llms.txt is a proposed convention. It is a single plain-text file that sits at the root of your site. It gives any AI tool that reads it a clean map of your most useful content, which you have chosen and organized yourself. Think of it as robots.txt's helpful cousin: where robots.txt says what a crawler may not touch, llms.txt says here is what is worth reading, in an order that makes sense.
# Northwind Analytics > Northwind is a product analytics platform for > subscription businesses. ## Docs - [Getting started](/docs/start): install and first event - [Pricing](/pricing): plans and limits - [API reference](/docs/api): endpoints and auth ## Guides - [Reducing churn](/guides/churn): the playbook
A minimal llms.txt: a plain-text map of your most useful pages, in the format AI models are learning to read.
What goes in it
The format is deliberately simple, plain Markdown, so both a model and a human can read it:
- A title and a one-line description of what the site or product is, to help a tool that reads it identify the entity.
- Curated links, grouped under headings, each with a short note on what the page covers.
- Optionally, an llms-full.txt that inlines the full content of those pages, so a model can read them without crawling.
Does it actually do anything yet?
The major engines have adopted it only in limited ways, and no one should promise that it moves rankings today. Google states that llms.txt is not required for its search features, and that it neither improves nor harms your visibility there (Google Search, developers.google.com/search/docs/appearance/ai-features). The first version is cheap to create, and it may help in cases where a tool or a workflow you use can read it. It also makes you do a useful exercise, which is deciding which of your pages actually matter.
How to add one
- Create a plain-text file at /llms.txt.
- Start with a # title and a > one-line description.
- Group your best pages under ## headings, one bullet per page: the page title in square brackets, its full URL in round brackets, then a one-line description.
- Keep it curated, not exhaustive. The point is your best pages, not every page.
llms.txt FAQ
Is llms.txt the same as robots.txt?
No. robots.txt gives crawling rules to the crawlers that cooperate, and it does not keep private content safe from access. llms.txt chooses what is worth reading and describes it. They do different jobs, and both of them live at the root of the site.
Will llms.txt improve my rankings?
There is no evidence it moves rankings today, and you should be wary of anyone who claims otherwise. It is a bet on early adoption. It costs little, and it could pay off if adoption grows. The file and its links still need maintenance so that they stay accurate.
Should I have llms.txt and llms-full.txt?
llms.txt is the map; llms-full.txt inlines the content so a model need not fetch each page. The map is the higher-value, lower-effort of the two, so start there.
Auditaar reports whether llms.txt and llms-full.txt are present, and how well the file has been curated. It also explains why the file matters for your site, rather than treating its absence as an automatic defect. Run an audit, or read how to make your site answer-ready.