llms.txt Generator
Draft an llms.txt from your sitemap, ready to edit. It appears on screen and arrives by email.
What the llms.txt generator does
Enter your domain, then give a work email. The generator reads your sitemap and your key pages, groups the URLs into sections and makes one model call to write a short note for each link. The result is a draft in the format the llmstxt.org specification describes. The draft appears on screen when it is ready and arrives by email as llms.txt. If you close the tab while it runs, the email still arrives.
The draft describes up to 60 pages, picked by importance: the home page, pages linked from your navigation, top-level sitemap sections, then the rest. If your sitemap has more than 60 pages, the result says how many pages the draft covers, and you can request a complete file with every page. The Tenpoint team writes those by hand and replies within one business day.
It is a draft. A model wrote the notes, so read each one against its page before you publish.
The generator and the llms.txt checker work as a pair. The generator drafts the file, and the checker tests the published version, including its links. The spec itself is short, and its only requirement is an H1, so a draft that looks sparse can still be valid.
What the draft looks like
# Example Co > Example Co makes invoicing software for freelancers. ## Docs - [Quickstart](https://example.com/docs/quickstart): Set up an account and send a first invoice - [API reference](https://example.com/docs/api): Endpoints, authentication and rate limits ## Guides - [Invoice templates](https://example.com/guides/templates): Template options and when to use each ## Optional - [Changelog](https://example.com/changelog): Release history
The H1 names the site and is the only element the spec requires. The blockquote is a one-sentence summary. Each H2 holds a list of links with a note after the colon. The Optional heading follows the spec's convention for links an agent can skip when it needs a shorter context.
What to edit before you publish
The notes
Check each against the page. Models can state a feature the page does not have or describe an old version of a product.
The page list
Remove login pages, thank-you pages, thin tag archives and anything you would not send a new reader to.
The grouping
Rename sections to match how your visitors think about the site. Move secondary pages into Optional.
The summary
Rewrite the blockquote in your own words. It is the first thing an agent reads.
The length
A shorter file is easier to keep accurate. List the pages that matter and stop.
How to write the notes yourself
The note after each link is the part an agent reads in place of the page, so it earns the most editing time. Write one plain sentence that says what the page answers, such as "Endpoints, authentication and rate limits". Skip adjectives and marketing language. If two pages need the same note, one of them probably does not belong in the file.
Where to put the file and how to test it
Save the file as llms.txt and serve it from your site root, so it loads at https://yourdomain.com/llms.txt. The spec also allows a file at other paths, covering the pages under that path, and says an agent should use the most specific file that applies. Most sites need only the root file. A plain-text or Markdown response works best, and your server should return the file itself when the path is requested, with no HTML page in its place.
Then run the llms.txt checker. It tests the structure and requests the links, so a mistyped URL in the draft shows up before an agent finds it. Put a reminder in your calendar to re-run it after site changes. Dead links accumulate in a file nobody revisits.
A pricing change, a renamed guide or a retired product leaves a stale link in the file. Regenerate or edit after any change to your main pages, and run the checker to catch the dead links. The generator reads your live sitemap on each run, so a new draft reflects the site as it stands today. A site with no sitemap is drafted from the links on its home page. Compressed sitemaps (.xml.gz) are not read yet.
Should you publish one?
Be clear about what the file does. Google's AI optimization guide says Search does not use llms.txt and that keeping one won't help or hurt your visibility. OpenAI, Anthropic and Perplexity publish the file for their own documentation, and we found no statement from them that their crawlers read it on other sites. Ahrefs found that 97 percent of files got no requests across 137,210 domains in May 2026, with Claude Code the second most frequent named AI fetcher after GPTBot.
Our reading is that a file helps most when you run developer documentation that coding agents consult, a case we set out in our llms.txt primer. For other sites it is low priority. It takes minutes to generate and review, it can sit alongside everything else you do for search, and we make no promise that it changes what any AI engine says about you.
Tenpoint's angle
A generated file lists your pages. What AI engines say about your company is a separate question, and it decides whether a buyer who asks ChatGPT or Perplexity about your category hears your name. Tenpoint Labs measures that across engines, and fixes the content and entity gaps behind it. The free AI visibility audit shows where you stand today.
More free checks: check your llms.txt, check which AI crawlers can reach your site and generate schema markup.
Key facts
A plain summary of the generator, written for people and for the systems that summarize pages.
- Input
- Your domain and a work email
- Reads
- Your sitemap and up to 60 pages, picked by importance
- Output
- A draft llms.txt in the llmstxt.org format, on screen and by email
- Checked by
- The same rules as the Tenpoint llms.txt checker
- Larger sites
- Request a complete file; the Tenpoint team writes it by hand within one business day
- Price
- Free
- Made by
- Tenpoint Labs
Questions about the llms.txt generator
What is an llms.txt generator?
An llms.txt generator drafts the llms.txt file for a site. This one reads your sitemap and key pages, groups the URLs into sections and writes a short note for each link with one model call. You edit the draft and publish it at your site root.
How many pages does the generator cover?
Up to 60, picked by importance: the home page, pages linked from your navigation, top-level sitemap sections, then the rest. If your sitemap has more than 60 pages, the result says how many the draft covers and offers a request for a complete llms.txt with every page. The Tenpoint team writes those by hand and replies within one business day.
How does the generator work?
You enter your domain and a work email. The generator then reads the sitemap and your key pages, organizes the links under headings and writes the notes. The draft appears on the page when it is ready, and a copy is emailed to you. If you close the tab, the email still arrives.
What should I edit in the generated file?
Read each note against its page, remove pages a new visitor would never need, rename the sections and rewrite the summary in your own words. The notes come from a model, so treat them as a first draft.
Where do I put the llms.txt file?
Serve it from your site root so it loads at yourdomain.com/llms.txt. The spec also allows a file at any path, covering the pages under that path, and an agent should use the most specific file that applies.
Will llms.txt make ChatGPT, Perplexity or Google cite my site?
No evidence supports that. Google says Search does not use llms.txt files and that keeping one won't help or hurt your visibility. OpenAI, Anthropic and Perplexity have made no statement that their crawlers read the file on other sites. Treat the file as low-cost housekeeping, mostly useful for documentation and coding agents.
Do I need llms-full.txt too?
The spec text we read does not define llms-full.txt. Documentation platforms often generate one as a single file of full page text. The generator drafts llms.txt only. The checker reports whether you have an llms-full.txt, and warns if it is served as a web page or is very large.
How often should I update the file?
Update it when you add, move or remove important pages, and re-run the checker after each change. A file with dead links is worse than a short one with working links.
