llms.txt Generator: Create an llms.txt File for Your Site

What is llms.txt?

llms.txt is a proposal, first published by Jeremy Howard in September 2024 at llmstxt.org, for a Markdown file at /llms.txt that gives language models a short, curated map of a website. A sitemap lists every page for search engines; llms.txt picks the pages that explain your product and says in a few words what each one covers.

The format has five parts, in this order:

1. An H1 with the name of the site or project, such as # Acme. This is the only required part.
2. A blockquote with a short summary, such as > Acme is invoicing software for freelancers.
3. Optional paragraphs or lists with more detail. No headings here.
4. Zero or more H2 sections, each a list of links written as - [Page name](https://example.com/page): what it covers. The note after the colon is optional.
5. By convention, a section titled ## Optional holds links a tool can skip when it needs a shorter context.

The current version of the proposal also allows a file in a subfolder, such as /docs/llms.txt, which covers the pages under that path. Before you run it, the tool above shows a complete example file.

How the generator builds your draft

• Finding pages: it reads the Sitemap lines in your robots.txt, or tries /sitemap.xml, /sitemap_index.xml and /wp-sitemap.xml. It follows sitemap indexes two levels down and keeps up to 300 pages, reading page sitemaps before post, tag and media sitemaps. With no sitemap, it uses up to 60 links from your home page.
• Skipping noise: files, feeds, search results, tag and author archives, login, cart and account pages are left out, and so are URLs with query strings. On a site with several languages it keeps one version of each page.
• Naming pages: it reads the title and meta description of up to 20 key pages (home, pricing, docs, about, product and other top-level pages first). Other pages are named from their URLs, so check those names.
• Grouping: pages are grouped into H2 sections by the first part of their path, so /docs/... becomes ## Docs. Single pages go under ## Main pages, and privacy, terms, careers and similar pages go under ## Optional.
• Keeping it short: each section is capped (about 20 links for docs, 8 for blog posts, 12 for other sections), because a curated list is more useful than a copy of your sitemap.

It does not run JavaScript, so links that only appear after a page's scripts run are not found, and it does not write llms-full.txt.

Where to upload llms.txt

The file has to load at https://yourdomain.com/llms.txt as plain text, with HTTP status 200.

• Next.js: save the file as public/llms.txt. Next.js serves files in the public folder from the root of your site. To build the file from your content instead, add a Route Handler at app/llms.txt/route.ts that returns the text.
• WordPress: upload llms.txt to your site's root folder (the one that answers at your domain) with your host's file manager or SFTP. Or turn on the llms.txt option in the free Yoast SEO plugin (Yoast SEO, Settings, Site features, AI tools), which builds the file for you and lets you choose the pages it lists.
• Webflow: upload the file in your site's SEO settings, then publish. Webflow serves it at the root of your domain.
• Shopify: some Shopify stores already answer /llms.txt with a default file aimed at shopping agents, so check yours with the llms.txt checker first. To publish your own, use an app with an llms.txt setting, such as Yoast SEO for Shopify (Settings, Advanced, llms.txt).
• Any other host: put the file in the folder your server serves static files from at the domain root, and make sure it is served as text/plain or text/markdown.

Then run the llms.txt checker on your domain. If it reports an HTML page instead of text, your site's catch-all route is answering instead of the file.

How to turn the draft into a good file

• Rewrite the summary. One or two sentences a stranger can use: what the product is, who it is for and what makes it different. The generator copies your home page's meta description, which is often written for clicks, not for clarity.
• Cut hard. Keep the pages that answer questions people ask about your product: pricing, docs, key features, comparisons and your best guides. Delete thin pages and duplicates.
• Add a note to important links. A few words after a colon (: Plans and what each includes) help a model choose the right page.
• Use full URLs. Every example in the proposal uses absolute links, and a model that sees the text on its own cannot resolve /pricing.
• Use the Optional section. Legal pages and older posts can stay, but under ## Optional, so tools with little room can skip them.
• Keep it current. Update the file when you add or retire key pages, or generate it from your content with a route so it never goes stale.

What llms.txt can and cannot do

llms.txt is a proposal, not an official web standard, and it is not a ranking factor in Google. Google's guide to its generative AI features says it is fine to keep an llms.txt file, but Google Search ignores it, so it neither helps nor hurts your visibility in Search.

It is still cheap to add and some tools use it. AI developer docs such as OpenAI's and Anthropic's publish one, as do companies like Stripe, Vercel and GitHub. Chrome's Lighthouse has an experimental Agentic Browsing category with an llms.txt audit, and coding assistants and agents can read the file when someone points them at your site.

It does not control crawling or AI training. That is robots.txt: see which AI bots your robots.txt allows with the AI crawler checker. For the full picture, read the llms.txt guide.