BridgeToAgent
Get the kit — $49
All posts
Explainer10 min read

The 2026 agent-ready Shopify checklist — beyond Agentic Storefronts

Shopify's Agentic Storefronts terms take effect May 25, 2026. They cover product discovery via AI shopping channels. They do not cover your blog, FAQ, policies, custom landing pages, or any non-product surface an agent reads. Here's the full four-surface checklist for making a Shopify store actually agent-ready — what Shopify covers, what it doesn't, and what you ship yourself.

BridgeToAgentEditorial team

The 2026 agent-ready Shopify checklist — beyond Agentic Storefronts

May 25, 2026 is the date Shopify's Agentic Storefronts terms take effect for qualifying merchants. That date sits exactly one week from the time this post goes up. Every Shopify merchant who got the email is in the same position right now: a notice they half-understand, a feature they need to opt into, and an open question about whether opting in is the whole picture or just the first move.

The honest answer is that it's the first move. Agentic Storefronts is a high-quality piece of infrastructure — it does what it says, it solves a real problem, and Shopify is right that it changes the surface of how AI shopping agents read your catalog. It is also, by design, scoped to your catalog. Everything else an agent reads when answering questions about your store — your blog, your FAQ, your policies, your About page, your campaign landing pages, your customer support content — sits outside the Catalog feed and outside Agentic Storefronts.

That's the gap this checklist closes. Four surfaces, in priority order, with what Shopify handles natively and what you ship yourself for each.

The 30-second version

Surface 1 — Product discovery. Shopify Agentic Storefronts handles this. Opt in via Admin → Catalog → Agentic Storefronts. Pick your channels. Done.

Surface 2 — Everything that isn't a product. Ship the three kit files (agents.json, llms.txt, agent-instructions.md) so agents can find your blog, policies, FAQ, custom landing pages, and brand content. The BridgeToAgent kit generates them from a crawl of your store in under two minutes; install via Shopify Files + URL Redirects per the Shopify install guide.

Surface 3 — Per-product Schema density. Your theme handles this if it's any modern theme (Dawn, Sense, Refresh, most Shopify Plus themes). Verify with Google's Rich Results Test on three random product URLs. If the theme is custom or old, install Yotpo / Judge.me / a Shopify Schema app, or fix the theme.

Surface 4 — WebMCP readiness. Not actionable yet at scale in 2026. Lighthouse audits for it; nobody passes cleanly. Don't spend hours on this right now — track the spec, ship the other three surfaces first.

That's the whole checklist. The rest of this post walks through each surface in enough depth to actually do the work, and surfaces the half-dozen places Shopify merchants reliably miss something.

Surface 1 — Agentic Storefronts (product discovery)

What it does. Agentic Storefronts is the structured product-data pipeline Shopify announced May 12, 2026. It takes your existing catalog, structures and enriches it (categorization, attribute normalization, image classification), and distributes the resulting feed to AI shopping channels where buyers transact — Perplexity Shopping, ChatGPT shopping integrations, and the partner list Shopify keeps current in the Admin panel.

What it covers. The /products/* surface of your store. Product titles, descriptions, variants, prices, inventory state, images. The buyer who asks an AI shopping agent "find me a lightweight cotton summer dress under $80" gets your matching products surfaced as structured options rather than as scraped HTML.

What it does not cover. Anything outside the product catalog. Your blog (/blogs/*), your FAQ pages, support content, custom landing pages, About, Contact, Press, Policies, refund pages, sustainability statements, and any content surface that isn't a Product resource. Custom apps and checkout flows are also outside scope.

Checklist for Surface 1:

  1. Open Shopify Admin → Catalog → Agentic Storefronts (the panel rolls out for qualifying merchants ahead of the May 25 effective date).
  2. Review the updated Additional Terms. They cover what data Shopify shares, with which AI channels, under what controls.
  3. Pick which AI channels your product data is shared with. The default is opt-in per channel — review the partner list and check the boxes you're comfortable with.
  4. Verify your product images are high-quality and your titles + descriptions are clean — the enrichment pipeline does heavy lifting, but garbage in, lower-quality categorization out.
  5. Spot-check the feed using Shopify's preview tooling (in the same panel) for three random products. Make sure the categorization and attribute extraction match what the product actually is.

Time required: 15–30 minutes for a typical SMB store. Larger catalogs may need a product-data clean-up pass before opting in.

Heads-up. Agentic Storefronts is a meaningful policy change as well as a feature. Read the Additional Terms before opting in, especially the sections on data sharing with third-party AI partners and on inventory-state propagation. The terms aren't unusual for a structured-feed product, but they're worth understanding rather than skipping.

Surface 2 — The three kit files (everything else)

This is the surface most Shopify merchants miss after opting into Agentic Storefronts. The reasoning gap goes like this: "I opted into the product feed, AI agents can now read my catalog, I'm done." The catalog is one piece of what agents read. The other pieces — increasingly, the more important pieces for buyers in research mode rather than ready-to-transact mode — are the content surfaces.

A concrete example. A customer asks Claude or ChatGPT: "Is YourBrand a sustainable company? Where are their products manufactured?"

With Agentic Storefronts alone, the agent has structured product data (good — it can answer "what do they sell?") but has no structured access to your About page, your sustainability report, your supply-chain blog post, or your manufacturing transparency page. It scrapes the HTML, gets a partial answer, and fills in the rest with a plausible-sounding guess. That guess might be wrong. It will be confident either way.

With Agentic Storefronts plus the three kit files:

  • agents.json declares the typed actions available on your site (browse_products, search_products, view_policy, contact_support) so agents know how to act, not just what to read.
  • llms.txt is a curated reading list pointing agents at the canonical brand pages — About, sustainability, refund policy, key blog posts — with one-line summaries so the agent picks the right URL the first time.
  • agent-instructions.md is the behavioral runbook — "don't quote prices excluding tax", "refunds policy is at /policies/refund", "the sustainability page is the canonical source for manufacturing questions". This is what stops the model from making up answers when the structured data doesn't cover the question.

The three layers compose. Agentic Storefronts handles the transaction surface; the kit files handle the content and behavior surface. They aren't substitutes for each other.

Checklist for Surface 2:

  1. Generate the three files. The BridgeToAgent kit does this in two minutes from a real crawl of your store ($49 one-time, no subscription). Hand-writing is possible — the formats are intentionally simple — but plan for maintenance every time your content meaningfully changes.
  2. Install via Shopify's Files + URL Redirects workflow. Shopify doesn't let you upload arbitrary files at the root path, so the install uses Content → Files for upload, then Online Store → Navigation → URL Redirects to map /agents.json, /llms.txt, and /agent-instructions.md to the CDN URLs Shopify generates. Full step-by-step at the Shopify install guide.
  3. Add the three <link rel="alternate"> tags to your theme's <head> (edit layout/theme.liquid in Online Store → Themes → Actions → Edit code). These are what tells AI agents the files exist — without them, even agents that look for the convention may miss the files.
  4. Verify each URL serves the file content. Open https://your-store.com/agents.json, /llms.txt, and /agent-instructions.md in browser tabs and confirm each one returns the file (Shopify performs a 301 redirect behind the scenes; agents follow redirects fine).

Time required: 10–15 minutes for the install if you have the files ready.

Heads-up. A common mistake is uploading the three files via Content → Pages instead of Content → Files. Pages render the content as HTML, which corrupts the JSON in agents.json and breaks parsing. The install only works through Files + URL Redirects so the original file content-type is preserved.

Surface 3 — Per-product Schema density

This is the surface Lighthouse's schema-org-density audit scores you on. It checks whether your product pages ship Schema.org JSON-LD blocks at minimum density — at minimum, a Product object with name, description, image, offers.price, offers.priceCurrency, and offers.availability. Aggregate review data (AggregateRating, Review objects) earns additional weight.

The good news for Shopify merchants. Modern Shopify themes — Dawn, Sense, Refresh, Studio, and most Shopify Plus themes shipped since 2024 — emit Product Schema natively on every product page. The theme reads from your product data and outputs a JSON-LD block in <head> automatically. You don't need an app, a code change, or a plugin to pass this audit if your theme is current.

The bad news for some Shopify merchants. Older themes (2022 and earlier), heavily customized themes, and some Shopify Plus themes that were custom-developed don't ship native Schema. If you fall into one of those buckets, your schema-org-density audit fails silently and your product pages look identical in a browser but score differently in Lighthouse and in Google's Rich Results Test.

Checklist for Surface 3:

  1. Test three random product URLs against Google's Rich Results Test. Paste the product URL, run the test, look at the structured-data panel. You want to see at least a Product block with offers, ideally with review data.
  2. If the test shows the Product block present → you're done on this surface. Move on to Surface 4.
  3. If the test shows no structured data, or only minimal WebPage data → your theme isn't emitting Product Schema. Options in order of effort:
    • Switch to a modern Dawn-derived theme if you can do so without redesign work. Cleanest fix; gets you Schema for free on every product page going forward.
    • Install a Schema app like Yotpo SEO, Judge.me, or a dedicated structured-data app from the Shopify App Store. Adds Schema via Liquid snippets without touching the theme code directly.
    • Add Schema manually by editing templates/product.liquid (or templates/product.json on Online Store 2.0 themes). Insert a JSON-LD <script> block that reads from the product's Liquid variables. This is the most fragile option — every theme update needs review — but it's the only path that works for heavily customized stores that can't switch themes or install apps.
  4. After fixing, re-run the Rich Results Test on the same three URLs. Confirm the Product block now appears.

Time required: 5 minutes to test, 1–4 hours to fix depending on which path you take.

Heads-up. Don't add review Schema (AggregateRating) without real reviews to back it. Google penalizes review markup that doesn't match visible content on the page. If you don't have real reviews yet, ship Product + offers Schema and add AggregateRating later when reviews accumulate.

Surface 4 — WebMCP readiness (the forward-looking one)

WebMCP is the emerging spec for declaring agent-callable actions inline on a page — annotating <button>, <form>, and <input> elements so an agent can interact with them as typed tools. Chrome 146 ships a browser-level API, Edge 147 follows, and the spec is stable enough that Lighthouse audits for it.

Where this stands for Shopify merchants today. Nobody passes the webmcp-annotations audit cleanly in 2026. The annotation patterns aren't settled, the theme ecosystem hasn't shipped WebMCP-aware templates, and the audit is low-weighted inside Lighthouse precisely because the spec is still moving. This is the surface where reading the news regularly is more valuable than acting now.

What to do: track the spec, don't sink hours into manually annotating your theme today. The kit ships WebMCP-annotated output as a planned add-on in version 1.3 — the version cuts when the annotation conventions stabilize. Until then, Surface 4 is honest to leave failing on your Lighthouse audit. The other three surfaces are where the actionable wins are.

How to verify the full picture

After working through the three actionable surfaces, run the verification pass:

  1. Run the free BridgeToAgent audit on your store domain. Five-second readiness check, no email needed. Tells you which signals are present and which aren't, with a Shopify-specific detection that accounts for Agentic Storefronts.
  2. Run Chrome Lighthouse Agentic Browsing on your homepage and on a product page. Chrome DevTools → Lighthouse → enable "Agentic Browsing" under Categories → Analyze. Expect to pass six of nine audits if you shipped the kit files; the remaining three (sitemap-discoverable, schema-org-density, webmcp-annotations) depend on Shopify-side defaults and your theme.
  3. Run Cloudflare's isitagentready.com for a cross-check. The Content dimension will move once the kit files are live; the Capabilities dimension partially moves with agents.json. The Bot Access Control dimension is a publisher decision you make separately by editing per-bot rules in your robots.txt.

The two free audits give you the same signal Lighthouse and Cloudflare give themselves, which is the signal that buyers, search engines, and AI agents will increasingly use as the default. If both audits show the kit files present and parseable, the major work is done.

What changes May 25 and what doesn't

The May 25 effective date is for the Agentic Storefronts Additional Terms — which means if you've opted into the panel, the data-sharing arrangements documented in those terms become live for your store. It does not mean Lighthouse, Cloudflare, or third-party AI agents change their behavior on that date. The browser-side audit category Chrome ships keeps scoring your site exactly the way it scored before May 25. The kit files keep being the way you cover everything outside the catalog.

The reason the date matters is that the Agentic Storefronts launch is the single largest piece of public attention on agent-readiness an SMB merchant is likely to see in 2026. The window where customers, partners, and Shopify itself are paying explicit attention to "is my store ready for AI?" runs roughly six weeks on either side of May 25. Sites that shipped the other three surfaces during that window get compounding traffic from the broader conversation — sites that didn't ship them have to do the work later, when the urgency has faded but the underlying need is the same.

Common mistakes to avoid

A short list of the failure modes that show up in real audits of Shopify stores:

  • Opting into Agentic Storefronts and stopping there. The most common one. Catalog discovery is solved; content discovery is still unsolved, and agents in research mode (which is most of them, most of the time) hit the same broken-HTML scraping wall they did before.
  • Uploading the kit files to Content → Pages instead of Content → Files. Pages render as HTML, JSON gets corrupted, parsing breaks. The install only works through Files + URL Redirects.
  • Skipping the <link rel="alternate"> snippet in theme.liquid. Files served but undiscoverable. Some agents probe the convention regardless; many don't. Ship the snippet.
  • Adding AggregateRating Schema without real reviews. Google deindexes the product page when it detects review markup that doesn't match visible review content. Ship Product + offers only until real reviews exist.
  • Manually annotating a theme for WebMCP in 2026. Spec isn't settled. The work doesn't compound. Wait for Kit 1.3 or for theme-side WebMCP support.

What to do this week

  1. Opt into Agentic Storefronts in Admin → Catalog → Agentic Storefronts. Pick your channels. Review the Additional Terms.
  2. Ship the three kit files. Either generate them with the kit ($49, two minutes) or hand-write them and accept the maintenance cost. Install via Files + URL Redirects.
  3. Test three product URLs against Google's Rich Results Test. If Product Schema is missing, fix the theme or install a Schema app.
  4. Run the free readiness audit on your store domain to confirm the kit signals are detected.
  5. Re-run Chrome Lighthouse Agentic Browsing in 14 days after shipping to confirm the audits flipped. If they didn't, the most common cause is the <link rel="alternate"> snippet missing from theme.liquid.

Six weeks of attention. One week to ship four surfaces. Two minutes per generated file. That's the whole opportunity.

Related

Keep reading

All posts →