How to Generate an llms.txt File for Your Shopify Store

Merchants keep searching for how to generate an llms.txt file for a Shopify store, and the honest answer in mid-2026 has a twist that most of the tutorials floating around have not caught up with. So here is the current, accurate version: what the file is, what a good one contains, and how generating one on Shopify actually works now.

What llms.txt is, in one paragraph

llms.txt is a Markdown file served at your domain root, yourstore.com/llms.txt, that gives AI assistants a curated map of your most important pages with a line of context for each. Where sitemap.xml dumps every URL for a search crawler to chew through, llms.txt is editorial. It says: if you only read a few things about this store, read these, and here is what each one is. The convention emerged across the wider web as a cheap way for language models to understand a site in a single fetch, and the full history and format live in llms.txt, explained.

What belongs in it

Whatever file ends up serving this job for your store, the content brief is the same. A useful llms.txt for a merchant contains:

• A brand description that a stranger could repeat. Two or three sentences: what you sell, who it is for, what makes you different. Assistants lift this language, so write it like you want it quoted.
• Your top collections and a few flagship products, each with a one-line annotation. Not the whole catalog. Curation is the entire value of the file.
• Shipping and returns policies. These are the questions shoppers ask assistants most, and an agent that cannot find your policy tends to omit you rather than guess.
• FAQ or help content that answers the recurring questions in your category.
• Anything that proves trust: your about page, guarantees, certifications that matter in your niche.

And what does not belong: URL dumps, marketing fluff with no facts in it, pages that expire in a week, and anything you would not want an assistant repeating verbatim.

The Shopify twist: llms.txt now mirrors agents.md

Here is the part that changed. In Spring 2026 Shopify made agents.md the canonical AI-discovery file for every store on the platform. Every Shopify store now serves it at the domain root automatically, and /llms.txt and /llms-full.txt mirror it. Three URLs, one file, one place to edit.

This is genuinely good news, because the old path for llms.txt on Shopify was awkward: the platform would not let you drop an arbitrary text file at the domain root, so you needed an app or an app proxy just to serve it. Now the plumbing is free. It also means agents.md is strictly the better file to think about, because beyond the curated index it carries the transaction handshake, the endpoints an agent uses to search your catalog and build a checkout. How the three root files divide the labor is mapped in robots.txt vs llms.txt vs agents.md, and the earlier history of the standalone file is preserved in llms.txt for Shopify.

So on Shopify, "generate an llms.txt" translates to: put the right content in your agents.md, and the mirror does the rest.

How to actually generate it

Step 1: Read what you are serving today. Open yourstore.com/llms.txt and yourstore.com/agents.md in a browser. You will find the platform default: technically present, generically worded, and near-identical to every other store's. That sameness is the problem you are about to fix, and why the default costs you is its own story.

Step 2: Draft the content. Work through the "what belongs" list above against your own store. The exercise that produces the best files is pretending you are briefing a new employee in 300 words: what we sell, who buys it, our best collections, our policies, why people trust us.

Step 3: Publish it through your agents.md. Shopify exposes the file for customization through the theme layer, so your edits ship like any other theme change and the llms.txt mirror updates with it. The mechanics, template, and worked examples are in how to customize your Shopify agents.md.

Step 4: Keep it current. A discovery file is a promise about your store, and promises rot. Collections change, policies change, bestsellers rotate. A file describing your 2025 catalog is worse than the default, because it is confidently wrong. The failure pattern is common enough that we wrote up why agents.md goes stale separately. Put a quarterly review on the calendar, or automate it.

The honest caveats

Two of them, as always. First, no discovery file guarantees you placement in AI answers. Assistants weigh relevance, price, availability, reviews, and signals nobody outside the platforms controls. What the file does is remove ambiguity about what your store is, which is the prerequisite for everything else. Second, the file is one layer of several: it does not fix blocked crawlers, thin product data, or a mis-categorized catalog entry. It is the introduction, not the whole interview.

If you want to know whether the rest of the interview holds up, the free AI-Readiness Checker reads your discovery files, your structured data, and your crawler access the way an agent would, and grades the lot in about a minute. And if you would rather the file be generated and kept fresh automatically as your catalog changes, that is one of the jobs AgentReady does on autopilot.