Shopify MCP connects AI assistants to live e-commerce data using the Model Context Protocol – an open standard that lets AI agents search products, manage carts, track orders, and even run store operations through natural language. This guide covers the full Shopify MCP ecosystem, setup instructions for every server type, and how to unlock cross-source e-commerce intelligence with a data warehouse approach.
Shopify powers roughly 26% of all e-commerce websites globally and nearly 29% of the US market. With 4.8 million active storefronts generating over $14.6 billion in sales during Black Friday-Cyber Monday 2025 alone, there is an enormous amount of commerce data flowing through the platform every second. The question for 2026 is no longer whether AI will interact with that data – it is how.
The answer, increasingly, is MCP. Shopify has gone all-in on the Model Context Protocol, shipping native MCP endpoints on every store, open-sourcing a developer AI toolkit, and building an entirely new interactive UI layer for AI-powered shopping. If you run a Shopify store, build Shopify apps, or manage e-commerce data for analytics, understanding Shopify MCP is no longer optional.
What is the Model Context Protocol (MCP)?
MCP is an open standard, originally developed by Anthropic in late 2024, that defines how AI applications connect to external data sources and tools. Think of it as a universal adapter between AI models and the systems they need to interact with. Instead of building custom integrations for every AI assistant, you expose your data through an MCP server, and any MCP-compatible client – Claude, ChatGPT, Cursor, or a custom agent – can connect to it immediately.
How MCP works
The protocol has been adopted by Google, Microsoft, OpenAI, and now Shopify – making it the de facto standard for connecting AI to external tools. For a deeper dive into MCP fundamentals, see our guide on the Model Context Protocol.
Why Shopify is betting on MCP
Shopify’s commitment to MCP reflects a fundamental shift in how commerce works. In 2026, the primary user of many storefronts is no longer a human browsing a screen – it is an AI agent with a purchase mandate. ChatGPT, Perplexity, Google AI Mode, and Microsoft Copilot all have live shopping integrations. Customers are asking AI assistants to find products, compare prices, and complete purchases on their behalf.
The numbers back this up. Shopify’s cumulative gross merchandise volume has surpassed $1.6 trillion, with revenue projected to exceed $12 billion in 2026 according to DemandSage market analysis. To capture the next wave of growth, Shopify needs its merchants’ stores to be discoverable and transactable by AI agents – not just human browsers. MCP is the infrastructure that makes that possible.
Since mid-2025, Shopify has shipped three distinct MCP integrations: a Storefront MCP server on every store, a Customer Accounts MCP server for order management, and a Dev MCP server for developers building Shopify apps with AI assistance. Combined with an active open-source community building admin-level MCP servers, the ecosystem is maturing rapidly.
Shopify’s MCP ecosystem: four server types
Understanding the Shopify MCP landscape starts with knowing which server does what. There are four distinct categories, each serving a different audience and use case.
1. Storefront MCP server (built-in, no auth required)
Every Shopify store automatically exposes a Storefront MCP endpoint at https://{shop}.myshopify.com/api/mcp. No API keys, no custom app setup, no authentication required. This is Shopify’s consumer-facing MCP implementation, designed for AI shopping agents.
The Storefront MCP server provides four core capabilities: product discovery through natural-language search with recommendations, cart management including creating carts and adding or removing items, store information covering policies, shipping, returns, and FAQs, and order management for tracking status and processing returns. As of the Winter 2026 Edition, every Hydrogen storefront on Oxygen became an AI agent endpoint by default, with the /api/mcp route live whether you configured it or not.
2. Customer Accounts MCP server
A companion to the Storefront server, the Customer Accounts MCP server helps authenticated customers interact with their account data through AI. This covers order tracking, return management, and account information access. Together with the Storefront server, it enables end-to-end AI-powered shopping experiences.
3. Dev MCP server (Shopify AI Toolkit)
On April 9, 2026, Shopify open-sourced the Shopify AI Toolkit – a Dev MCP server that connects coding tools like Claude Code and Cursor directly to Shopify’s Admin API documentation and GraphQL schemas. This is not for shopping – it is for building. Developers use it to search Shopify docs, validate GraphQL queries, and execute store operations through natural language.
4. Community and admin MCP servers
The open-source community has built several MCP servers that connect to the Shopify Admin GraphQL API for full store management. The most popular is shopify-mcp by GeLi2001 (169 stars, 31 tools), which provides complete CRUD operations for products, customers, orders, metafields, inventory, and tags. Unlike the Storefront server, these require API authentication and provide admin-level access to store data.
How to set up Shopify Storefront MCP
The Storefront MCP requires zero setup from the merchant side. Every Shopify store already has the endpoint active. To connect an AI agent, point your MCP client to:
https://{your-store}.myshopify.com/api/mcpNo API keys are needed. The server exposes read-only tools for product search, cart operations, and policy queries. This is ideal for building customer-facing AI shopping assistants that help visitors find products and complete purchases. For developers building these assistants, Shopify provides a full tutorial at shopify.dev.
How to set up the community admin MCP server
For store management tasks – creating products, processing orders, managing customers – you need an admin-level MCP server. The shopify-mcp package is the most comprehensive option with 31 tools covering the full Shopify Admin API.
Prerequisites
You need Node.js v18 or higher and a Shopify custom app with the appropriate API scopes. For apps created after January 2026, you will use OAuth client credentials (Client ID and Client Secret) from the Dev Dashboard. For legacy apps, a static access token (shpat_) works as well.
Claude Desktop configuration
Add the following to your Claude Desktop config file (on macOS at ~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"shopify": {
"command": "npx",
"args": [
"shopify-mcp",
"--clientId", "YOUR_CLIENT_ID",
"--clientSecret", "YOUR_CLIENT_SECRET",
"--domain", "your-store.myshopify.com"
]
}
}
}For legacy apps using a static access token, replace the client credentials arguments with "--accessToken", "YOUR_ACCESS_TOKEN".
Claude Code configuration
In your terminal, run a single command to add the MCP server:
claude mcp add shopify -- npx shopify-mcp \
--clientId YOUR_CLIENT_ID \
--clientSecret YOUR_CLIENT_SECRET \
--domain your-store.myshopify.comCursor configuration
Open Cursor Settings, navigate to Tools and MCP, and add a new MCP server with this configuration:
{
"mcpServers": {
"shopify": {
"command": "npx",
"args": ["-y", "shopify-mcp",
"--clientId", "YOUR_CLIENT_ID",
"--clientSecret", "YOUR_CLIENT_SECRET",
"--domain", "your-store.myshopify.com"]
}
}
}Once configured, the AI assistant gains access to all 31 tools for managing products (create, update, delete, manage variants and options), customers (CRUD, merge, address management), orders (lookup, cancel, fulfill, refund), metafields, inventory quantities, and tags.
How to set up Shopify Dev MCP (AI Toolkit)
The Dev MCP is Shopify’s official developer assistant. It does not manage store data directly – instead, it gives your coding AI access to Shopify’s documentation, API schemas, and query validation.
For Claude Code
Run one command:
claude mcp add --transport stdio shopify-dev-mcp -- npx -y @shopify/dev-mcp@latestThis connects Claude Code to all seven tools in the AI Toolkit: searching Shopify docs, exploring the Admin API schema, validating GraphQL queries, and executing store operations. For store operations, you also need to authenticate via shopify auth login with your Partner account.
For Cursor
Add this to your MCP server configuration:
{
"mcpServers": {
"shopify-dev-mcp": {
"command": "npx",
"args": ["-y", "@shopify/dev-mcp@latest"]
}
}
}Important: telemetry and privacy
The Shopify AI Toolkit sends code validated through the toolkit to Shopify’s servers during validation. If this is a concern, opt out by setting OPT_OUT_INSTRUMENTATION=true in your environment variables before running the server.
What you can do with Shopify MCP
The combination of Shopify’s MCP servers unlocks capabilities that span the entire e-commerce workflow. Here is what each audience can accomplish.
Shopify MCP use cases by role
The admin MCP server alone provides 31 tools. Product management covers full CRUD with variant and option handling. Customer management includes lookups, merging duplicate records, and address management. Order management spans from smart lookup (by order name, numeric ID, or full GID) to cancellation, fulfillment, and refund processing. Metafields, inventory quantities, and tags round out the toolkit.
Example workflows: what Shopify MCP looks like in practice
Understanding the tools is one thing – seeing how they work together in real scenarios makes the value concrete. Here are three workflows that show what becomes possible when you connect Shopify to AI through MCP.
Workflow 1: automated inventory monitoring
A store operator connects the admin MCP server to Claude and asks: “Which products have fewer than 10 units in stock and had more than 50 sales last month?” The AI uses the get-products tool with inventory filters, cross-references with recent order data via get-orders, and returns a prioritized reorder list – complete with suggested quantities based on sales velocity. What used to require exporting CSVs from Shopify and running spreadsheet formulas happens in a single conversation.
Workflow 2: customer service escalation
A support agent uses Claude with the admin MCP to handle a complaint. They say: “Customer #6276879810626 says they received the wrong item. Find their most recent order and start a return.” The AI calls get-customer-by-id, then get-customer-orders to find the order, and can initiate a refund with refund-create including the correct restock type. The entire lookup-and-action sequence that takes 5-10 minutes in the admin panel happens in seconds.
Workflow 3: cross-source revenue analysis
A data analyst connects Shopify through Peliqan alongside Google Ads and HubSpot data. They ask the AI: “What is the customer acquisition cost by marketing channel for customers who made repeat purchases?” The Peliqan MCP server runs a SQL query that joins Shopify order data with Google Ads spend and HubSpot attribution – something no Shopify-only MCP server can answer because the data spans three different systems. This is the power of the warehouse-backed approach.
MCP UI: interactive shopping in chat
One of Shopify’s most innovative contributions to the MCP ecosystem is MCP UI – a protocol extension that lets MCP servers return interactive UI components alongside text responses. Instead of an AI assistant describing a product with plain text, the Shopify MCP server can return a carousel of product cards with images, prices, and “Add to cart” buttons that the user can interact with directly in the chat interface.
This transforms the AI shopping experience from conversational text into a visual, actionable commerce interface. When a customer asks “Show me blue sneakers under $100,” the response includes swipeable product cards, size selectors, and one-click purchase buttons – all rendered within the chat window.
Comparing Shopify MCP server types
Limitations and challenges
Known limitations of Shopify MCP in 2026
- Storefront MCP is read-only for catalog data: You cannot create or update products through the Storefront endpoint – it is designed for shopping, not store management
- Admin MCP servers require careful scoping: Granting write access to products, orders, and customers means a misconfigured AI agent could delete inventory or cancel orders
- No cross-source context: Shopify MCP servers only expose Shopify data. If you need to combine Shopify orders with marketing spend from Google Ads or customer data from your CRM, you need a different approach
- Rate limiting applies: All MCP servers interact with the Shopify API underneath, so standard API rate limits (2 requests/second for REST, 50 points/second for GraphQL) still apply
- Token economics at scale: Every AI query against your MCP server consumes tokens. High-traffic stores need caching strategies to manage costs
- UCP transition in progress: Shopify is migrating catalog tools to the Universal Commerce Protocol (UCP), with new endpoints effective May 30, 2026 – expect breaking changes
Troubleshooting common issues
If your Shopify MCP server is not responding or producing errors, work through these checks systematically.
Authentication failures
For client credentials (post-January 2026 apps), verify that your Client ID and Client Secret are from the Dev Dashboard, not the legacy admin panel. The server automatically exchanges credentials for an access token that refreshes every 24 hours. If you see “SHOPIFY_ACCESS_TOKEN environment variable is required,” make sure you are using the shopify-mcp package (not shopify-mcp-server – a different package with different requirements).
Missing tools or incomplete responses
Check that your custom app has the correct API scopes. Product operations require read_products and write_products. Order operations need read_orders and write_orders. Customer tools require read_customers and write_customers. Missing scopes result in silent failures where the tool exists but returns empty or error responses.
Rate limit errors
If you see 429 (Too Many Requests) errors, implement backoff in your MCP client configuration or reduce query frequency. The Shopify GraphQL Admin API allows 50 cost points per second. Complex queries that fetch nested resources (products with variants, images, and collections) consume more points per request. For high-volume stores running frequent AI queries, consider caching frequently accessed data or using a warehouse-based approach that queries a synced copy instead of hitting the live API.
Storefront MCP returns empty product results
If the Storefront MCP endpoint returns no products, check that your products are published to the Online Store sales channel. Products that exist only in draft status or are published only to the POS channel will not appear in Storefront API responses. Also verify that your store has not restricted the Storefront API in its settings.
Debugging in Claude Desktop
Claude Desktop logs MCP server interactions. Check the logs for connection and runtime errors:
tail -n 20 -f ~/Library/Logs/Claude/mcp*.logThe multi-source problem
Shopify MCP servers give AI agents direct access to one thing: Shopify data. That is powerful for store operations and customer-facing shopping, but real e-commerce intelligence requires more. Customer lifetime value calculations need Shopify orders combined with marketing attribution data. Inventory forecasting needs Shopify sales velocity combined with supplier lead times from your ERP. Revenue reporting needs Shopify transactions reconciled with payment processor data and accounting records.
This is the same challenge that appears with every single-source MCP server. We covered this in depth for Snowflake MCP – the AI agent can only work with the data it can see. When that data lives across Shopify, your CRM, your ERP, your ad platforms, and your accounting system, a Shopify-only MCP connection leaves the AI working with incomplete context.
The data warehouse approach: connect Shopify to AI through Peliqan
The solution is to sync Shopify data into a governed data warehouse alongside all your other sources, then expose everything to AI through a single MCP endpoint. This is exactly what Peliqan does.
The architecture works in three steps. First, you connect Shopify to Peliqan – a few clicks, OAuth authentication, and your Shopify orders, customers, products, and inventory start syncing into Peliqan’s built-in Postgres and Trino warehouse. Second, you connect your other data sources. Peliqan has 250+ connectors, so your CRM, ERP, ad platforms, and accounting tools sync into the same warehouse. Third, you install Peliqan’s MCP server (pip install mcp-server-peliqan) and connect Claude, ChatGPT, or any MCP client. The AI agent now queries all your data through one governed endpoint.
Peliqan capabilities for Shopify data
This approach means your AI agent gets a complete picture. Ask “Which marketing channel drives the highest customer lifetime value?” and the AI can join Shopify order data with marketing attribution data from multiple sources – something no Shopify-only MCP server can do.
Real-world example: CIC Hospitality
CIC Hospitality connected 50+ data sources through Peliqan, saving 40+ hours per month on reporting that previously required manual data consolidation. Their board reports now pull from all sources automatically through the same governed warehouse. Read the full case study.
The key difference from direct Shopify MCP is the architecture. With Peliqan, the flow is always: Shopify (and your other sources) sync into Peliqan’s warehouse, the Peliqan MCP server exposes the warehouse to AI agents, and the AI never connects to Shopify directly. This protects your live store from AI-driven API load and rate limits while providing governed, quality-checked data for every query.
Comparing approaches: direct Shopify MCP vs data warehouse MCP
Decision framework
Which Shopify MCP approach fits your needs
- If you want AI-powered shopping on your storefront: Use the built-in Storefront MCP – it is already active on your store, no setup required
- If you need AI to manage store operations: Set up the community admin MCP server (
npx shopify-mcp) with properly scoped API credentials - If you are building Shopify apps with AI coding tools: Install the Dev MCP (
npx @shopify/dev-mcp@latest) for documentation and API schema access - If you need cross-source e-commerce analytics: Connect Shopify to a data warehouse through Peliqan and expose everything via the Peliqan MCP server
- If you need governed data for enterprise AI: The warehouse approach with data lineage and quality monitoring ensures AI agents work with trusted, auditable data
Many organizations end up using multiple approaches simultaneously. The Storefront MCP handles customer-facing shopping. The admin MCP handles internal operations. And a warehouse-backed MCP server handles analytics and cross-source intelligence where data quality and governance matter most.
Getting started
For customer-facing AI shopping, you are already set – the Storefront MCP endpoint is live on every Shopify store. For admin operations, install the community MCP server with npx shopify-mcp and configure your Claude Desktop, Cursor, or Claude Code client with your API credentials.
For cross-source analytics that combine Shopify with your other business data, connect Shopify to Peliqan in under five minutes. Peliqan syncs your store data into a governed warehouse, lets you combine it with 250+ other sources using SQL or low-code Python, and exposes everything to AI agents through the Peliqan MCP server. The setup takes minutes, and your AI agents get a complete picture of your e-commerce business – not just what lives in Shopify.
If you are already using Peliqan for other connectors – like Exact Online MCP for accounting or Snowflake for your data warehouse – adding Shopify takes minutes. All sources land in the same warehouse, so your AI agent can answer questions that span e-commerce, finance, and operations in a single query.
The MCP ecosystem around Shopify is evolving fast. With UCP integration underway, MCP UI maturing, and the open-source community expanding the toolkit weekly, the best time to connect your Shopify data to AI is now.
