Free llms.txt generator written from your real pages.
Enter your site. We read your sitemap and pages, write a spec-compliant llms.txt with sections and descriptions, check it against the standard, and build llms-full.txt alongside.
Your llms.txt will appear here
We read your sitemap and up to 60 pages, write the file from what they say, check it against the spec and build llms-full.txt alongside.
How the llms.txt generator works
It does what a careful technical writer would do with an afternoon: read the site, choose what matters, describe it plainly, and follow the format to the letter.
Enter your site
The generator reads robots.txt and your sitemap, or follows links from the page you give it, and picks the pages that matter: documentation, product, pricing, guides, the home page.
It reads the pages and writes the file
Each page's title, description and text go to the writer, which organises them into sections and writes a one-line description per link. Only real pages are linked; anything else is removed.
Check, edit, publish
A checklist shows how the file measures up to the specification. Ask for changes in a sentence or edit it directly, then download llms.txt and its llms-full.txt companion and put them at the root of your domain.
What a good llms.txt contains
# Acme Ledger > Acme Ledger is bookkeeping software for small agencies: invoicing, expenses and VAT returns, with an API for accountants. ## Documentation - [Getting started](https://acme.example/docs/start): create an account, connect a bank, send a first invoice- [Invoicing](https://acme.example/docs/invoices): templates, recurring invoices, reminders and payment links- [VAT returns](https://acme.example/docs/vat): how returns are calculated and filed, with the rules per country ## API reference - [Authentication](https://acme.example/api/auth): API keys, scopes and rate limits- [Invoices endpoint](https://acme.example/api/invoices): create, list and void invoices ## Optional - [Changelog](https://acme.example/changelog): release notes since 2023
- The H1
- The project or site name, on the first line. Exactly one.
- The summary
- A blockquote that says what the site is and who it is for, in a paragraph a model can quote.
- Context paragraphs
- Optional. Terms a model needs, how the product is structured, what the docs cover. No headings.
- Sections
- H2 headings grouped by purpose, each a list of links. Three to six is usual.
- Link lines
- “- [Title](absolute URL): what the page contains.” The description is what makes the file useful.
- Optional
- A final section for pages that can be skipped when context is short. Legal, careers, older posts.
- llms-full.txt
- The companion with the full text of the linked pages, so one fetch gets everything.
Who publishes an llms.txt
Anyone whose site gets described by AI assistants and would rather do the describing.
Developer tools and APIs
Coding assistants read llms.txt before answering questions about your SDK. A good file means correct answers and fewer invented endpoints.
SaaS and product companies
AI search engines and assistants describe your product from what they can read. Tell them, in your words, what it is and where the details are.
Documentation teams
One file that maps the docs by purpose, kept in the repo next to them and regenerated on release.
Marketing and SEO teams
The answer-engine counterpart to a sitemap: a curated, described index that models can use, instead of whatever a crawler happens to find.
Agencies
Generate one per client site in a minute, review the checks, publish.
What the file can and can’t do
llms.txt is a proposal that a growing number of tools honour, not a guarantee that every model will read it. What it reliably does is give any assistant that looks a clear, curated map of your site, written by you rather than inferred from a crawl. The generator gets the format right and the descriptions grounded in your pages; the choices about what to feature are yours to review.
Before you publish
- Read the summary line. It is the sentence a model will repeat about you.
- Check that the sections match how you think about the site, and move anything secondary to Optional.
- Make sure the pages you care about most are in the first section.
- Regenerate when you add or retire pages, the same way you would update a sitemap.
Questions about llms.txt
What is llms.txt?
A plain Markdown file served at /llms.txt that tells language models what your site is and where its most useful pages are. It was proposed by Jeremy Howard in 2024 (llmstxt.org) as the AI-era counterpart to robots.txt: robots.txt says what crawlers may fetch, llms.txt says what is worth reading and how it is organised. Assistants, coding tools and AI search engines read it when someone asks about your product.
What does the file contain?
One H1 with the project name, a blockquote summarising what the site is, optionally a few paragraphs of context, then H2 sections (Documentation, Guides, API, Pricing, Blog and so on), each a list of links in the form “- [Title](https://…): one-line description”. A final “Optional” section holds links a model can skip when it is short on context. That is the whole format, and this tool follows it exactly.
What is llms-full.txt?
A companion file that contains the full text of the pages llms.txt links to, in one document, so a model can read everything in a single fetch instead of following each link. The generator builds it from the pages it read, headed by the llms.txt itself. Serve it at /llms-full.txt next to llms.txt. It can be large; that is expected.
How does the generator find my pages?
It reads robots.txt for sitemap locations, then your sitemap or sitemap index. If there is no sitemap, it follows links from the page you entered, two levels deep. It then reads the most useful pages, up to 60, favouring shallow paths and documentation, product and pricing pages, and reads each one for its title, description and content. It respects robots.txt disallow rules and only ever reads public pages.
Can it make up links or pages?
No. The writer may only use URLs from the pages the crawler found, and the result is checked against that list: any link that is not a real page of your site is removed before you see the file. Descriptions are written from each page's own title, description and text.
My site already has an llms.txt. What happens?
The generator fetches it and shows it to the writer, which keeps what is accurate and well organised, fixes what breaks the format, and adds pages that are missing. The checks panel tells you what was wrong with the current file.
What do the checks look for?
The things the specification requires or recommends: exactly one H1 as the first line, a blockquote summary, H2 sections with lists of links, absolute links on your own domain, no duplicates, a description on every link, no headings deeper than H2, and the Optional section last. Errors break the format; warnings make the file less useful.
How do I publish it?
Put the file at the root of your domain so it is served at https://yourdomain.com/llms.txt, as plain text (text/plain or text/markdown). Most hosts let you drop it in the public or static folder; on Next.js that is public/llms.txt, on WordPress the web root or a plugin. Do the same for llms-full.txt. Then open the URL in a browser to confirm, and keep the file updated when your pages change.
Is it free? Are there limits?
It is free and needs no account. Reading a site takes real work, so there is a daily allowance of sites per visitor, and up to 60 pages are read per site. A generation takes between fifteen seconds and a minute depending on how fast the site responds.
Is my site's content stored?
Pages are read and kept in memory for a short while so a refinement does not have to read the site again, then discarded. Page text is sent to a third-party language model to write the file and is not used to train models. We store nothing about your site beyond counts, such as how many files were generated today.