Free llms.txt Generator for Shopify Stores

Shopify serves llms.txt natively. Here is how to replace it.

Since May 2026 every Shopify storefront serves an agents.md file at /agents.md, and /llms.txt and /llms-full.txt point to the same content by default. Shopify’s developer changelog entry of 28 May 2026, "Customize /llms.txt, /llms-full.txt and /agents.md", describes the customisation path: add templates/llms.txt.liquid in the theme code editor and that template controls /llms.txt on its own; templates/agents.md.liquid controls /agents.md and is the fallback for the other two paths; templates/llms-full.txt.liquid controls /llms-full.txt. If none exists, the store falls back to Shopify’s generated default.

The default is worth reading once. It is a set of agent instructions: it recommends the Shop skill for buy-for-me agents, describes the store’s Universal Commerce Protocol endpoints, and lists a handful of pointers. It is not a catalogue. It does not name your collections, describe your products, or link your policies, which is what an answer engine or an agent researching a purchase would want. The generated file here adds exactly that, and installs through llms.txt.liquid without touching /agents.md.

The older method, uploading a file under Content, Files and adding a URL redirect from /llms.txt, stopped working when Shopify’s own route took precedence. If you set that up in 2025, the redirect is now ignored and the store serves the default instead. Move the content into the template.

How collections, products, policies and pages map into the file

The H1 is the store name from the homepage, and the blockquote summary is the homepage meta description set under Online Store, Preferences. Then four sections. Collections lists up to 25 collections, the ones linked from your header, navigation and footer first, each with its description cut to the first sentence; empty collections and internal ones (upsell, cart, secret, test and the like) are left out and the rest are counted on one line under Optional. This is the section an agent uses to browse, so it comes first. Products lists the first fifty products in sitemap order with the first sentence of each product description, read from /products.json; on a catalogue over fifty the file says how many more exist and points at /collections/all rather than listing them, because a 5,000-line file defeats the purpose. Policies lists whichever of the refund, shipping, privacy and terms pages answer with a 200. Pages lists about, contact, FAQ, shipping and similar pages found in the pages sitemap, with their meta descriptions.

Optional holds one link per blog with its article count, and the remaining-products pointer. Everything is the store’s own text: no product description is rewritten, and a product with an empty description gets a link without a note. If a store has /products.json turned off, the generator reads the same product and collection URLs from the Shopify sitemaps instead, and the file has titles from handles but no notes.

Markets, locales and password pages

Stores using Markets publish a sitemap per locale, with URLs under prefixes like /en-us/ or /fr/. The generator lists the primary locale only and says so in the notes. If you want a file per market, run it once and duplicate the sections with the localised URLs, or serve a different llms.txt.liquid per market through Liquid conditions on request.locale.

A password-protected store returns the password page to every visitor, including every AI crawler, and the generator stops there and tells you so. Headless storefronts built on Hydrogen usually do not expose /products.json or the Shopify sitemaps on the storefront domain; those fall back to the generic sitemap reader like any other site.

What AI shopping answers actually pull from a store

When ChatGPT, Claude or Perplexity recommend products, the pages they cite are the ones their crawlers could fetch and extract a claim from: a product page that states what the product is, who it is for, the price and whether it is in stock, and a collection page whose description says what ties the products together. Structured data helps, and so does a description that reads like a sentence rather than a keyword list. An llms.txt does not change any of that, and as of September 2026 no AI vendor documents reading it. What it does is give any agent that lands on it a curated way in, and it forces the descriptions that answer engines do read to be short and clear, because the file quotes them back at you.

The order of operations, then: make sure the crawlers are not blocked (the crawler checker linked below the result reads your robots.txt and firewall the way they do), write real descriptions for the collections and products you care about, and only then install the file. Run this generator again after you edit descriptions; the file updates with them, and the linter tells you if the live file drifted from the format.

Source: Shopify developer changelog, 28 May 2026: Customize /llms.txt, /llms-full.txt and /agents.md.

Not on Shopify? The general llms.txt generator reads any site’s sitemap.