Your Shopify Store Became an AI Agent Endpoint on 5 August

On 5 August 2026 Shopify switched on WebMCP across every Liquid storefront. Your store now hands AI agents a list of things they can do on it, including searching your catalog, editing a cart, and moving to checkout. There was no announcement in your admin, no setting to approve, and no migration to run. Shopify's changelog marks this as requiring no merchant action, which is true. It is also the least interesting part of the story.

What actually shipped

Shopify's developer changelog entry is short and worth reading in full. The line that matters is that this is "live today on every Liquid storefront and on the Hydrogen developer preview" and that "There's nothing to install or configure."

So this is not a feature you adopt. It happened to your store a week ago, and it happened to your competitors at the same time. If you run a Liquid theme, which is nearly everyone, you are already in.

The tools your storefront now exposes

An agent visiting your store can call a defined set of tools rather than guessing its way around your HTML. Shopify lists them:

• Catalog. search_catalog, browse_store, get_product, show_variant
• Cart. get_cart, update_cart, cancel_cart
• Checkout and orders. proceed_to_checkout, manage_orders
• Support. search_shop_policies_and_faqs

Read that list as a shopping trip. An agent can find a product, pick a variant, put it in a cart, and take the shopper to checkout without ever rendering your homepage the way a human sees it. Your theme, your hero banner, your carefully placed trust badges and your popup are not part of that path.

What WebMCP actually is, minus the hype

WebMCP is a proposal that lets a page register tools the browser can hand to an agent. The specification repository describes it as a way to expose "either JavaScript functions or HTML <form> elements" as tools "with natural language descriptions and structured schemas, designed for AI agent ingestion." The interface is document.modelContext, and pages register tools through document.modelContext.registerTool(). Plenty of secondary coverage says navigator.modelContext. That is wrong, and it is a useful signal about how much of the current writing on this subject is being copied rather than checked.

It was authored by engineers from Microsoft and Google and lives under the webmachinelearning organisation. Be clear about status: it is a proposal in an origin trial, not a ratified standard, and Shopify's own entry describes agent support as limited to Chromium-based browsers for now.

That is the honest ceiling on this. Today the addressable agent traffic is small. The reason to care now is that the work it implies takes months and the switch has already been flipped.

Why "no action required" is only half true

Shopify is right that nothing breaks. But "no action required" answers a technical question, not a commercial one. The commercial question is what an agent sees when it calls search_catalog on your store and on three competitors, and which of the four it puts in the cart.

When a human lands on a product page, a thin description is survivable because photography, reviews, layout and price do the persuading. When an agent calls get_product, the structured fields are the entire pitch. Missing material, missing dimensions, vague variant names, no size guidance, empty product type, and specifications buried in an image become the whole reason your product is not selected. Merchants have spent a decade treating product data as an SEO chore. It has quietly become a merchandising surface.

The one file you actually control

There is a lever here, and it predates WebMCP by two months. Since 28 May 2026 you can control what AI agents read about your store by adding Liquid templates:

• templates/agents.md.liquid controls /agents.md, and is the default for the other two paths
• templates/llms.txt.liquid controls /llms.txt only
• templates/llms-full.txt.liquid controls /llms-full.txt only

The fallback order is documented plainly. If no template exists for a path, it falls back to your agents.md template, then to the Shopify-generated default. Which means almost every store in India is currently running Shopify's generic default, because almost nobody has added these templates.

This is the cheapest work on the list. It is a theme file. You can state what you sell, what your shipping and returns terms actually are, which categories matter, and what makes a given product the right pick. An afternoon of work, and it is currently a default file on virtually every competitor you have.

What we do not know yet, and will not pretend to

Two questions matter commercially and neither has a documented answer as of 12 August 2026.

Can you opt out? Shopify's changelog documents no opt-out switch and states there is nothing to configure. We could not find documented merchant-facing controls for disabling the tools. If someone tells you confidently how to turn this off, ask them for the source.

Does agent activity separate cleanly in your reporting? Not documented. A cart built by an agent and a checkout reached through proceed_to_checkout still produce an order, but whether that session attributes the way a human one does is an open question. For stores that just rebuilt their tracking around the 26 August checkout deadline, that is worth watching rather than assuming. Our notes on the analytics worth setting up cover the baseline this sits on.

What is actually worth doing this month

Ordered by return, not by novelty.

• Write an agents.md template. Cheapest, entirely in your control, and currently a default file on nearly every store.
• Fix product data on your top 50 SKUs. Product type, variant names that mean something, materials, dimensions, care, compatibility. Specifications sitting inside an image are invisible to a tool call.
• Move policy content out of images and PDFs. search_shop_policies_and_faqs is one of the exposed tools. Shipping timelines, COD terms and return windows should be text.
• Watch your order sources. Look for sessions that convert without the usual browsing pattern. Do not restructure reporting yet, just start looking.
• Do not rebuild your theme for agents. Anyone selling you an agent-optimised redesign in August 2026 is ahead of the evidence.

What is WebMCP?

WebMCP is a proposed browser capability that lets a website register its own functions as tools an AI agent can call directly, instead of the agent reading the page and simulating clicks. It is defined through document.modelContext, authored by engineers from Microsoft and Google, and currently in an origin trial rather than being a finished standard.

Do I need to install anything for WebMCP on Shopify?

No. Shopify's changelog states there is nothing to install or configure, and that the tools are live on every Liquid storefront as of 5 August 2026. The work worth doing is not installation, it is improving what those tools return about your products.

Can I stop AI agents from using my store?

Not through any documented merchant setting we could find as of 12 August 2026. Shopify's entry describes the tools as live with nothing to configure and does not document an opt-out. Treat any confident claim otherwise as unsourced until Shopify publishes guidance.

How do I customise agents.md on a Shopify store?

Add templates/agents.md.liquid through Online Store, Themes, Edit code. That file controls /agents.md and acts as the default for /llms.txt and /llms-full.txt unless you add their own templates. With no template present, Shopify serves its generated default.

Does this replace SEO for Shopify stores?

No, and treating it that way would be expensive. Human search still drives the overwhelming majority of Indian D2C revenue today. The overlap is large though, because clean structured product data, real text policies and clear category structure serve both. Doing this well is mostly doing your existing job properly.

Is it worth spending money on agent readiness right now?

Spend on the parts that pay off regardless. Better product data improves conversion, search and agent selection at the same time, so it carries no risk if agent traffic stays small. Do not spend on speculative rebuilds, agent-specific themes, or anything priced on the promise of traffic that nobody can currently measure.

Get an agent readiness check

Free check. Email hello@exactwhy.com with subject "Agent readiness" and your store URL. We look at what your /agents.md currently serves, how complete your product data is on your top sellers, and whether your policies are readable as text. We respond within 4 hours with the specific gaps. If your store is already in decent shape, we will say so.

Paid pass, Rs 25,000 to Rs 1 lakh. We write the agents.md template, clean product data across your priority SKUs, move policy content into structured text, and set up the reporting to watch for agent-driven sessions.

Ongoing Shopify development, Rs 40,000 to Rs 1.5 lakh a month. For brands that would rather have platform shifts like this one handled as they land instead of read about six months later.

The reason to move now is not that agents are sending you revenue this month. They are not. It is that the cost of being ready is a theme file and better product descriptions, and both were already worth doing. Our CRO audits have been flagging thin product data for two years. This just added a second reason to fix it.