llms.txt generator
Fill in the fields and get an llms.txt that passes the four checks the format actually has — plus the llms-full.txt skeleton for the pages you want an assistant to read in full. Nothing is uploaded; the whole thing runs in this tab.
Your files
Upload this to the root of your domain so it resolves at /llms.txt. Most hosts: drop it next to index.html.
llms-full.txt carries the page text, not just links. We cannot read your pages from here, so this is the skeleton with slots marked — paste each page's opening paragraphs into its slot.
The four checks, run on what you just wrote
What the format actually requires
The first line is an H1 and it is your site name. The second block — optional in the proposal, and the one people skip — is a single line of blockquote text saying what this site is. Without that line an assistant has only your link titles to guess from, and guessing is where it goes and reads someone else instead.
Sections are H2. Under each one is a list of links, written as <code>- [title](URL): a note</code>. The note is not decoration. It is the only thing an assistant has to decide whether opening that page is worth it, and a list of bare links is a list of pages it has no reason to prefer over the ones it already knows.
<code>## Optional</code> is a convention, not a rule: contact page, licence, changelog — links that help if they are there and do not break anything if they are missing. Keeping them in their own section is what stops them from diluting the section that matters.
# Acme > Acme builds invoicing software for freelancers. ## Docs - [Pricing](https://acme.com/pricing): what each plan costs and what is included - [Getting started](https://acme.com/docs/start): set up your first invoice in ten minutes ## Optional - [Contact](https://acme.com/contact)
What we measured about llms.txt
We ran five real queries, collected the 48 unique sources a web search returned, and asked a fixed model the same question six times per query, with the full page text in context. Then we correlated every page-level feature we could extract against whether that page was cited.
At n=48 the two-tailed 0.05 threshold is about 0.285. Only JSON-LD clears it. llms.txt does not, and neither does anything else in the list — so on our own data we cannot tell you that writing this file will get you cited, and we are not going to pretend otherwise.
What we can tell you is why it is still worth ten minutes. It costs nothing, it cannot hurt, and it is the one place where you state, in your own words, what each page answers. The day an assistant does read it, that statement is the difference between being summarised correctly and being summarised as whatever your title happened to say.
Where this file helps, and where it stops
It helps when the crawler can already reach you. If a retrieval crawler is blocked in robots.txt, or your body text only exists after JavaScript runs, an llms.txt describing that text is a description of something the assistant never got. Fix reachability first — that is the one with a mechanism behind it, and it is what the <a href="../tools/robots-txt-generator.html">robots.txt generator</a> is for.
It stops at anything that requires the file to be read. There is no enforcement, no registry, and no penalty for ignoring it. A site with no llms.txt and a well-structured page will beat a site with a beautiful llms.txt and an unreadable one, every time.
Questions about llms.txt
Is llms.txt an official standard?
No. It is a proposed convention, not a specification anyone is obliged to follow, and adoption is uneven. We generate it because it is cheap and cannot hurt you — not because we have evidence it moves citations. Across 48 sources the correlation between having one and being cited was 0.168, which is below the 0.285 that a sample this size needs before we would call it a signal.
Does adding llms.txt get my page cited?
We cannot show that it does. Of the page-level features we measured across 48 sources, llms.txt had a correlation of 0.168 with being cited, FAQ blocks 0.095 and summary blocks 0.070. JSON-LD was the only one that cleared the threshold, at 0.391. That is not an argument against writing llms.txt — it is an argument against buying it as a ranking fix.
What is a fake 200 llms.txt?
A single-page app answers every path with its own HTML shell and a 200 status. Ask it for /llms.txt and you get 200 back — with HTML inside, not a plain-text file. A checker that only looks at the status code marks that site as having an llms.txt when it does not have one. Check that Content-Type is text/plain and that the first line starts with #.
Should llms.txt list every page on my site?
No. It is a map for an assistant that has to decide which pages are worth opening, and a map that lists everything is the same as no map. List the pages that answer a question someone would actually ask, and give each one a note saying what it answers. Ten good entries beat four hundred exported URLs.
Do I need llms-full.txt as well?
Only if you want an assistant to be able to use your text without fetching each page. llms.txt is a list of links; llms-full.txt carries the text itself. Most sites start with llms.txt and add the full version for the three or four pages that matter most.
The other tools
robots.txt generator for AI crawlers
Choose a policy and get the AI-crawler block, with retrieval and training kept apart.
FREE Open →JSON-LD generator
Organization, WebSite, Article, FAQPage and BreadcrumbList, validated as you type.
FREE Open →AI crawler reference
Every user-agent that decides whether your page enters an AI answer, and what blocking it costs.
FREE Open →Fix generator — all four files at once
If you want llms.txt, robots.txt and JSON-LD in one pass instead of one at a time.
FREE Open →AI readability check
Eight checks on a pasted page, including whether the two files above are even reachable.
FREE Open →