llms.txt generator, free

Short answer: fill in your site name, a one-line summary and a list of key pages, and the generator writes a Markdown file in the llmstxt.org format. Save it as llms.txt in your web root so it loads at yourdomain.com/llms.txt. Be aware there is no evidence yet that AI search engines use it.

How to make an llms.txt file

1. Type the site or project name. It becomes the H1.
2. Write one sentence on what the site is. It becomes the blockquote, which is what a model reads first.
3. Optionally add a sentence or two of context, such as who the site is for.
4. List your key pages, one per line, as Title | https://url | short description. The description is optional.
5. Copy the output, save it as a UTF-8 plain text file named llms.txt, and upload it to the root of your domain.
6. Open https://yourdomain.com/llms.txt in a browser to confirm it loads as text, then run it through the llms.txt validator.

The llmstxt.org proposal defines a simple structure, and this generator follows it:

1. An H1 with the site or project name. It's the only required element.
2. A blockquote right after, summarising what the site is (recommended, not required).
3. Optional plain paragraphs with more detail (no headings).
4. One or more H2 "file list" sections, each entry formatted - [Name](url): description.
5. An optional ## Optional section: links an AI may skip when context is tight.

Companion files you can add later: /llms-full.txt (the full expanded content in one file) and .md versions of your HTML pages served at the same URL + .md.

Worked example

With the placeholder inputs (name "Acme Analytics", the summary sentence, the context line and the three page lines), the generator outputs:

# Acme Analytics

> Acme is a privacy-first web analytics platform for SaaS teams.

Docs, guides and API references for developers integrating Acme.

## Key Pages

- [Docs](https://acme.com/docs): Getting started & API reference
- [Pricing](https://acme.com/pricing): Plans and limits
- [Blog](https://acme.com/blog): Product and analytics guides

What the generator doesn't do

• It writes a single section called Key Pages. If you want several H2 sections (Docs, Guides, an ## Optional list), copy the output and add the headings by hand.
• It doesn't crawl your site or check that the URLs work. It formats what you type.
• It doesn't build llms-full.txt, which needs the full text of each page.
• Square brackets are removed from titles, and spaces and closing brackets in URLs are encoded, so the Markdown links don't break.

Troubleshooting

• "N lines skipped: no URL." A page line has no second part. Check you used the pipe character | between title and URL, not a comma or dash.
• The file downloads instead of displaying, or shows garbled characters. Your server is sending the wrong content type. It should be served as text/plain or text/markdown with UTF-8.
• 404 at /llms.txt. The file is in a subfolder. It has to sit at the domain root, next to robots.txt.

Be honest about llms.txt: it's a proposal, not an established ranking factor. Google has said it doesn't use it. Large public studies (Ahrefs across about 137,000 domains, SE Ranking across about 300,000, and Otterly's 90-day log test) found no sign that AI systems rely on it, and a 90-day log test on this site found the file was barely requested, and almost never by AI retrieval bots. That test measured fetches, not citations, so the effect on citation is still open.

It costs a few minutes, so publishing one is harmless. Just don't count it as progress. Effort is better spent where there is evidence: the GEO tactic evidence scoreboard grades each common tactic, and how to get cited by ChatGPT covers one engine in detail. Already have a file? Check it with the llms.txt validator.