llms.txt generator: introduce your site to AI assistants
Paste your homepage address to pre-fill the site name, its summary and its main pages, or fill in the fields by hand. The tool writes the file in the format proposed on llmstxt.org, ready to copy or download. No sign-up.
The llms.txt introduces your site to AI assistants. To go the other way and ask Claude about your site, the SEO Cartograph MCP connector gives it access to your crawl: technical audit, pages, topic clusters and on-page fixes, from the free account.
What an llms.txt file contains
A Markdown text file at the root of the site, readable in a few seconds.
A name and a summary
The file opens with the site name as a level one heading, followed by a summary as a blockquote. This is the part an assistant is sure to read: it should say in one sentence what the site does and for whom.
Lists of annotated links
Each section groups pages as a Markdown link followed by a short note. The tool puts your main pages in a first section and secondary pages in the “Optional” section.
An optional section
The proposal reserves this name for what a model can skip when it runs short of room: legal notices, archives, side pages. The rest of the file is meant to fit in a short context.
The limits, stated before you start
llms.txt is a proposal published in September 2024 by Jeremy Howard, with no standard status. To date, no major search engine has committed to reading it, and Google has said it does not use it for search. The file replaces neither robots.txt nor the sitemap, and it does not lift a page in Google. It mostly serves assistants and agents that are given a site address, especially for technical documentation. Reading an address covers the HTML returned by the server, without running JavaScript, and takes at most 40 links from the same domain. It is limited to twenty reads per hour per connection. Typing by hand is unlimited and never leaves your browser.
llms.txt: what it is really for
A file that is simple to produce, whose effect depends on who reads it.
A table of contents built for a model
A web page mixes menus, cookie banners, scripts and footers. The llms.txt gives an assistant the short version: who you are, what you do, and the few pages to read first, with a sentence for each.
Useful first for documentation
Software and technical documentation publishers adopted it first: a developer who hands the documentation address to a coding assistant gets a clean list of the useful pages. For a business website, the benefit is more modest and the cost close to zero.
AI crawlers read raw HTML
Most crawlers that feed AI assistants fetch the page as the server returns it, without running JavaScript. A site whose text only appears after rendering stays invisible to them, llms.txt or not. The Googlebot simulator shows that raw HTML.
The file goes at the root
It must answer at the /llms.txt address of the domain, as plain text encoded in UTF-8. On WordPress, a file uploaded to the root over FTP is enough. Keep it current when an important page changes address.
Creating and publishing an llms.txt, in order
Read the homepage
Paste your site address. The tool takes its name, its meta description and the menu links first.
Sort the pages
Keep as main pages those that describe your offer, your prices, your area and how to reach you. Move the rest down to the secondary section, or delete the line.
Write a note per page
After the second vertical bar, half a sentence saying what the page teaches. That note is what helps an assistant pick the right page.
Publish at the root and check
Download the file, upload it to the root of the site, then open yoursite.com/llms.txt in the browser to check that it displays as plain text.
Beyond the file
Your site, read by Claude
The SEO Cartograph MCP connector gives Claude access to your crawl. You ask in plain English, it reads the data and suggests fixes.
- Technical audit and issues page by page
- Topic clusters and coverage
- On-page fixes to approve
- Covered by your Claude subscription, no AI credits
A whole site
Before introducing your site to an assistant, check that it holds up: error pages, broken links, missing tags, orphan pages.
- 4XX and 5XX statuses, redirects
- Broken internal and external links
- Missing or duplicate titles and meta descriptions
- Badly nested H1 and H2
In the same series: the SERP simulator and the JSON-LD structured data generator. To see what a crawler gets from your pages, the Googlebot simulator. All tools are listed on the free SEO tools page.
Frequently asked questions
What is an llms.txt file?
Does llms.txt improve Google rankings?
robots.txt, the sitemap and links.How is it different from robots.txt?
robots.txt tells crawlers what they may explore. llms.txt summarises the site and points to the pages to read first, without blocking or allowing anything. To keep an AI crawler out, robots.txt is the file that counts.Where should the file go?
https://yoursite.com/llms.txt. It must be served as plain text, encoded in UTF-8. On WordPress, uploading it over FTP or through the host’s file manager is enough.Which pages belong in llms.txt?
What about llms-full.txt?
llms-full.txt holding the full text of their pages, following technical documentation practice. That file is not part of the original proposal. It is useful for documentation, rarely for a business website.