Shopify launches · Jun 30, 2026 · 3 min read
Shopify's Storefront MCP Endpoint, Explained for Merchants
Every Shopify store now speaks MCP — AI agents can search your catalog, read product details, and quote your policies over a standard protocol. What the endpoint exposes, what it can't answer, and how to prepare.
MCP — the Model Context Protocol — is the plug standard AI agents use to talk to external systems. In 2026 Shopify put an MCP endpoint on storefronts, which means an AI assistant no longer has to scrape your website to answer questions about it. It can ask your store directly.
This is the single most consequential piece of Shopify's agentic stack, and most merchants don't know it exists. Here's what it serves, in plain terms.
What an agent can ask your store
The storefront endpoint exposes a small set of tools. The names vary by version, but the capabilities are stable:
- Catalog search — natural-language product search over your live catalog, returning matching products with prices and availability.
- Product details — full data for a specific product: variants, options, images, descriptions.
- Policies and FAQ — your shipping, returns, and store policies, served as answerable content rather than a page to parse.
- Cart actions — on supported setups, building a cart the shopper can complete.
In other words: discovery, detail, trust, and transaction. An assistant with those four can walk a shopper from "do they have something like this?" to a checkout link without ever rendering your theme.
What it can't answer
The MCP endpoint returns what's in your data, verbatim. It does not reason about your products, reconcile contradictions, or fill gaps. Ask it "will this jacket work for a Chicago winter?" and it can only return whatever your product description happens to say.
The practical failure modes we see: search summaries that flatten variants (an agent recommends a colorway that's out of stock), attributes that live in images (fit notes on a size-chart JPEG are invisible), and policy answers that go stale the moment your policy page and your Knowledge Base disagree.
The endpoint is honest plumbing. If the answers are wrong, the data was wrong first — and now the wrong answer is being served programmatically, at scale, in a channel you don't watch.
How to prepare your store
Three moves cover most of it. Write product data as if a very literal intern is reading it aloud to a customer — because that is functionally what's happening. Move buying-decision facts (fit, materials, compatibility, dimensions) out of images and into text and structured fields. And keep policies in one canonical place so every surface serves the same answer.
Then verify from the agent's side. Kinect probes stores through the same channels agents use — MCP, feeds, crawl — and shows you the answers shoppers are getting before you hear about it from a support ticket. The same product intelligence powers the AI sales rep we put on your storefront, so fixing legibility once improves both channels.
Frequently asked questions
Is the storefront MCP endpoint on for every Shopify store?
It's platform infrastructure rolled out broadly through 2026 — most stores have it live without having configured anything. Check your store's status rather than assuming either way.
Can I see what agents are asking my store over MCP?
Shopify gives you very little visibility into agent traffic today. That reporting gap is one of the main reasons brands run a separate agent-readiness layer — you want to know what's being asked and answered in your name.
Does MCP replace SEO and product feeds?
No — assistants blend live protocol access with crawled pages and shopping feeds. You want all three telling the same story. Contradictions between them are where hallucinated answers come from.
Related reading
See Kinect on your store
An AI sales rep on your storefront and an agent-ready store behind it — trained on your catalog and policies, live the same day, measured against your own baseline.