AI search
llms.txt: what it does and how to add one
The quick answer
llms.txt is a short plain-text note that introduces a website to AI tools and links to its most useful pages. Google says Search ignores it, with no ranking benefit and no penalty. Documentation sites use it for coding assistants. Add one only for a clear purpose and keep it accurate.
On this page
Key takeaways
- llms.txt is a proposed plain-text introduction to a site, written with simple heading and link marks called Markdown.
- Google says Search doesn’t use llms.txt and that publishing one will neither help nor harm your visibility or rankings.
- OpenAI and Perplexity link to indexes of their own documentation and Microsoft describes files for coding assistants; none of those pages promises search citations for publishers.
- llms.txt doesn’t control who can visit your site; that belongs to robots.txt and your website’s security settings.
- If you add a file, put it at the site root, link only pages that work and assign someone to keep it current.
llms.txt is a short note at the front door of your website. It introduces your business to AI tools and points them to your most useful pages. Google says its search doesn’t use the file, so for most local businesses it’s a small extra and not the thing that gets you found. If someone tells you to add one, ask which tool will actually read it.
Here’s an example. Say someone selling website services tells you that you need an llms.txt file to show up in ChatGPT. Before you pay for it, ask which tool reads the file and what that tool does with it. This guide explains the file, checks what the companies behind the tools have said and shows a simple example you can keep up to date.
If your real goal is more enquiries from Google and ChatGPT, the file is a small part of the job. Our AI SEO agency starts with the pages a customer reads and the search access around them. Work on how a business is described in AI answers is called generative engine optimisation (GEO). Our guide on what is GEO covers the whole practice.
Does a small business website need an llms.txt file?
Probably not yet. Google says Search doesn’t use the file and the companies we read make no promise to publishers.2 A local service site gets more from clear service pages, accurate business details and pages that search tools can reach.
Three kinds of site should consider one:
- Documentation and developer sites. Their readers use coding assistants and Microsoft publishes the file for that purpose.7
- Sites with a known use for the file. Identify the tool that will read it and the task it needs to complete before you add one.
- Businesses with a person who will own it. A stale file that lists a closed service is worse than none.
Check your important service pages before adding another file. Make sure Google can reach them and that they clearly answer a customer’s question. Add llms.txt when you have a use for it and someone responsible for keeping it current.
What is llms.txt and what is it for?
An llms.txt file is a plain-text introduction to a website with links to its most useful pages. The “llms” stands for large language models, the kind of AI behind ChatGPT. The file is written with simple heading and link marks, a format called Markdown. Its author proposes it as a guide for software that acts on a person’s behalf, often called an agent.1
The proposal invites community input and isn’t an official standard.1 Each tool decides for itself whether to use it, so find out which tool that is before you treat the file as a requirement.
The proposal sets out the file in this order:
- A single heading with the name of the site or project. This is the only required part.1
- A short summary written as a quote block.1
- Optional notes in plain paragraphs.1
- Sections that each list links, with a short note after each link.1
In the example further down, that turns into your business name, a one-line summary and a short list of your best pages.
Does Google use llms.txt for AI search?
No. Google says Search doesn’t use llms.txt or need special AI text files to include a website. Publishing and maintaining these files for another tool will neither help nor harm your visibility or rankings in Google Search.2
Google’s Search advice doesn’t cover every Google product. Lighthouse, its separate website audit tool, includes a check for software that browses on a person’s behalf. A server failure when retrieving llms.txt triggers that check. A missing file that returns a 404, the “page not found” error, is marked Not Applicable because the file is optional.3
So Google gives two answers. Search ignores the file, while a Chrome audit looks at it for tools that browse on a person’s behalf and doesn’t require it.
What have OpenAI, Anthropic, Perplexity and Microsoft documented?
OpenAI and Perplexity link to indexes of their own documentation, while Microsoft describes files for coding assistants.467 Anthropic’s support article doesn’t mention llms.txt. It explains the permissions for automatic page readers, which sit in a public file called robots.txt.5 None of these pages says that publishing your own file earns search citations.
Scroll across to read every column
| Company | What its page shows | What we did not find |
|---|---|---|
| OpenAI | Its page-reader documentation points to an llms.txt index and offers plain-text page versions by adding .md to the address4 | A statement that a publisher’s llms.txt affects ChatGPT search results. The documented access controls use the permission file called robots.txt4 |
| Anthropic | A support article dated 7 April 2026 explains its automatic page readers and the permission file called robots.txt5 | Any mention of llms.txt in that article |
| Perplexity | Its page-reader documentation points to an llms.txt index6 | A statement that publishers’ files influence Perplexity answers. Its access controls use the permission file called robots.txt6 |
| Microsoft | The Teams SDK publishes llms.txt files, described as plain-text documentation for AI coding assistants7 | Anything about Bing or consumer search inclusion. That page is developer documentation only7 |
| Google Search | A statement that it does not use llms.txt and that the file neither harms nor helps rankings2 | A statement of any benefit |
These are documentation examples, so they don’t show a search benefit for a local service business. The company pages we reviewed give no basis for promising that an llms.txt file will get a plumber, dentist or builder recommended.
Kavera note: In our Sydney trades AI search study, ChatGPT’s 50 completed answers each linked a business or other website. We did not record whether any of those sites had an llms.txt file. So the study says nothing for or against the file. We make no claim either way.
How is llms.txt different from robots.txt and a sitemap?
Picture three things at the front of a property: a welcome note, a gate and a paper map. llms.txt is the welcome note that introduces your content to a tool. robots.txt is the gate that sets your access preferences. A sitemap is the map that lists your page addresses so search engines can find them. They do different jobs and one doesn’t replace another.
Scroll across to read every column
| File | Main job | Who reads it | Does it control access? |
|---|---|---|---|
| llms.txt | A short introduction and selected page links for software helping a user with a task1 | Tools that choose to read it | No |
| robots.txt | Tells search crawlers which URLs they can access on your site8 | Crawlers, which are programs that request pages automatically | Yes, for crawlers that follow it |
| sitemap.xml | Lists page addresses for search engines. The llms.txt proposal notes that the linked documents together may be too large for an AI tool to read at once1 | Search engines | No |
Google also warns that robots.txt is not a way to keep a page out of Google.8 If a page must stay private, protect it with a password. Allowing or blocking AI crawlers is a separate decision with its own rules. Our guide to AI crawlers and robots.txt walks through them company by company.
What should a simple llms.txt example contain?
A simple file has your business name, a short factual summary and links to the pages you most want read. Below is a fictional example for a Sydney tiler. The web addresses are placeholders.
# Harbour Tiling Co
> Harbour Tiling Co is a Sydney tiling business. We tile bathrooms, kitchens and outdoor areas in the Inner West and Eastern Suburbs.
## Services
- [Bathroom tiling](https://www.example.com.au/bathroom-tiling): What we tile, how a job runs and how we quote
- [Outdoor tiling](https://www.example.com.au/outdoor-tiling): Patios and pool surrounds
## Business information
- [Service areas](https://www.example.com.au/service-areas): The suburbs we work in
- [About](https://www.example.com.au/about): Who runs the business
- [Contact](https://www.example.com.au/contact): How to request a quote
The proposal prefers links to clean plain-text page versions written in Markdown. Use those only where they actually exist. Otherwise link the working page rather than inventing a .md address. A section headed ## Optional can hold secondary links an agent may skip.1 Keep services, areas and contact details consistent with your website.
How do you add and check an llms.txt file?
Write the file, place it at the root of your website, then open the address and follow every link. Chrome’s guidance is to create the file, put it in the root directory of your website and follow the specification.3
- Write the file using the structure above. Keep it short and factual and leave out private details.
- Publish it at the root, so the address is yourdomain.com.au/llms.txt. Ask whoever manages your website to upload it.
- Open the exact address and confirm the server returns your text and not an error. Lighthouse flags a page when the server returns an error for the file.3
- Follow every link in the file. Each one should open the page you named, so remove any that do not.
- Re-check after site changes. Update the file whenever a service, area or contact detail changes.
Your hosting logs may show requests for the file. That proves something fetched it. It does not show an AI tool used it and it is not a recommendation or a ranking. To find out whether AI answers mention your business, see our guide to measuring AI visibility. It shows how to test with a fixed question list.
Kavera note: We froze 50 questions covering 10 trades for our Sydney trades study. A fixed question list also helps you record answers before and after adding llms.txt. Keep other site changes in the log, because a difference between two checks doesn’t establish that the file caused it.
What else does the llms.txt proposal say?
The page that sets out the proposal labels the current version v2. It gives 3 September 2024 as the original publication date and 10 August 2026 as its latest modification date.1
A file can cover the whole site or only the pages beneath a particular path. Where several files apply, the proposal tells agents to use the most specific one. It describes reading the file while helping a user answer a question, called inference, rather than mainly collecting material to build a model, called training.1
What should you do first?
Start with the page a customer reads before they call. Once that page is clear and accessible, decide whether an llms.txt file serves a specific tool or documentation task and who will maintain it.
If you want a plan, our AI SEO agency can set the priorities with you. Our SEO audit is free. It reviews your technical setup, keywords, competitors and Google Business Profile, then gives you a prioritised roadmap. Implementation is separate unless retained. Monthly plans start at Suburb SEO, $890/month + GST. See the SEO packages for scope and terms. No SEO package guarantees rankings, leads or revenue.
Good questions
Common questions Clear answers
What is an llms.txt file?
An llms.txt file is a short introduction to your website, written in plain text with simple heading and link marks called Markdown. It names the site, sums up what it does and links to useful pages. You might place it at yourdomain.com.au/llms.txt. It’s an open proposal and not an official standard, so each tool chooses whether to use it.
Is llms.txt still used?
Yes, mainly for documentation. Software companies publish llms.txt files so coding assistants can find the right pages and Google Chrome’s Lighthouse tool has an optional check for the file. Google says Search itself doesn’t use it. We found no company page saying that a business’s file earns recommendations in answer engines.
Does llms.txt really work?
It works as a tidy map for a tool that chooses to read it, such as a coding assistant reading documentation. Google says it neither harms nor helps visibility or rankings in Search. The company pages we read make no promise of citations for publishers, so don’t expect more business from the file alone.
Does my small business website actually need an llms.txt file?
Probably not yet. Start with clear service pages, accurate business details and working search access, because Google says Search ignores llms.txt. Consider a file if a particular tool will use it and someone will keep it accurate. Ask your website provider what adding and maintaining it involves before treating it as a useful extra.
What should I include in an llms.txt file for a local service business?
Use your business name as the heading, then add a one or two sentence summary of what you do and where. Link your main service pages, service area page, about page and contact page with a short note for each. Leave out private pages, prices you don’t publish and anything that’s out of date.
Does llms.txt control how AI uses my content or do I need separate settings?
No. The file doesn’t set access permissions. It only guides tools that choose to read it. Automatic page readers are called crawlers and their permission file is robots.txt, which OpenAI, Anthropic and Perplexity document separately. Review each company’s guidance and your hosting security before changing who can reach a page.
How can I check that my llms.txt file is accessible and whether any AI tools are reading it?
Open the exact address in a browser and confirm it shows your text, then follow every link in it. Your hosting logs show requests for the file, which proves something fetched it and nothing more. They don’t show that an AI tool used the file or recommended your business, so never read a log entry as a result.
Sources
- The /llms.txt file, v2llmstxt.org (Jeremy Howard). Accessed .
- Optimizing your website for generative AI features on Google SearchGoogle Search Central. Accessed .
- llms.txt (Lighthouse agentic browsing)Google Chrome for Developers. Accessed .
- Overview of OpenAI CrawlersOpenAI. Accessed .
- Does Anthropic crawl data from the web and how can site owners block the crawler?Anthropic. Accessed .
- Perplexity CrawlersPerplexity. Accessed .
- llms.txt (Teams SDK developer tools)Microsoft Learn. Accessed .
- Introduction to robots.txtGoogle Search Central. Accessed .
