Short answer: llms.txt is a plain Markdown file at the root of your site (yourdomain/llms.txt) that gives AI tools a short summary of the site and links to its most useful pages. It is a proposed convention, not a standard that crawlers are required to follow, and we cannot point to an assistant that has promised to rely on it. It takes about ten minutes, so it is worth adding, but only after the things that decide whether AI can use your site at all: letting crawlers in, putting the text in the HTML, and saying plainly what you are. Unvibe checks those on your live site.
What llms.txt is
Websites are built for people: menus, images, scripts, pop-ups. An AI tool that wants a quick, clean picture of a site has to dig through all of it. llms.txt is a proposal to make that easier: put a short Markdown file at a known address that says what the site is and links to the pages worth reading. The name echoes robots.txt, and the file is meant for language models (the “LLMs”).
Some sites also publish llms-full.txt, which contains the full content of the key pages in one file, and Markdown versions of individual pages. Both are optional extras.
Does it work?
Be careful with anyone who says yes without qualification. llms.txt is a convention that people have adopted, not a standard crawlers must follow, and we cannot point to a major assistant that has said it relies on the file. Some AI tools and coding agents can use it when you give them your address. Treat it as a cheap extra, not a ranking trick.
What is certain is the cost-benefit: it takes about ten minutes, it cannot hurt if it is accurate, and writing it forces you to state what your site is and which pages matter, which is useful in itself.
What goes in it
The proposal asks for a simple structure in Markdown: a title, a one-paragraph summary, and then sections of links with a short description for each. This is a made-up example for a fictional invoicing tool. Replace everything with your own facts:
# Tallyleaf > Tallyleaf is invoicing software for freelancers. It sends invoices in under a minute, accepts card and bank payments, and tracks who still owes you money. Free for up to 3 clients, then $12 a month. ## Product - [Features](https://example.com/features): invoice templates, reminders, card and bank payments - [Pricing](https://example.com/pricing): the Free plan and the $12 Pro plan, with what each includes - [Integrations](https://example.com/integrations): works with Stripe and PayPal ## Help - [FAQ](https://example.com/faq): common questions about plans, security and cancelling - [Contact](https://example.com/contact): how to reach the team ## Compare - [Tallyleaf vs spreadsheets](https://example.com/compare/spreadsheets): when a spreadsheet is enough and when it is not
A few rules of thumb:
- Keep it short. A summary and a handful of the pages a buyer actually needs. It is not a sitemap.
- Say what is true now. Prices and plan names in the file must match the site. A file that contradicts your pricing page is worse than none.
- Describe each link. “Pricing: the Free plan and the $12 Pro plan” is more useful than “Pricing”.
- Link pages that load without JavaScript. There is no point pointing an AI tool at an empty page.
Where it sits in your priorities
Do these before you write llms.txt, because they decide whether AI can use the site at all:
| Priority | Why it comes first |
|---|---|
| 1. Let AI crawlers in | If robots.txt or a firewall blocks them, nothing else is read. See letting AI search in. |
| 2. Put the text in the HTML | Many crawlers do not run JavaScript. See why ChatGPT can’t see your app. |
| 3. Say plainly what you are | Title, description and headline. See Does AI know what your business is? |
| 4. Publish readable prices and answers | The things buyers ask about. See making prices readable. |
| 5. llms.txt | A tidy summary that points tools at the pages above. |
How to publish it
1. Write the summary
One paragraph: what it is, who it is for, what it costs.
2. List your best pages
Pricing, the product or service pages, FAQ, a comparison, contact. One line each.
3. Save it in your public folder
Name it llms.txt and put it where your site serves static files, so it loads at yourdomain/llms.txt. If you build with a tool like Lovable, Bolt or v0, ask it to “add a static llms.txt to the public folder with this content”.
4. Check it loads
Open the address in a private window. You should see the file, not your homepage or an error page. Many single-page apps return the homepage for every address, which would hide the problem.
5. Keep it current
Add it to your checklist for price or page changes.
Where Unvibe fits
Unvibe publishes an llms.txt of its own, and every documentation page is also available as Markdown, so you can see a working example. A note on scope: the checkup does not score llms.txt. It focuses on the things above it in the table.
The things that come before llms.txt
- Crawler access: robots.txt, firewalls and noindex, checked by visiting as the real AI crawlers.
- Readable HTML: whether the text is in the page without JavaScript.
- Clear identity: whether your title, description and headline say what you are and who it is for.
- Consistency: prices and claims that disagree between pages or with your structured data, shown side by side.
Once those pass, an accurate llms.txt is a pleasant finishing touch. The glossary explains the other terms used here.
Questions
Does llms.txt replace robots.txt?
No. robots.txt tells crawlers what they may fetch. llms.txt is a summary for tools that want a clean overview. You still need robots.txt to be right, and a firewall that lets the crawlers in.
What is llms-full.txt?
An optional companion file that puts the full text of your key pages in one Markdown file. It is useful for documentation sites and less so for a typical business site.
Will adding llms.txt make ChatGPT recommend me?
We cannot promise that, and we cannot point to an assistant that has said it relies on the file. Add it because it is cheap and helps tools that do use it, but fix access, HTML and clarity first.