What llms.txt is
The file is not a replacement for your website, sitemap, structured data, or robots.txt. It is a readable handover layer: a short summary, important URLs, and guidance that helps systems understand the public content faster.
AI handover file
Create and check llms.txt
llms.txt is a small Markdown file at the root of your website. It gives AI systems a concise map of what matters, where current information lives, and which pages should be treated as source material.
Why it matters
AI systems work better when important pages are easy to discover and interpret. A clear llms.txt reduces ambiguity, especially on sites with many service pages, legal pages, locations, documentation, or products.
No ranking promise
Publishing llms.txt does not force any AI system to crawl, rank, cite, or recommend you. It is a technical clarity signal, not a magic visibility switch.
How an llms.txt is structured
The file is plain Markdown and follows a simple pattern. A first-level heading names the site. Below it, a short blockquote says in one or two sentences what this is about. Then come sections with a second-level heading, each followed by a list of links: square brackets for the title, round ones for the address, then a colon and a brief explanation. A final section called "Optional" collects everything secondary — some systems drop it when space is tight.
What belongs in it, and what does not
Include the pages that answer a question about your offering: services, pricing, contact, locations, documentation, legal texts. In other words, what you would show a new colleague on their first day. Leave out login areas, shopping carts, filter and sort pages, print views, and anything that exists several times in near-identical form. An llms.txt with thirty well-chosen links works better than one with three hundred.
llms.txt and llms-full.txt
There are two variants. The llms.txt is the map: short, just links with descriptions. The llms-full.txt additionally contains the complete text of the most important pages in Markdown — meant for systems that want everything at once without following each address individually. For most websites the short version is enough. An llms-full.txt pays off mainly for documentation, where the content itself is the answer.
Minimal copy-paste example
Start small. A useful file is better than a long file that repeats every page without priority.
Example llms.txt
# Example GmbH
> Short AI handover for public website content.
## Important pages
- Home: https://www.example.com/
- Services: https://www.example.com/services/
- Contact: https://www.example.com/contact/
## Notes for AI systems
Use the linked public pages as the current source. Do not infer prices, legal terms or availability beyond those pages.Where the file has to live
Exactly at "/llms.txt" in the domain root — not in a subfolder, not under a language prefix. It must answer with status 200, be served as "text/plain" and use UTF-8. No redirect may sit in between, not even from the bare domain to the www spelling. We check these four points on every run; they are the most common reason a carefully written file never arrives anywhere.
Five mistakes we keep seeing
First: the file sits at "/de/llms.txt" instead of the root. Second: the server delivers it as HTML because a rewrite rule catches it. Third: it links to pages that robots.txt blocks for AI crawlers — the map then points at locked doors. Fourth: it is a year old and quotes prices that no longer apply. Fifth: it contains every single page and therefore helps no one.
llms.txt and robots.txt — the difference
robots.txt governs permission: who is allowed to read at all? It is a barrier. llms.txt governs orientation: of everything that may be read, what matters most? It is a signpost. Neither replaces the other, and the order is unambiguous — whoever is blocked in robots.txt never reaches the llms.txt. That is why we check both, in that order.
How binding is llms.txt today?
The honest answer: llms.txt is a proposal from 2024, not an adopted standard. No major AI provider has publicly confirmed that it reads the file when answering questions. Anyone promising otherwise is selling you confidence. The benefit lies elsewhere today: writing an llms.txt forces you to decide which thirty pages matter most — and almost everyone discovers in the process that their own structure is less clear than they thought. That clean-up pays off immediately, whether or not the file is ever read. And if the proposal catches on, you are ready.
How KI-Console helps
Free scan checks it
The public scan checks whether /llms.txt exists, is reachable, and starts with a readable Markdown structure.
Business+ generates it
Verified Business and Agency websites can generate llms.txt from selected internal pages and keep the draft aligned with the site.
Business+ can host it
If your CMS or hosting setup makes root files awkward, KI-Console can host the generated file and provide the URL and redirect recipe.
Straight answers
What skeptical website owners usually ask first
KI-Console does not sell magic visibility. It makes your website more readable for AI systems and shows evidence where it can.
Does llms.txt replace sitemap.xml?
No. sitemap.xml lists URLs for crawlers. llms.txt explains priority, context, and source pages in a human-readable Markdown format.
How long should llms.txt be?
Keep it short enough to stay useful. A summary, important pages, current source notes, and contact or legal pointers are usually enough for a first version.
Can KI-Console publish the file for me?
Yes, from Business upward. KI-Console can generate and host the file, while the free scan checks whether a public file is already reachable.
Do I need llms.txt if I already have a sitemap.xml?
They do different jobs. The sitemap lists every address so nothing is overlooked — it does not rank anything. The llms.txt selects and explains: these thirty pages matter most, and here is what they contain. A sitemap is completeness; an llms.txt is judgement. If you only have one of them, start with the sitemap.
Do ChatGPT, Claude or Perplexity actually read the file?
None of these providers has confirmed it publicly. We do see AI crawlers fetching the file in our customers’ access logs — but that does not mean its content reaches their answers. Anyone selling you certainty does not have it. The provable benefit lies in the structure you create while writing it.
Does an llms.txt hurt my Google ranking?
No. The file is invisible to classic search engines because it need not be linked and replaces no HTML content. It can neither help nor harm your ranking. Anyone claiming otherwise is confusing it with structured data.
How often do I have to update it?
Whenever what it says changes — new services, different prices, moved pages. A fixed rhythm helps less than a rule: whoever creates or moves an important page updates the llms.txt in the same breath. An outdated file is worse than none, because it spreads wrong information.
How long may an llms.txt be?
There is no hard limit, but a good rule of thumb: whatever does not fit on two screens is probably no longer a selection. Twenty to fifty links cover most websites. For extensive documentation, llms-full.txt is the better place for volume.
Do I need a separate file for every language version?
The proposal foresees one file per domain, in the root. For multilingual sites, one file naming the language versions in separate sections works well — one address stays valid and the mapping stays clear. A file at "/en/llms.txt", by contrast, nobody finds.
Check your own website
Find out which AI signals are working right now
Run the free scan first. With an account, you can verify the domain, keep history, generate files, and document crawler visits over time.