What is llms.txt, and does your Shopify store need one?

What is an llms.txt file?

It is a short Markdown file, placed at a site’s root, that tells AI agents what the site is and where its most useful pages are. The name comes from LLMs, short for large language models, the kind of models behind AI assistants. Markdown is plain text with simple marks for headings and links.

The idea is a proposal, not a standard. Jeremy Howard published it on September 3, 2024, and the current second version was last updated on August 10, 2026 (the proposal on llmstxt.org). It asks site owners to put a curated overview at the root of the site, with links to clean text versions of key pages.

“Agents are best served by concise, expert-level information gathered in a single, accessible location.”

Jeremy Howard, author of the proposal, llmstxt.org

That is the whole pitch: web pages are built for people, and a tidy file saves an agent from wading through menus and scripts. If a vendor has told you your store is invisible without one, you are right to be skeptical.

What does the file look like?

A short Markdown document in a set order. The proposal says only the first line, an H1 heading with the site’s name, is required. Everything else is optional:

The proposal also suggests clean Markdown versions of key pages, for example page.md next to page.html, so each link leads to text an agent can read in one pass. Those Markdown files hold the detail, and the summary stays small.

Does your Shopify store already have one?

Yes. As of September 2026, Shopify’s help center says every store automatically serves three agent discovery addresses: /agents.md, /llms.txt and /llms-full.txt (Shopify Catalog and product discovery for agentic storefronts). By default, all three return the same content.

Shopify’s developer changelog dated May 28, 2026 says the same: stores have a default agents.md, and the other two paths point to it (Shopify’s developer changelog). The help center adds that you do not need a third-party app to generate these files.

“We’re making every Shopify store agent-ready by default”

Tobi Lütke, CEO, Shopify, Introducing Shopify Agentic Storefronts

So before anyone sells you one, open https://yourstore.com/llms.txt in a browser. You will most likely find it already there.

What is in the llms.txt Shopify serves?

Shopify lists the default content as your store name, store URL, sitemap link, store policies, and discovery endpoints for other store information. Its developer docs say the managed file stays aligned with your store’s commerce setup without extra theme work.

Did Shopify replace the llms file with agents.md?

Not exactly: on Shopify, agents.md is the main file, and the two llms addresses serve the same content. Shopify’s developer docs call agents.md the canonical, agent-facing description of a store (agents.md.liquid reference). It points shopping agents to the store’s commerce endpoints, read-only product and collection URLs, and published policies.

The help center says the older file names exist for compatibility with older AI crawlers that look for them. Shopify decides what each address serves in this order:

In plain terms: change agents.md and all three follow. Add a template for one address and only that address changes.

Why did your own llms file stop working on Shopify?

Because Shopify now serves that address itself, and a URL redirect cannot override a live page. Shopify’s help center says you can redirect only from broken URLs, the ones that show a Page not found or 404 error. If the URL still loads a valid page, the redirect will not work (Creating and managing URL redirects).

That matters if you once uploaded your own version under Content, then Files, and pointed a redirect at it. Once every store began serving /llms.txt by default, the address stopped being broken, so the redirect had nothing to catch. If your curated version vanished overnight, you did not break anything.

The supported way to change the address now is a theme template: agents.md.liquid to change all three files, or llms.txt.liquid to change only that one. If an app manages it for you, open the address in a browser and check that it shows only published products, pages and policies.

Is llms.txt actually used?

Yes, mostly for software documentation. The proposal’s author says these files are used most heavily for developer docs, where coding agents follow them to find API references and tutorials. For a store selling products, the evidence is thin.

Notice what is missing: none of these sources says the file gets a store cited or ranked. That is the claim to question when someone sells one.

Does an llms file help SEO or AI Overviews?

No. Google’s guide to optimizing for generative AI says you do not need new machine readable files, AI text files or Markdown to appear in Google Search, including its AI features, because it does not use them (Google’s guide). It adds that keeping one for other services is completely fine, and will neither harm nor help your rankings.

ChatGPT is similar. OpenAI’s help center lists what makes a site eligible for ChatGPT search: allow its OAI-SearchBot crawler, and make sure your host accepts that crawler’s traffic (Searching the web with ChatGPT). The file is not on that list.

The AI Overviews guide and the generative engine optimization guide cover what does count.

Did llms.txt get this site cited?

No. This site publishes an llms.txt file and names every AI crawler in robots.txt, so I can say what the file did and did not do in its first eight days.

The site went live on September 16, 2026. Its robots.txt names GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot, Google-Extended and others as allowed, and on September 24 eleven AI crawler user agents all fetched a guide with a 200 response. The file was live at the root the whole time.

On September 24, eight days after launch, a ChatGPT-type model with search described the site accurately and cited the /expertise/ page, with utm_source=openai on the link. Grok described the domain accurately.

Perplexity, Gemini and Google’s AI Mode did not have the site at all. AI Mode said the domain was “not an active or official website.”

The pattern matters more than the score. The assistants that found the site used live search and read the pages themselves; the ones that rely on their own index did not have it yet. The file made no difference either way: it is not what got the site cited, and it did not put the site in front of the engines that lacked it.

So the file costs nothing and earns nothing on its own. Mine sits at the root, it is accurate, and I check it when the site changes. That is the right amount of attention for it.

Is the file mandatory?

No. Chrome’s Lighthouse documentation says providing the file is optional at the moment, and its audit marks a missing file as not applicable rather than a failure (Lighthouse documentation). It flags the page only if the server returns an error when asked for the file.

Shopify’s developer docs go further for stores: for most stores, the managed file is all you need (Shopify’s template docs).

How is it different from robots.txt and sitemap.xml?

They do three different jobs. robots.txt sets rules for crawlers, sitemap.xml names the pages you want indexed, and the llms file offers a short guide for AI agents.

Shopify generates the sitemap for you, according to its help center (Finding and submitting your sitemap).

Does robots.txt still work? Yes. It became a standard in 2022 as RFC 9309, and OpenAI’s crawler documentation still uses it to control OAI-SearchBot and GPTBot (Overview of OpenAI crawlers).

Shopify notes that robots.txt rules are advisory, and not all crawlers follow them.

“the REP is a public standard, no one entity can make unilateral changes to it”

Gary Illyes, Search Relations team, Google, Robots Refresher: Future-proof Robots Exclusion Protocol

REP is the Robots Exclusion Protocol behind robots.txt, which search engines and AI crawlers read. The llms file is still a proposal, open for community input. The Shopify sitemap guide covers the sitemap side, and the ChatGPT SEO guide covers which crawlers to allow.

Where do you submit an llms.txt file?

Nowhere. The proposal has no submission step. The file sits at the root of your domain, and agents request it when they need it.

On Shopify it is served at the bare primary domain, with no locale or Markets prefix, according to Shopify’s developer docs. Since Google ignores the file, there is nothing to submit in Search Console either.

The proposal’s site lists a few public directories of these files, but joining one is not required.

When should you customize the file on Shopify?

Rarely. Shopify’s developer docs say that in most cases you should not add a separate template for that address. If you want different content, edit agents.md.liquid instead, and all three files follow it.

The template has limited access: only the request object and the agents object, which supplies values such as the store name, store URL and endpoint links. Before adding a template, decide who will keep it current.

If you do customize it, follow the steps in Shopify’s help center:

1. From your Shopify admin, go to Online Store.
2. For the theme you want to edit, open the menu and choose Edit code.
3. Click Add a new template, then select agents.md, llms.txt or llms-full.txt.
4. Click Create template.
5. Make your edits, then click Save.

A hand-written file that points to last season’s collections is worse than Shopify’s default.

How can you check if your llms.txt file is valid?

Open it and read it. A valid file starts with a single H1 line, and each link should point to a live page. Then run three quick checks:

• Structure. Compare it with the format in the proposal: H1, optional blockquote, then H2 sections of links.
• Server response. Run a Lighthouse report in Chrome. Its audit flags a server error when the file is requested.
• Usefulness. The proposal suggests giving an agent only your file and asking it questions about your content.

If the answers are wrong, fix the pages and store settings the file points to. The file only summarizes. It cannot fix a product page that leaves out the facts.

What should a Shopify store work on instead?

The product data and pages that AI tools actually read. Shopify’s guidance is to focus on complete, well-structured product information in Shopify Catalog and on following SEO best practices, for product data accuracy and discoverability with AI agents.

Its help center is direct about the files, too: Shopify Catalog is the authoritative product data feed to your AI channels, and the agent discovery files do not replace any of its capabilities.

• Titles and descriptions that say what each product is, who it suits and what it is made of.
• Sizes, care, delivery times and policies in plain text on the page.
• Items published to Google through the Google & YouTube channel, so Merchant Center has full data.
• robots.txt settings that let the AI crawlers in.

On one direct to consumer store on Shopify, the best sellers were published through that channel on August 24, 2026. Free listing impressions went from 610 (July 26 to August 24) to 1,567 (August 25 to September 15), according to Merchant Center. The product pages were rewritten in the same weeks, so not all of the change comes from publishing.

The product page optimization guide covers the pages, the Shopify feed guide covers the data, and the ChatGPT shopping guide covers how Shopify items reach ChatGPT.

What does getting llms.txt right look like for your store?

It looks like very little work. Your store serves an accurate llms.txt file by default, you have checked it once, and nobody charged you to create a file Shopify already makes. Your time and resources go into the product data and pages that engines actually use, and the AI search service does that work inside your own accounts.

If you want to see where your store is losing sales first, the free Traffic to Sales Check reads Shopify, Google and your email side by side. It is the first step of The Traffic to Sales Method. Skip it, and the budget goes to files while the pages stay thin.

Does this sound like your store?

If it does, the service that fixes it is AI search for Shopify stores. You can see the before and after on one store in the case study.