llms.txt generator
Enter a URL and get a draft llms.txt built from your sitemap, page titles and descriptions, ready to edit.
- Free
- No account required
Pages in the file
# Example
> An online home-goods store. Ships across the EU, 30-day returns.
## Main pages
- [Home](https://example.com/): Categories, best sellers and current offers.
- [Shipping and payment](https://example.com/shipping): Delivery costs and times, accepted payment methods.
- [Returns](https://example.com/returns): How to return a product within 30 days.
## Category
- [Lighting](https://example.com/category/lighting): Table lamps, ceiling lights and outdoor lighting.
- [Textiles](https://example.com/category/textiles): Cushions, throws and curtains.
## Guides
- [How to choose a lamp for a room](https://example.com/guides/choosing-a-lamp): Brightness, colour temperature and hanging height.
Publish the file at https://example.com/llms.txt
This llms.txt generator reads the robots.txt and sitemap of the site you enter, picks up to 30 key pages in one language, reads their titles and meta descriptions, and assembles an llms.txt file in the llmstxt.org format: a heading with the site name, a short summary and link lists in sections. You can edit the name, summary and page list before downloading. llms.txt is a proposal, not a standard. Google says no special files are needed to appear in its AI features.
What llms.txt is for
llms.txt is a Markdown file at /llms.txt. Jeremy Howard proposed it in September 2024 (llmstxt.org). The only required element is an H1 with the site name; after that, optionally, a short summary in a blockquote, paragraphs of context and H2 sections with link lists in the form [name](url): notes. A section called "Optional" marks links an agent may skip.
It is not a standard or a ranking lever. Google says you do not need to create new machine-readable files to appear in its AI features. llms.txt does not control crawler access either, robots.txt does that. Treat it as a cheap, optional map for tools and agents that choose to read it.
The generator only knows what the sitemap shows. It prefers pages higher in the structure, at most eight from one section, and takes their titles and meta descriptions. It cannot know which pages matter most to you, so review the list and untick what does not belong.
llms-full.txt, a single file with the full text of many pages, is a convention some tools use. It is not in the llmstxt.org proposal, and this generator does not create it.
Edit the draft before you publish
The generator knows only the sitemap and meta tags. Rewrite the summary so that two sentences say what the site is and who it is for, untick legal and technical pages, and if you have documentation or pricing, make sure they are on the list.
llmstxt.org, The /llms.txt fileHow to create an llms.txt file for your site
An llms.txt file is a short Markdown map of a site at <code>/llms.txt</code>. Five steps take you from a draft built from your sitemap to a file on your server.
-
Generate a draft from your sitemap
Enter your site URL in the generator above. The tool reads the sitemap, picks up to 30 pages in one language and takes their titles and meta descriptions.
-
Edit the name and summary
The H1 is the site name and the only required element. Below it, in a blockquote, write one or two plain sentences on what the site is and who it is for, without advertising slogans.
-
Arrange sections and link notes
Group links under H2 sections, each in the form
[name](url): notes. Move less important pages to an## Optionalsection, which an agent can skip when it is short on context. -
Publish the file in the site root
Upload the file so it answers at
https://example.com/llms.txt, next to robots.txt. Open the URL in a browser and check that the server returns the file with a 200 status. -
Update the file along with the site
When you change key pages, their URLs or your offer, generate a new draft and compare it with the published version. A stale link sends the agent to a 404.
Example llms.txt files
The URLs are examples on example.com.
# Example
> Example is a store for everyday ceramics. We ship orders across the EU and the US.
## Products
- [Mugs](https://example.com/mugs): prices, sizes and shipping times
- [Returns](https://example.com/returns): policy and return formAn H1, a summary blockquote and one section of links are plenty for a small site.
- [Page](https://example.com/pricing)- [Pricing](https://example.com/pricing): monthly and annual plans, prices in USD and PLNThe notes after the colon tell the agent what it will find at the URL before it fetches it.
## Docs
- [Quick start](https://example.com/docs/start): setup in three steps
## Optional
- [Changelog](https://example.com/changelog): list of releases
- [Blog](https://example.com/blog): articles and newsThe name Optional has a special meaning in the proposal: it marks links that can be skipped.
Common llms.txt mistakes
-
Treating llms.txt like robots.txt
llms.txt neither blocks nor allows anything. You set crawler access in robots.txt, separately for each token, such as GPTBot or OAI-SearchBot.
-
Hundreds of URLs copied from the sitemap
The file should help an agent find the most important pages quickly. A full list of URLs turns it into a second sitemap.
-
Links to blocked or missing pages
A URL with noindex, blocked in robots.txt or returning 404 leads the agent to a dead end. Click every link before you publish.
-
A summary written like a slogan
A line like “the best solution on the market” says nothing about the site. Write what you offer, to whom and in what scope, as you would for a new colleague.
-
Expecting better rankings
Google says its AI search features need no AI text files. Treat llms.txt as a cheap extra for tools that choose to read it.
Status and structure of llms.txt
| Value | What it means | Source |
|---|---|---|
| 3 September 2024 | Publication date of the llms.txt proposal by Jeremy Howard. It has the status of a proposal open to community input. | llmstxt.org |
| 1 element | What the proposal requires: an H1 with the project or site name. The summary and link sections are optional. | llmstxt.org |
| Optional | An H2 section with this name marks secondary links an agent can skip when it needs a shorter context. | llmstxt.org |
| .md | The proposal also recommends Markdown versions of pages at the same URL with .md appended. | llmstxt.org |
| No AI files | Google says you do not need new machine readable files, AI text files or special markup to appear in its AI search features. | Google Search Central, AI features |
| robots.txt | Where you set consent for model training: OpenAI reads the GPTBot token, and Google reads Google-Extended. | OpenAI, crawlers |
Sources checked:
File structure
- #Name
- An H1 with the site name, the only required element.
- >Summary
- A short blockquote: what the site is and who it is for.
- ##Sections
- Link lists in the form [name](url): notes, grouped by topic.
- OptOptional
- A section of links an agent can skip when it needs a shorter context.
A Markdown file at /llms.txt that briefly describes a site and points to its most important pages. Jeremy Howard proposed the format in September 2024 at llmstxt.org. Only the H1 with the name is required; the summary and link sections are optional.
There is no evidence that it does. Google says no special files are needed to appear in its AI features, and llms.txt remains a proposal rather than a standard. The file costs nothing and may help tools that read it, but it does not replace well-described content that crawlers can reach.
In the site root, so it is served at https://your-domain/llms.txt, the same place as robots.txt. The proposal also allows files in subdirectories, such as /docs/llms.txt, which describe only the URLs under that path.
robots.txt tells crawlers what not to fetch. llms.txt neither blocks nor allows anything. It is a map with a description of the site and a list of its key pages. To block a particular AI crawler, use robots.txt.
A convention some tools use: one file with the full text of many pages instead of a list of links. It is not part of the llmstxt.org proposal, so this generator produces llms.txt only.
It looks for a sitemap in robots.txt, then at /sitemap.xml. It keeps same-domain pages in one language, without parameters or files, shallower pages first, at most eight per section and 30 in total. Titles and descriptions come from each page's meta tags. Without a sitemap it uses the links on the home page.
Our server fetches the pages, so the address reaches our logs. We do not keep the generated file: you copy or download it yourself.
An llms.txt file works without Markdown versions. The proposal does recommend a clean copy of each page at the same URL with .md appended, because it is easier for an agent to read than HTML. That is separate work on your site, and this generator does not do it.
An llms.txt file does not set training consent. You do that in robots.txt: OpenAI reads the GPTBot token, Anthropic reads ClaudeBot, and Google reads Google-Extended. Blocking these tokens leaves the same companies' search crawlers unaffected.
You update llms.txt when the pages it lists change: a new offer, changed URLs or a removed page. The simplest habit is to check it with every larger release, the same way you check your sitemap.
What else might come in handy
File or code ready? Check the result on the live page
Indexability, metadata and structure, with priorities and a client report.