An llms.txt generator builds the Markdown file that tells AI tools which pages on your website matter most. Enter your site name and a short summary, add curated links or import them from a sitemap, then copy or download a ready-to-upload llms.txt file. Everything runs in your browser.
The result follows the llms.txt proposal: a plain Markdown file named llms.txt that sits at the root of your site. It acts like a short, human-readable map for AI assistants, listing the pages you would want them to read first and explaining each one in a line.
How to use this llms.txt generator
- Type your site name and a one-paragraph summary. Say what the site is, who it is for and what makes it useful.
- Add sections such as Docs, Tools, Products or Policies, then give each page a title, a URL and a short description.
- Or paste the contents of a sitemap.xml file and press Import URLs. The tool reads every loc address, suggests a title from the URL and places the links in a Pages section.
- Remove noise pages such as tags, search results and login screens, and rewrite descriptions so they say when a page is worth citing.
- Check the live preview, then press Copy result or Download llms.txt.
- Upload the file to the root of your domain so it opens at https://yourdomain.com/llms.txt.
What an llms.txt file contains
The format is deliberately simple, so both a person and a language model can read it. Only the first element is required.
| Element | Markdown | Required? |
|---|---|---|
| Title | # Site name | Yes |
| Summary | > One or two sentences | Recommended |
| Details | Plain paragraphs or lists, without headings | Optional |
| Link sections | ## Section name, then lines like - [Title](URL): description | Recommended |
| Optional section | ## Optional | Optional |
A section named Optional has a special meaning in the proposal. It marks links that an AI tool can skip when it has limited space, such as a changelog or older guides. This generator keeps whatever section names you type, so you can create an Optional section yourself.
llms.txt example: a small documentation site
Suppose you run the docs for a product called Acme. You enter the site name Acme Docs, the summary "Developer documentation for the Acme API.", a section called Docs and two links. The tool produces a file with 1 section and 2 links:
| Line in the file | What it does |
|---|---|
| # Acme Docs | Title of the site |
| > Developer documentation for the Acme API. | Summary that agents read first |
| ## Docs | Section heading |
| - [Quickstart](https://acme.example/docs/quickstart): Install the SDK and send your first request | Link with a short description |
| - [API reference](https://acme.example/docs/api): Endpoints, parameters and error codes | Second link in the same section |
Page titles can be left blank. If you import https://acme.example/docs/getting-started, the generator suggests the title Getting started from the last part of the address, and a root address becomes Home. Descriptions are never invented, so write them yourself. Parentheses in URLs are encoded as %28 and %29, and square brackets in titles are escaped, so the Markdown links stay valid.
llms.txt vs robots.txt and sitemap.xml
| File | Aimed at | Job |
|---|---|---|
| robots.txt | Crawlers and bots | Allows or blocks URL paths |
| sitemap.xml | Search engines | Lists the URLs you want crawled and indexed |
| llms.txt | AI assistants and agents | Highlights a curated set of pages with short descriptions |
They do different jobs, so do not merge them. Keep access rules in robots.txt and let llms.txt point out the pages you want AI tools to find.
Does an llms.txt file help SEO? Honest limits
llms.txt is a community proposal, not an official web standard, and adoption is uneven. No search engine has confirmed that it uses the file as a ranking signal, and AI providers differ in whether they read it at all. Publishing one is cheap and harmless, and some documentation tools and coding assistants do look for it, but do not expect traffic from it on its own. Think of it as one small signal that supports the work that matters most: clear content, good internal links and a clean sitemap.
Tips for a better llms.txt file
- Curate. Ten well-described pages beat five hundred bare URLs.
- Describe the use. Say what question a page answers, not a marketing slogan.
- Use absolute URLs with https, pointing to the canonical version of each page.
- Keep private pages out. Staging, admin and login pages do not belong in a public file.
- Group by task, for example Docs, Pricing and Policies, so a model can choose the right section quickly.
- Refresh it when you launch a new tool or restructure your documentation.
- Test it. Open /llms.txt in a browser after uploading and confirm it loads as plain text.
Need more utilities for your site? Browse our other free online tools for developers, writers and site owners.