lunalink.ai
← lunalink.ai blog
September 28, 2026 · The lunalink.ai team

Can you check whether AI can read your Shopify store without installing an app?

A vibrant arrangement of assorted canned goods on wooden shelves in a store.
Photo by Warren Yip on Pexels

You don't need to install an app to find out whether an AI can read the public parts of your Shopify store.

You do need to separate a few different questions that often get mixed together.

Can an AI see your product pages? Can it search your catalogue? Can it answer a shopper's question? Can it see inventory, orders, or customer data?

Those are not the same thing. The first three can involve public storefront data. The last one needs your approval.

Start with what anyone can read

Open your store in a private browser window. Don't log in to Shopify. Then open three pages:

  • A product page
  • A collection page
  • A policy page, such as shipping or returns

These are the pages a person, crawler, or public-facing shopping tool can reach without being granted account access.

Look for the basics on the product page:

  • Product name
  • Price
  • Variant names and options
  • Product description
  • Availability details
  • Shipping or return information where it matters

Say your store sells linen shirts. A shopper may ask an AI for a washable linen shirt with a relaxed fit. If the product page only says “premium linen,” the public page has not answered the question. The fit and care details need to be written on the page, not held only in a staff member's head.

This check takes a minute, and it is useful even if you never use an AI tool yourself. A clear product page helps shoppers read it too.

Shopify already provides public AI-facing routes

Shopify says WebMCP tools are available on every Liquid storefront. They require no installation or configuration. Compatible agents can use them to search the catalogue, manage carts, and navigate the store.

Shopify currently limits agent support for WebMCP to Chromium-based browsers. So a result from one AI product is not proof that every AI product can use the same route.

Shopify also gives each store a Storefront MCP endpoint:

`https://your-store.myshopify.com/api/mcp`

Replace `your-store` with your permanent `myshopify.com` store name, not your custom domain.

Shopify says requests to this endpoint do not require authentication. It can expose store-policy and other storefront functionality, though a store may restrict access.

There is a useful distinction here. Public access does not mean every AI can do every task. It means Shopify has provided a way for compatible agents to interact with public storefront information without you installing a separate app first.

Check the endpoint, but don't overread the result

You can paste your store's `/api/mcp` address into a browser.

If you get a response, you have confirmed that the route exists and is publicly reachable from where you are checking. If you see an error or restriction, that is worth investigating.

But this is only a route check. It does not prove that an AI has understood your catalogue correctly.

Shopify's current documentation separates catalogue access into a UCP MCP endpoint:

`https://your-store.myshopify.com/api/ucp/mcp`

That endpoint includes catalogue tools such as `search_catalog`, `lookup_catalog`, and `get_product`. Shopify says requests require an agent profile.

This means you should not expect a browser visit to tell you the whole story about catalogue search. It can tell you whether a public endpoint is present. It cannot replace testing with a compatible agent.

Shopify has also removed the older catalogue and cart tools from `/api/mcp` and directs developers to the UCP endpoint for those jobs. That is a good reason not to rely on an old setup guide or a screenshot from last year.

Your public catalogue still needs checking

A public route is only as useful as the information behind it.

We ran into a version of this problem while building our own public scanner. We decided it would read only public URLs: `robots.txt`, the catalogue endpoint, product markup, and policy pages. No install. No login. A merchant should not have to trust a stranger's app just to learn the basics.

Then we found a quiet limitation in Shopify's product catalogue endpoint. `/products.json` returns at most 250 products per page.

A catalogue read can look complete when it is not.

We had to handle pagination in our own `llms.txt` builder so it would continue through the catalogue rather than stopping after the first page. The lesson applies to any manual check too: don't use a sample of your store and call it the whole store.

If your store has more than 250 products, open this URL:

`https://your-store.myshopify.com/products.json?limit=250`

Then check whether the response includes a next page reference. If it does, your catalogue needs more than one request to read fully.

You do not need to become an API developer to learn from this. Just avoid assuming that a tool has seen every product because it found some products.

Product discovery may already be on

For eligible stores, Shopify says products are available to ChatGPT product discovery through Shopify Catalog by default. No merchant action is required.

Eligibility currently requires selling to customers in the United States. Shopify says there are no fees beyond standard payment-processing fees.

You can check this in Shopify admin:

  • Go to Sales channels
  • Open Agentic
  • Review the ChatGPT catalogue setting

You can remove ChatGPT's Shopify Catalog access there. Shopify says that removal can take up to seven days.

It also warns that removing catalogue access does not make a product disappear from the wider web. Products may still be found through crawling or other external listings.

That is the practical point: catalogue participation and web visibility are related, but they are not the same control.

Private data needs a different kind of connection

No public storefront check can tell an AI your order history, customer records, or private inventory details.

Shopify's ChatGPT plugin can access products, collections, inventory, orders, customers, discounts, and analytics after a merchant connects an owned store, signs in, and approves access.

That is permissioned access, not public reading.

Its order access is read-only. Shopify says it can list up to 50 recent orders at a time and open an individual order with its line items, fulfillment status, shipping address, and tracking. It cannot edit, refund, or cancel orders.

This is a useful boundary to keep in mind when someone says an AI can “read your Shopify store.” Ask which store data they mean, and what permission it has.

Install an app when the job is customer chat

If you want an AI chat tool inside your storefront for customers, that is a separate job.

Shopify says Shopify Inbox must be installed from the Shopify App Store. Its Inbox agent can use product catalogue data, policies, Knowledge Base facts and files, and published storefront content such as pages and blog posts.

In that case, installing an app is the answer because you are adding a customer-facing feature. It is not required just to check what a public AI can already reach.

A sensible first check

Before adding anything to your store, do this:

  • Open one product page while logged out.
  • Check that its key buying details are written as text.
  • Open your shipping and return policies.
  • Visit your `/api/mcp` endpoint.
  • In Shopify admin, review Sales channels → Agentic if you sell to U.S. customers.

That will not answer every question about every AI. It will show you the line between your public storefront, Shopify's agent routes, and the data that needs your approval.

If the public pages are thin or inconsistent, fix those first. They are the part you control without installing anything.

Sources

shopifyai searchstorefrontcatalog

See how AI assistants actually read your store — a free 30-second scan, no install and no signup.

Check my store