Free llms.txt Generator

What is llms.txt, and what can a generated one do for you?

A two-year-old proposal, a Markdown file at the root of your site, and one of the most oversold ideas in search right now. Here is what the file is, how it differs from robots.txt and a sitemap, what is actually known about who requests it, and how to decide whether yours is worth ten minutes.

What the file is

An llms.txt file is a Markdown document served at https://yourdomain.com/llms.txt that gives an AI agent a short, curated map of your site: who you are, in a sentence, and a list of the pages worth reading, each with a line saying what is on it. Jeremy Howard of Answer.AI proposed it in September 2024 and published a second version, titled v2 on llmstxt.org, in August 2026. It is a proposal open for community input, not a standard: no standards body has adopted it, and nothing obliges a search engine or an AI company to read it.

The idea behind it is practical. A web page wraps its information in navigation, scripts and layout, and an agent that has to read a whole site to answer one question wastes most of its effort on that wrapping. A small file that says "start here" is cheap to read. The proposal's own words describe it as used "on demand, when an agent needs information about a topic while assisting a user", and says it has been used mostly for that, rather than for training.

The format, section by section

The spec fixes the order of the sections and very little else. Only the first is required.

• An H1 with the name of the site or project. The one mandatory line.
• A blockquote with a short summary, holding the key information a reader needs to make sense of the rest.
• Optional paragraphs or lists with more detail, as long as they contain no headings.
• Zero or more H2 sections of links, each a Markdown list where every item is [name](url), optionally followed by a colon and a note about the page.

v1 gave a section named Optional a special meaning, as the links a tool could drop when space was short. v2 removed that mechanical meaning; the heading is still a sensible home for secondary links. v2 also added two things this generator does not do for you: files at subpaths, where /docs/llms.txt covers only the pages under /docs/, and clean Markdown copies of pages at the same URL with .md added, announced with rel="alternate" and rel="describedby" links.

An llms.txt example, generated and hand-edited

The clearest way to see what a generator can and cannot do is to put its draft beside a file a person wrote for the same site. Willowdale Equity, a multifamily syndication firm and one of the two sites Verand is tested on, serves a hand-edited file today. Its opening lines, fetched on 27 September 2026:

# Willowdale Equity

> Willowdale Equity is a private real estate investment firm that acquires Class B and C value-add multifamily assets across the southern United States, then syndicates the equity alongside accredited (and occasionally non-accredited) passive investors. [...]

[...] All content reflects the views of Willowdale Equity LLC, which is not a registered investment advisor; nothing on this site constitutes investment, tax, or legal advice.

## About the firm

- [About Willowdale Equity](https://willowdaleequity.com/about/): firm overview, founders, mission, values, and track record summary
- [Editorial Process](https://willowdaleequity.com/editorial-process/): how content is researched, written, reviewed, and updated; author credentials

And the opening of the draft this tool wrote for the same domain, on the same day:

# willowdaleequity.com

> Content and resources from willowdaleequity.com (willowdaleequity.com).

## Key pages

- [Passive Real Estate Investing | Willowdale Equity](https://willowdaleequity.com/): Private real estate investment firm acquiring Class B & C value-add multifamily across the Southeastern U.S. Join the Investor Club for first-look deals.
[the blog index, then:]
- [10 Year Treasury, How it relates to Multifamily CAPs and values | Blog](https://willowdaleequity.com/blog/10-year-treasury-what-it-is-and-how-it-relates-to-multifamily-cap-rates-and-values/): The 10-year Treasury and multifamily cap rates: how the benchmark rate flows through to commercial real estate valuations and what it means for investors.

The draft is correct and it is thin. It knows the site's pages and what each says about itself. It does not know that the firm wants its disclosure read first, which nine guides are its pillars, or that the tenth article in sitemap order is not the tenth most important. And the order it lists pages in is the order it read them, not the order the firm would choose. Those are the edits the card lists after every run, and they are the ten minutes that turn a draft into a file worth serving.

Where it lives

At the root of the host, as /llms.txt, served as plain text with a 200 status. Not in a folder, not behind a redirect to a login page, and not as an HTML page: a theme that answers every unknown path with a styled 404 page and a 200 status looks, to anything requesting the file, like a site that has one full of HTML. Verand's own checks treat an HTML body at that path as no file for that reason. Under v2 a file can also sit at a subpath and describe only the pages beneath it, which suits a site whose documentation or resource library lives in one section.

llms.txt, robots.txt and sitemap.xml

Three plain files at the root, three different jobs. They are often confused because they sit side by side.

So an llms.txt never overrides robots.txt. A page listed in llms.txt but disallowed for a crawler stays disallowed for it, and a crawler you have blocked from the whole site will not be reading the llms.txt either. If you are unsure what your robots.txt allows, the Robots.txt Checker reads it by crawler name. If you are unsure your sitemap is found at all, the Sitemap Checker looks. This generator uses that sitemap to pick its pages, so a missing one changes which 25 pages you get.

Who actually requests it: what is known

This is the question every generator page answers with confidence, and the honest answer is mixed. Here is what the primary sources say, so you can weigh it yourself.

• Google says you do not need one. Its Search Central page on AI features, last updated 10 December 2025, says: "You don't need to create new machine readable files, AI text files, or markup to appear in these features." That page covers AI Overviews and AI Mode. Read it on developers.google.com.
• Google's John Mueller said in June 2025 that no AI system used it. On Bluesky: "FWIW no AI system currently uses llms.txt", adding that the chatbots fetch pages but "none of them fetch the llms.txt file", as reported by Search Engine Roundtable. In June 2026 he was still sceptical, arguing on Google's Search Off the Record podcast that the file cannot help a system decide which sites to surface, per Search Engine Journal.
• The proposal's author reports real use, mostly by coding agents. The v2 text says thousands of sites publish the file, documentation platforms generate it automatically, and coding agents use it to find API references. OpenAI, Anthropic and Google publish llms.txt files for their own developer documentation. Publishing one for your own docs is not the same thing as your assistant requesting one from every site it reads, and we have not found a statement from any of them that their assistants do the latter.
• Chrome's Lighthouse now audits for one, in its agentic browsing checks, which describe the file as "an emerging convention". An audit for a file is a signal that a browser vendor thinks agents may use it, not evidence that any does.

We have no request logs of our own to add to that list. The one source that answers the question for your site is your own server log: search it for requests to /llms.txt and note the user agent on each. If the answer matters to you, look there before and after you publish the file.

Why a regulated firm might publish one anyway

Most generators sell the file as a way into AI answers. That claim is unproven, and this page will not make it. There is a smaller case that holds regardless: an llms.txt is a list of pages you chose. If any agent does read it, it lands on the versions you have reviewed, the ones that carry your disclosures and your credentials, rather than on whatever it happened to crawl. The file costs one upload and grants nothing, so the downside is ten minutes.

For a firm whose content is regulated, that means leading with the pages that establish who is speaking and on what terms. An About page with credentials and licensing. The editorial or review process. Author pages. The disclosure or disclaimer page, and for a registered investment adviser, the page that links Form CRS and Form ADV. Then the pillar guides. Willowdale's file ends with a short note telling agents that offerings are made only by private placement memorandum to verified accredited investors, which is the kind of line only the firm can write.

One thing to know about this generator in particular: any URL containing privacy, terms, cookie, legal or disclaimer is listed under its own ## Disclosures heading at the end of the file, never mixed into the content lists and never dropped. On a regulated site that section may matter more than any other, so read it before you upload. It only holds the disclosure pages among the 25 the tool read; if yours sits deeper in the site, add it there by hand, and consider moving it to the top.

Platforms that write it for you

llmstxt.org lists several tools that generate the file on their own, including the Yoast SEO and AIOSEO plugins for WordPress, Wix for every Wix site, and the documentation platforms Mintlify and GitBook. If your site runs on one of them, you may already serve a generated file without having written it. The card checks for an existing /llms.txt on every run for that reason; compare before you replace one.

Common mistakes

• Shipping the placeholder summary. "Content and resources from yourdomain.com" says nothing. The blockquote is the one line the spec asks you to write; write it about the firm, in plain words.
• Listing every URL. That is the sitemap's job. A file with 400 links is a sitemap in Markdown, and the proposal is explicit that the file should stay small.
• Leaving descriptions cut mid-sentence. Meta descriptions are written for search results and often run long. This generator stops them at 160 characters and ends each cut one with an ellipsis, so they are easy to find; finish the sentence or shorten it.
• Serving HTML at /llms.txt. A theme's error page returned with status 200. Open the URL in a browser: you should see raw Markdown, not your site's design.
• Listing pages robots.txt blocks. Advisory never beats access. Check the two files agree.
• Letting it go stale. A file generated once and forgotten keeps pointing at renamed and retired pages. Regenerate it when the site's structure changes.
• Treating it as a ranking lever. Google says it is not needed for its AI features, and nobody has shown that it moves a citation. Publish it as a courtesy and a curation, not as a strategy.

How to generate llms.txt and put it live

Generate the draft above, make the edits the card lists, and save the result as a UTF-8 text file named exactly llms.txt. On WordPress, either upload it to the web root with your host's file manager or SFTP, the folder that holds wp-admin and wp-content, or let an SEO plugin serve its own and paste your edits there. On a static site, put the file in the folder that becomes the root of the build, usually public/ or static/, and deploy. Then open https://yourdomain.com/llms.txt in a browser and confirm you see plain text. The llms.txt file needs no registration anywhere and no line in robots.txt.