Xt4b

llms.txt

A summary file for the machines that read your site.

llms.txt is a plain markdown file at the root of a domain that tells a language model what the site is and which pages matter. Xt4b serves one here, and one on every blog domain it hosts — generated from what is actually published, not written by hand and left to go stale.

Start with your domain Start with your domain

Served per domainPublished posts onlyNo guarantees attached

What is in the file.

Six parts, and the reason for each. A summary file is only useful if it is short enough to be read in one pass and specific enough to be worth reading.

An H1 with the name
One heading, the site's name, so a reader — model or human — knows immediately whose file this is before parsing anything below it.
A one-line summary
A blockquote on the app host that says what Xt4b is in a sentence, because a summary is the thing an answer engine is most likely to reuse.
Facts, one per line
On the app host: the price, the unit sold, what is published and what is explicitly guaranteed not to be. Short declarative lines, no marketing adjectives.
The pages that matter
Links to the landing page, the guides, the API reference and the legal pages, each with a line saying what a reader would find there.
The posts themselves
On a customer's blog: one line per published post, with its URL and its meta description, newest first. Nothing unpublished appears.
Plain markdown, nothing else
No JavaScript, no markup the reader has to execute, no login and no cookie wall — a text file that can be read with one request, which is the entire point of the convention.

What a hosted blog serves.

The shape of the file on a customer's domain. One line per published post, newest first, with the meta description as the detail.

https://yourdomain.com/llms.txt

# Northbrew Blog

> Latest posts from Northbrew — the blog of
northbrew.example.

- [How to descale a steel kettle](https://northbrew.example/how-to-descale-a-steel-kettle): What
  descaling actually removes, how often to do it in
  a hard-water city, and the one thing not to use.
- [Grinder burrs: conical or flat](https://northbrew.example/grinder-burrs-conical-or-flat): What
  the shape changes in the cup, and when it does
  not matter at all.

The file is generated from the blog’s published posts, so it cannot drift from what is live: a post that is planned, in review or unpublished is not in it, and a post that is taken down leaves it.

There is no separate setting and nothing to paste. The same domain serves a sitemap, a robots.txt that allows crawling, an icon and one page per post alongside it.

The app host serves its own file at xt4b.com/llms.txt, which states what the product is, what it costs and what it refuses to promise alongside the links to the guides.

The files that matter more than this one.

llms.txt is a convenience. Everything underneath it is what decides whether a page can be read, quoted and attributed at all.

A model — or a crawler, or a person — reaches a page, and the page has to say what it is about in language that survives being extracted. That means a title that is not a slogan, one question answered per page, and the answer in the first screen rather than four hundred words in.

This is the part the pipeline is built around: one query per planned post, an outline approved before the writing starts, and the metadata, canonical URL and structured data emitted with the post rather than retrofitted months later.

  1. A sitemap.xml on the same domain, listing every published post
  2. A robots.txt that allows crawling rather than shutting it out
  3. A canonical URL on every post, so two paths never look like two pages
  4. Per-post structured data describing the article and its date
  5. A cover image with real alt text rather than a filename
  6. One answer per page, in the language the query was asked in
Publishing so an answer engine can quote you
SEO content API

What the file cannot do

It cannot make an assistant read it, cite it or rank it. llms.txt is a proposal rather than a standard, and nothing in this or any other text file is guaranteed to be fetched — ranking, traffic and citation are not guaranteed by us, and our terms say so.

Nor is it a shortcut around thin content. A model that reads a well-indexed file of pages that say nothing useful still comes away with nothing, and a site with one honest answer per page is far better served by that than by any file at its root.

More on content

Questions about llms.txt.

Including where ours is served, what is in it, and whether the crawlers actually read it.

What is llms.txt?

A proposed convention from llmstxt.org: a plain markdown file at /llms.txt that gives a language model a short, structured summary of a site — what it is, which pages matter and what they contain — instead of leaving it to infer all of that from rendered HTML.

Where does Xt4b serve it?

Two places. On the app host at xt4b.com/llms.txt, which describes the product, states the price and links the guides and the API reference. And on every customer's blog domain at that domain's own /llms.txt, which indexes that blog's published posts.

What is in a customer's llms.txt?

A heading with the business name, one blockquote line saying what the blog is, and then one bullet per published post: the title as a link, followed by the post's meta description. It is an index rather than a summary — the blog's own posts are the content.

Does it help a site get cited by AI?

It makes the site easier to read correctly, which is all any file can do. Whether an assistant cites a page is decided by that assistant, and no ranking, traffic or citation is guaranteed by us. The file removes an obstacle; it does not buy a mention.

Is llms.txt an official standard?

No. It is a proposal, not something the crawlers are obliged to read, and there is no guarantee that any given model fetches it. That is exactly why the file is treated here as a layer on top of the things that are standard: the sitemap, the canonical URLs, the structured data and the crawlable HTML.

Do I have to do anything to get one?

No. Both files are served by the same routes that serve the blog, so a connected domain gets a sitemap, a robots.txt, an icon and an llms.txt without an extra setting. If you would rather serve the posts from your own CMS, the read API hands you the markdown and the rendered HTML, and you can emit whichever index files you like.

Should I add llms.txt to my main website too?

It is cheap and harmless: a short file stating what the business is, who it serves and which pages matter. The higher-return work is almost always underneath it — titles that say what the page is about, one question answered per page, and pages that render without JavaScript.

Does the file include posts that are not published yet?

No. It is generated from published posts only, in newest-first order, and it changes as the blog does. A post that is planned, in review or unpublished is not listed anywhere in it — the same rule the sitemap follows.

Read by machines

Publish the pages worth summarising.

Connect a domain and read the plan: the queries your site is missing, the outlines behind them, and the dates each post goes live on.

Automate my blogStart with your domain

Sitemap and llms.txt per blog20€ / site / language / month3 days free, no card