Guides 9 min read

How to Create an llms.txt File for Your Website (Step-by-Step Guide)

If you care about how AI assistants and answer engines describe your website, there is a new file worth adding to your root directory: llms.txt. It works a little like robots.txt, except that instead of telling crawlers what to skip, it tells language models what your site is about and which pages matter most. In this guide you will learn exactly how to create an llms.txt file from scratch — the format rules, a copy-paste template, and a free llms.txt generator that assembles the whole file for you in under a minute.

What Is an llms.txt File?

llms.txt is a proposed open standard for making websites easily readable by large language models (LLMs). The file is written in plain Markdown, lives at the root of your domain (for example https://example.com/llms.txt), and follows a deliberately simple structure:

# Example Docs

> Example Docs provides guides and references for building with the Example API.

## Documentation

- [Getting Started](https://example.com/docs/getting-started): Set up your API key and make your first request in five minutes.
- [API Reference](https://example.com/docs/api): Complete endpoint, parameter and error-code reference.

The H1 states your site, product or company name. The blockquote directly beneath it carries a one-sentence summary of what you offer and who it is for. Each ## heading marks a section, and under every section sits a bullet list of links written as - [Page Title](URL): Description. The entire file is designed so a model can ingest it in a single, inexpensive request and immediately understand both the shape of your site and where the substantive content lives.

A companion file, llms-full.txt, is sometimes published alongside it. While llms.txt is the curated index — the table of contents — llms-full.txt concatenates the full text of your key pages for models that want everything in one pass. For most sites, starting with llms.txt alone is the right first move.

Why an llms.txt File Matters for Your Website

The way people find information is changing. A growing share of questions are now answered by AI chat assistants and answer engines that summarize the web instead of listing ten blue links. When one of those systems tries to understand your site, it has two options: guess from whatever fragments it crawled, or read a concise map you wrote yourself. An llms.txt file gives it the second option.

There are three practical reasons to publish one. First, it lets you control the narrative: the one-line blockquote is your elevator pitch, rendered in your words rather than a model’s paraphrase of a random page. Second, it is cheap for models to consume. An LLM reading your homepage has to strip navigation, cookie banners and boilerplate before it finds meaning; a well-formed llms.txt delivers the same understanding in a few hundred tokens. Third, it highlights pages that matter. Deep documentation, comparison pages and conversion-focused landing pages often sit far from the homepage and get under-weighted; a curated link list with one-sentence descriptions puts them back in play.

The precedent is familiar. robots.txt standardized how crawlers treat your site, and sitemap.xml standardized how they discover it. llms.txt aims to standardize how AI models interpret it. Adopting it early costs minutes today and positions your content for the answer-driven web that is clearly arriving.

How to Create an llms.txt File Step by Step

You do not need developer tools, a build pipeline or a plugin. The whole process is writing one small Markdown file. Here is how to do it properly.

Step 1: Audit your site and pick the pages that matter

Start by listing the URLs a first-time visitor (or an AI answering a question about your niche) would genuinely benefit from: your homepage, core product or feature pages, key documentation articles, your best blog posts, pricing, and support or contact pages. Be selective. An llms.txt file that lists four hundred URLs is a sitemap with extra steps; a file that lists fifteen to thirty genuinely useful pages is a map a model can actually use. For each page, note its title and — in your own words — the single question it answers or the single job it does.

Step 2: Write your one-sentence site summary

The blockquote under the H1 is the first part of the file that both readers and models see, so keep it sharp. Write one sentence that says what you offer, for whom, and what makes it distinct. “GeoTools offers free, browser-based map and coordinate utilities for travelers and developers” beats “Welcome to our website” by a mile. Keep it under roughly thirty words, and skip marketing fluff — models respond to specifics, not superlatives.

Step 3: Group your pages into sections

Sort your page list into a handful of logical groups and give each group a ## heading. Common sections include Docs, Blog, Product, Company and FAQ. Sections are not just decoration: they tell the model what kind of content each link leads to, which dramatically improves retrieval quality. Three to six sections is the sweet spot for most sites, and the order matters — put the section you most want cited first.

Step 4: Assemble the file by hand

Open any text editor, start with an H1 containing your site name, add the blockquote summary, then add each section and its links using the pattern below. Save the file as llms.txt with UTF-8 encoding.

# Your Site Name

> One sentence describing what your site offers and for whom.

## Docs

- [Getting Started](https://example.com/docs/getting-started): Set up your account in five minutes.
- [API Reference](https://example.com/docs/api): Every endpoint, parameter and error code.

## Blog

- [Best Practices](https://example.com/blog/best-practices): Field-tested tips from our team.

That is genuinely the whole format. If you can write a Markdown bullet list, you can create an llms.txt file.

Step 5: Or build it with a free llms.txt generator

Doing this by hand is fine the first time, but keeping a hand-built file in sync with your site gets tedious. The faster route is the free LLMS.TXT Generator: enter your site name, URL and one-line description, add your sections and pages with titles, URLs and short descriptions, and the tool produces a standards-compliant file you can copy to the clipboard or download as llms.txt in one click. Relative URLs starting with a slash are automatically joined to your site URL, incomplete rows are skipped, and your draft is saved locally in the browser, so you can update the file whenever your content changes.

Step 6: Publish to your site root and verify

Upload the file to the root of your web server so it is served at https://yourdomain.com/llms.txt, exactly like robots.txt. How you upload depends on your stack: copy the file via SFTP, add it to your static site’s public directory, or create a redirect if your docs live on a subdomain. Then verify it the way a model will — open the URL in a browser, confirm it returns the plain Markdown file with a 200 status, and check that every link in the file resolves without a redirect chain. Some publishers also reference llms.txt from their site footer or docs so human visitors and crawlers can discover it.

llms.txt Best Practices

A few habits separate files that models actually use from files that get ignored:

  • Keep it under about 500 lines. Curated beats comprehensive.
  • Write descriptions a stranger understands — assume the model knows nothing about your brand.
  • One sentence per link. The description tells the model when the page is relevant, not a summary of the page.
  • Use absolute URLs starting with https:// so the file works even when copied out of context.
  • Put the most important section first, and your most important link first within each section.
  • Revisit the file quarterly, and whenever you launch or retire a major page.
  • Optionally publish llms-full.txt alongside it if your documentation is long and frequently cited.

Frequently Asked Questions

What is the difference between llms.txt and robots.txt?

robots.txt gives crawl directives: it tells automated agents which parts of a site they may or may not fetch, and violating it can get a crawler blocked. llms.txt is purely informational: a Markdown overview of your site’s most important content, written to help language models understand and reference it accurately. They solve different problems and most sites that publish an llms.txt file keep both.

Where does the llms.txt file go on my server?

It belongs in the root directory of your domain, so it is served at https://yourdomain.com/llms.txt — the same location as robots.txt. There is no special configuration required; it is just a static text file that your web server returns with a 200 status when requested.

Does Google use llms.txt for SEO rankings?

Google has stated that llms.txt is not a ranking factor in classic web search, and there is no evidence it affects traditional SEO either positively or negatively. Its purpose is different: helping AI assistants and answer engines understand your content. Publishing one is a low-cost way to prepare for AI-driven discovery without touching your existing SEO.

How long should my llms.txt file be?

There is no hard limit, but shorter is better. Most sites do well with fifteen to thirty links organized into three to six sections, which keeps the file comfortably within a single context window read. If your file grows past a few hundred lines, prune it back to the pages you would actually want quoted in an AI answer.

How often should I update my llms.txt file?

Treat it like a sitemap rather than a blog post: update it whenever you launch a significant page, retire one, or change what your site focuses on, and review it at least quarterly. Because the free llms.txt generator stores your draft in the browser, regenerating and re-uploading the file takes about a minute.