Free tool · Websites & AI Crawlers

llms.txt Generator

Fill in a form and get a well-formed llms.txt file, checked against the format on llmstxt.org. Preview it as you type, import the file you already have, then copy or download it.

Runs entirely in your browser. Nothing you enter is uploaded; your draft is saved only on this device.

The llms.txt generator

Jump to the preview

About your site

The name is the only required part. The summary and details give an agent the context it would otherwise have to guess.

Written as the H1, the first line of the file.

Written as a blockquote. Keep it to a few sentences.

Paragraphs and lists are fine. Headings are not allowed here; use a section below instead.

Link sections

Each section becomes an H2 heading followed by a list of links. Group links by what a reader is trying to do, and say in a few words what is behind each one.

No sections yet. Add one, such as “Docs” or “Start here”.

What “Optional” means. A section named exactly Optionalis the proposal’s way to mark secondary links: an agent that needs a shorter context can skip them. Put background, migration notes, and anything nobody needs for a first answer there. It is a convention that readers may honour, not a switch that removes the links.

Import an existing llms.txt

The text is read in your browser. Importing replaces what is in the form; Undo brings it back.

Checks

Nothing to check yet. Fill in a name, or load the example.

Preview: llms.txt

Your file appears here as you type. Fill in a name, or load the example to see how the parts fit together.

1 line · 2 bytes (UTF-8)

Your draft is saved in this browser only. Nothing you enter is uploaded.

How to use it

  1. Name. Enter your site or project name. It becomes the H1, the only part the format requires.
  2. Summary. Write two or three sentences on what the site is, who it is for, and what a reader is likely to get wrong.
  3. Details. Add what is needed before the links: base URLs, versions, deprecations. Paragraphs and lists only.
  4. Sections and links.Group links by task, such as “Start here” or “API reference”. Give each a title, a full https:// address, and a short note on what the page covers.
  5. Optional. Move background and rarely needed pages into a section named Optional.
  6. Check, then copy or download. Fix every error, read the warnings, and serve the file from /llms.txt.

Already have a file? Paste it under “Import an existing llms.txt” and the tool lists anything that does not fit the format. “Load example” fills the form with a made-up product.

How it works

The tool writes the structure from the llms.txt proposal, which Jeremy Howard published on the Answer.AI blog in September 2024. We checked it against version 2, last modified in August 2026. A file has, in order, an H1 with the name (the only required part), a blockquote summary, any Markdown that is not a heading, and H2 sections that each hold a list of - [name](url) entries, optionally followed by a colon and notes.

A section named Optional marks secondary links that an agent can skip when it needs a shorter context. Version 1 also gave it a mechanical meaning for a context-expansion tool; version 2 dropped that tool, so treat Optional as a signal, not a switch.

Errors break the format or leave an entry unusable: no name, a heading or an unclosed code block in the details, a link without a title or an absolute http(s) URL, an empty section, duplicate section names. Warnings are our own judgment, not rules from the proposal: no summary, a summary over 600 characters, http instead of https, notes over 200 characters, repeated URLs, and reserved example domains. So simple parsers can read each entry, the tool writes square brackets in titles as parentheses and percent-encodes spaces, parentheses, and backslashes in URLs, and warns you when it does.

What this tool does not do

  • It does not fetch your site or sitemap. It makes no network requests, so it cannot list your pages for you.
  • It does not check that links resolve or that the pages say what the notes claim.
  • It does not create the Markdown versions of pages that the links should point to.
  • It does not host the file. You publish it at /llms.txt yourself.
  • It does not control which crawlers may read your content. That is robots.txt, not llms.txt.

Where to host it

At the root of your site: https://yoursite.example/llms.txt. Version 2 also defines files in subpaths such as /docs/llms.txt. A file covers the URLs under its path, and the most specific one applies, which helps when you control only a folder.

Serve plain UTF-8 text with status 200. The proposal sets no content type; vercel.com and developers.openai.com send text/plain, and docs.stripe.com sends text/markdown. Open the URL afterwards and check that the first line is your H1: a catch-all route can answer a missing file with your home page and a 200.

Keeping it in sync with your sitemap

A sitemap lists every page you want search engines to find, up to 50,000 URLs per file. The llms.txt proposal says a sitemap is no substitute: sitemaps often lack LLM-readable versions of pages, leave out external links, and cover more than fits in a context window. Choose from the sitemap; do not copy it.

  • When a page is renamed, moved, or deleted, search the file for its URL.
  • Each quarter (our suggestion), check that every internal link is in the sitemap and returns 200 without a redirect.

Tips

  • Spend the summary on what an agent would get wrong: units, version names, authentication, deprecated endpoints.
  • Give every link a note. A title says what the page is; a note says why to open it.
  • Where you can, link to Markdown versions of pages (page.md or page.html.md), which the proposal recommends.
  • Treat the file as public code: keep it in version control and limit who can edit it, since an agent that trusts it can be misled by an altered link. Ahrefs gives the same advice.

What AI providers do with it

Support is limited and still changing. Google’s guide to generative AI features in Search (updated 10 July 2026) says Google Search does not use these files and that creating one will “neither harm nor help” visibility there. Ahrefs looked at 137,210 domains in its own web analytics that received traffic in May 2026: 28% served a /llms.txt, and 97% of those received no requests that month. The proposal itself says these files are used most heavily for software documentation, where coding agents follow them to find API references and tutorials. The details are in What is llms.txt?. For access rules, see the guide to AI crawlers and robots.txt.

Frequently asked questions

Is anything I type sent to a server?

No. Everything runs in your browser and the tool sends nothing anywhere. Your draft is kept in local storage under lhp.llmsTxt.v1; “Clear form” removes it. If storage is blocked, the tool works but cannot save.

Does llms.txt stop AI crawlers from using my content?

No, it blocks nothing. Access rules belong in robots.txt, and RFC 9309 notes that even those rules “are not a form of access authorization”.

Which links does the tool accept?

Absolute http and https addresses. It checks that each URL is well formed, not that the page exists, so run a link checker after you publish.

Do I need an llms-full.txt too?

The proposal does not define one. Some documentation platforms, such as Mintlify, generate a file that combines a whole docs site; this tool builds only llms.txt.

How many links should the file have?

The proposal sets no number, but it expects the file itself to stay small enough to fit in context, with the detail behind the links. Include a page only if it would appear in a human “start here” list, and leave the rest to the sitemap or Optional.

Related guides

Sources