code blocks for each API. This tool makes a request to shopify.dev to get the most up-to-date instruction for how to best work with the API the user would need to use for their prompt.
search_docs_chunks	Search across all shopify.dev documentation to find relevant chunks matching your query. Useful for getting content from many different documentation categories, but may have incomplete context due to chunking
fetch_full_docs	Retrieve complete documentation for specific paths from shopify.dev. Provides full context without chunking loss, but requires knowing the exact path. Paths are provided via learn_shopify_api
introspect_admin_schema	Access and search the Shopify Admin API GraphQL schema

Tool Usage Guidelines

When to use each documentation tool:

• learn_shopify_api: Always call this first when working with Shopify APIs. It provides essential context about supported APIs and generates a conversation ID for tracking usage across tool calls.

• search_docs_chunks: Use when you need to find relevant information across multiple documentation sections or when you don't know the specific path. This tool searches through chunked content which allows for better token matching within smaller content pieces, but may miss some context from individual pages.

• fetch_full_docs: Use when you need complete documentation for a specific API resource and know the exact path (e.g., /docs/api/admin-rest/resources/product). This provides full context without any information loss from chunking.

Available prompts

This MCP server provides the following prompts:

Prompt Name	Description
shopify_admin_graphql	Help you write GraphQL operations for the Shopify Admin API

Development

The server is built using the MCP SDK and communicates with Shopify Dev.

1. npm install
2. Modify source files
3. Run npm run build to compile or npm run build:watch to watch for changes and compile
4. Run npm run test to run tests
5. Add an MCP server that runs this command: node <absolute_path_of_project>/dist/index.js

License

ISC