What is llms.txt, and does it work?

llms.txt is one of the most talked-about files in AI search, and one of the least proven. Some guides call it essential. Others say it does nothing. This guide sets out what llms.txt is, what the companies behind ChatGPT, Gemini, Claude and Perplexity have actually said about it, what your Shopify store already serves at /llms.txt, and whether it deserves an hour of your time.

What is llms.txt?

llms.txt is a proposed Markdown file, placed at yoursite.com/llms.txt, that gives AI tools a short map of a website: its name, a one-line summary and links to its most useful pages. Jeremy Howard proposed it on llmstxt.org in September 2024 to help AI tools read sites at answer time.

The llmstxt.org proposal describes a plain file with a simple structure:

1. An H1 with the site’s name. This is the only required section.
2. A blockquote with a short summary of the site.
3. Optional notes in plain paragraphs or lists.
4. H2 sections of links, each link followed by a short note on what the page holds. A section called “Optional” holds links a tool can skip when space is short.

The proposal also suggests offering a clean Markdown copy of each page at the same URL with .md added. Its author expected the file to be most useful “for inference”: when an AI tool reads a site to answer a question, rather than when a model is trained.

The idea is reasonable. A product page carries menus, scripts and theme code around the text that matters. A short Markdown map tells a tool where the useful pages are. Whether any AI assistant actually reads that map is a separate question.

Is llms.txt mandatory?

No. llms.txt is a proposal, not a standard, and no search engine or AI assistant requires it. Chrome’s Lighthouse requests the file and flags server errors, but marks a missing file as not applicable, because it is optional for now.

The spec calls itself a proposal, now in its second version. No standards body has adopted it, and no assistant maker lists it as a requirement.

Another signal comes from Google’s Chrome team. Lighthouse added an llms.txt audit to its agentic browsing checks. A site without the file is marked “Not Applicable (N/A)”, because providing it is optional at the moment, the page says. It adds that agents may spend more time crawling a site without one. That is a hint about efficiency, not a statement that any assistant ranks or recommends sites by it.

Does Google use llms.txt?

No, not for Google Search. Google’s optimization guide lists llms.txt among things you can ignore, and says Google Search itself does not use such files. Adding one will neither harm nor help your visibility in Google Search, including AI Overviews and AI Mode.

Google’s optimization guide, last updated July 10, 2026, is unusually direct. Under things you can ignore for Google Search, it lists “LLMS.txt files and other ‘special’ markup.” You do not need new machine-readable files to appear in Google Search, it says, “as Google Search itself doesn’t use them.”

Two details matter for a store owner:

• The statement covers Google Search. That includes its generative AI features, AI Overviews and AI Mode. We found no official Google statement either way about the Gemini app.
• Google may still crawl the file. The guide notes that Google can discover and index many kinds of files. Crawling a file does not mean Google gives it special weight.

Do llms.txt files actually work?

There is no first-party evidence that they do, for shopping answers. OpenAI, Anthropic and Perplexity describe their crawlers without saying they read other sites’ llms.txt files, and Google says its Search ignores them. AI companies publish llms.txt for their own developer docs, which is a different use.

Here is what each party has said in its own documentation, as of October 15, 2026:

The pattern is clear. The companies that run AI assistants publish llms.txt files for their own developer documentation: OpenAI, Anthropic, Perplexity and Google’s Gemini API docs all serve one. That helps developers who paste docs into coding tools. None of these companies says its assistant reads your store’s llms.txt when a shopper asks what to buy.

What the assistants do document is robots.txt. OpenAI, Anthropic and Perplexity all explain which crawlers they run and how robots.txt applies to each. For shopping, OpenAI points merchants to product feeds and says Shopify catalogs are already integrated. Those are the inputs with documented readers.

So does llms.txt work? It may help a coding agent or a future tool read your site. For appearing in shopping answers today, nothing on record says it does. If a guide promises more, ask for the source.

Does Shopify create llms.txt for your store?

Yes, for most online stores. Since May 2026 Shopify serves a default /llms.txt that mirrors the store’s /agents.md, a file of instructions for AI agents and shopping assistants. You can replace it with an llms.txt.liquid template in your theme. Headless storefronts may not have one.

Shopify documented how to customize these files in its developer changelog on May 28, 2026. Its agents.md docs say Shopify manages an agents.md file by default for every store, and that /llms.txt and /llms-full.txt mirror it by default.

Shopify’s default file is not the site summary the llmstxt.org proposal describes. Open yourstore.com/llms.txt and you will likely see agent instructions: how an assistant can browse and buy from the store, with links to its shopping integrations and policies. That is Shopify’s design for agent commerce.

You can change what each path returns with theme templates:

If no template exists for a path, Shopify falls back to your agents.md template, then to its own generated default. One caution: replacing the default /llms.txt with a site summary removes the agent instructions from that path. They stay at /agents.md.

What should be in an llms.txt file?

Your store’s name as the H1, a one-line summary of what you sell and for whom, then your key collections, best products, policies and help pages as links, each with a short factual note. Keep it current and short. It should say what your pages already say.

For a store, a useful file covers:

• Who you are. One sentence: what you sell, to whom, and what makes it different. The same facts your About page states.
• Collections. Your main collections, each with a note on what is in it.
• Products that matter most. Best sellers and hero products, not the whole catalog.
• Policies. Shipping, returns and contact.
• Help content. Size guides, FAQs and how-to pages.

A short file for a small soap brand might look like this:

# Example Soap Co.

> Handmade bar soaps in six scents, sold online in the US. Every bar is listed with its full ingredients.

## Collections
- [Bar soaps](https://example.com/collections/bar-soap): all six scents, 4 oz bars
- [Gift sets](https://example.com/collections/gift-sets): three-bar and six-bar boxes

## Policies
- [Shipping](https://example.com/policies/shipping-policy): where we ship and how long it takes
- [Returns](https://example.com/policies/refund-policy): the 30-day return window and how to start one

Each line points to one page and says in plain words what a reader will find there.

Keep claims out of it. The file should restate facts your pages already show, never add new ones. If your product pages are thin, a good llms.txt will not make up for that, because an assistant that reads your pages will find what is missing.

If you decide to make one, our free generator reads your Shopify store’s public catalog and writes the file in your store’s own words:

On Shopify, the documented way to replace the default is an llms.txt.liquid template in your theme, which changes /llms.txt only. Shopify’s own docs advise that most stores do not need one, and you will have to keep the file current by hand.

Should I add llms.txt to my website?

Add or edit one if it takes you little time, but expect nothing measurable from it. Google says it neither helps nor harms, and no assistant maker says it reads one. Spend your effort first on crawler access, product data and product pages, which have documented readers.

A simple way to decide:

• You run a standard Shopify store. You already serve a default file. Leave it unless you have a reason to change it.
• You want a summary of your store at /llms.txt. Generate one and add it as an llms.txt.liquid template, knowing Shopify says most stores do not need one. It costs an hour and does no harm in Google Search.
• You want to be recommended by AI assistants. llms.txt is not the place to start.

Three things have readers on record. Check that your robots.txt lets the assistants’ search crawlers in. Keep your product data complete, because ChatGPT’s shopping results draw on it. And make each product page say what the product is, who it is for and how it differs. Our guide to AI visibility explains how to measure whether any of it worked.

That measurement is what we built SourceVision for. We ask ChatGPT, Gemini and Perplexity the questions your shoppers ask, many times over, and report how often your products appear, with the range across runs. Then we draft what each product page is missing from facts you confirm, and nothing is written without your approval. See how it works.

SourceVision is in an invite-only beta.

Request an invite

If you are new to how assistants reach store data at all, our guide to Agentic Storefronts covers how a Shopify catalog reaches ChatGPT and other AI channels.