What is llms.txt and Does Your Store Need One?

llms.txt is a proposed Markdown file that gives AI tools a short guide to your site and links to useful pages. Your store can add one as a small maintenance task, but it isn’t a requirement for appearing in AI search. Google Search explicitly ignores it, including for its generative AI features.

If your SEO agency recommends one, ask what it will contain and which tool will use it. A concise file can help an agent that reads it find your returns policy or a buying guide. It doesn’t make your products eligible for AI shopping results by itself.

What an llms.txt File Does

An llms.txt file summarises a website and points language models towards selected pages. Markdown is plain text with simple formatting for headings and links. The usual whole-site address is /llms.txt, directly after your domain name.

Jeremy Howard proposed the format. His llms.txt proposal is now version 2 and also allows files beneath a site’s root, such as /docs/llms.txt. A file at that path describes the pages beneath it.

The proposal begins with the site’s name as a heading. A short description can follow, then sections containing links and brief notes. The reader gets a route into the material without loading every page at once. The proposal recommends linking to clean Markdown versions where those exist.

For a store, that could mean directing an assistant to delivery terms before it answers a shipping question. The file gives it somewhere to look. The policy page remains the source of the answer.

Which AI Tools Read It

AI companies publish llms.txt files for their own documentation, but that doesn’t establish that their search crawlers automatically read every store’s file. Separate a documented instruction to fetch a file from a claim that publishing one improves visibility.

Anthropic’s Claude Code documentation and Perplexity’s developer documentation both point readers to their documentation index. Each says, “Use this file to discover all available pages before exploring further.” That is an explicit use for navigating those documentation sites.

OpenAI publishes llms.txt for its developer documentation too. Its crawler guidance describes OAI-SearchBot for search and GPTBot for potential training content. That guidance does not confirm automatic use of retailers’ llms.txt files by either crawler.

Google gives a firmer answer for Search. Its generative AI search guidance says, “Google Search ignores them.” Publishing llms.txt neither helps nor harms Google Search visibility or rankings. This statement covers Search, including its generative AI capabilities, rather than every Google product.

Neither Anthropic’s index instruction nor Perplexity’s crawler guidance confirms automatic discovery of retailers’ llms.txt files or their use as a search signal. Treat claims of guaranteed ChatGPT, Claude or Perplexity citations as unsupported by these statements.

llms.txt Compared With robots.txt and Sitemaps

robots.txt gives compliant crawlers instructions about which paths they can crawl. A sitemap lists URLs for discovery. llms.txt summarises selected content and explains where an agent can find more detail.

Adding a page to llms.txt does not allow a blocked crawler to visit it. Leaving a page out does not block access. The file has no crawler access-control role, and it does not replace server security or authentication.

For example, a store might include its returns page in llms.txt and its public product URLs in a sitemap. Its robots.txt handles crawl instructions separately. For the settings that control bot visits, use the guide to AI crawlers.

Keep the broader work focused on readable public pages and accurate product information. An llms.txt file is one optional aid within generative engine optimisation, not a substitute for those pages.

What to Put in an llms.txt File for a Store

Put your store description and a small selection of public policy pages, category pages and buying guides in llms.txt. Choose pages that help an assistant understand what you sell or answer a recurring customer question.

This hypothetical example is for Banksia Home, a fictional Australian homewares store. The example.com links illustrate the format and must be replaced with your store’s working URLs.

# Banksia Home

> An Australian online store selling homewares and lighting.

## Store policies
- [Delivery](https://example.com/delivery/): Delivery areas and shipping terms.
- [Returns](https://example.com/returns/): Return eligibility and the process.

## Categories
- [Table lamps](https://example.com/table-lamps/): Browse the lighting range.
- [Bed linen](https://example.com/bed-linen/): Browse bedding by material and size.

## Buying guides
- [Choosing a table lamp](https://example.com/guides/table-lamps/): Light output and placement advice.

Don’t list every product. A large catalogue turns the file into another URL inventory and adds constant upkeep as stock changes. Category pages give an agent a stable starting point, while product pages and feeds carry product details.

Avoid copying changing prices or stock counts into the summary. That creates another place where old information can survive. Keep AI-ready product data accurate at its source, and link to the pages that explain it.

Using an llms.txt Generator

An llms.txt generator turns selected site content into a Markdown index, usually with page titles, links and descriptions. Some also create an llms-full.txt file containing page content. Review the output before publishing it because an automatic inventory can include far more than your store needs.

On Shopify, the platform manages /llms.txt by mirroring /agents.md, its main agent-discovery document. It serves the file at the bare primary domain, without a language or market subfolder. Most stores can use that managed default.

Shopify’s theme documentation says to customise agents.md.liquid when the shared agent-discovery content needs changing. A separate llms.txt.liquid template overrides only /llms.txt when it needs different content. You then take responsibility for keeping that text current. Inspect the existing file before paying for an app that creates another version.

On WordPress, the Website LLMs.txt plugin generates a file from public content and has settings for selecting what to include. Its plugin documentation explains the setup. Installing a generator establishes that you can publish a file, not that an AI search provider will use it.

After generation, remove irrelevant links and edit descriptions that misstate a policy or category. Open /llms.txt on your public domain and confirm it returns text instead of an error or a normal webpage. Follow its links and check that they reach the intended public pages.

Keep the file if it stays useful with little maintenance. Give fixes to your catalogue and customer-facing policies priority over expanding an index whose readers have no confirmed need for more detail.