Control how AI models interact with your website content
Build an llms.txt file: a plain-text map of your site written for language models rather than for crawlers. Free, no account.
Site Information
AI Training Permissions
Content Sections
Important pages AI should know about
Paths AI should not access or reference
llms.txt
Check llms.txt
About llms.txt
What is llms.txt? A proposed standard for websites to communicate preferences to AI systems, similar to robots.txt for search engines. Place it at yoursite.com/llms.txt
Training vs Synthesis: Training means using content to improve AI models. Synthesis means using content to answer user queries. You can allow one without the other.
Growing adoption: Major AI companies are beginning to respect llms.txt. Adding one now future-proofs your site for AI-driven search.
Also consider: llms-full.txt for complete site content, and .well-known/ai-plugin.json for AI assistant integrations.
What This Generates
The file goes at the root of your domain and describes what your site is, which pages matter, and where the authoritative version of each topic lives. It is Markdown, it is meant to be read rather than parsed, and it exists because a model working from a rendered page has to reconstruct context that you could simply state.
Be clear on the status: this is a proposed convention, not a standard, and adoption is uneven. Publishing one is cheap and low risk. Expecting every engine to read it is not warranted.
Reading the Output
Generator output, llms.txt
# Acme
> Project management software for agencies
> of two to twenty people. Founded 2019,
> based in Toronto.
## Product
- [Pricing](https://acme.com/pricing/):
current plans and limits
- [Integrations](https://acme.com/integrations/):
supported third-party tools
## Documentation
- [Getting started](https://acme.com/docs/start/)
- [API reference](https://acme.com/docs/api/)
## Optional
- [Changelog](https://acme.com/changelog/)
The blockquote under the title does most of the work. One or two sentences stating plainly what you are and who you serve gives a model the framing it would otherwise infer from marketing copy, and marketing copy is exactly the material that produces vague or wrong summaries.
The Optional section is part of the convention and means what it says: content that can be skipped if the model is working with limited context. Putting your changelog there rather than omitting it lets you include things without making them compete for attention.
What It Cannot Do
It does not control anything. It is not an access rule, it does not block or permit crawling, and nothing is obliged to read it. A model that ignores the file will behave exactly as it did before.
The file that actually controls crawler access is a separate one, and the generator for it is also free. Publishing a helpful map while blocking the crawlers that would use it is a contradiction worth checking for.
Questions
Do AI engines actually read this?
Some tools and agents do. Major engines have not committed to it. The cost of publishing one is close to zero, which is the honest argument for doing it.
Where does the file go?
At the root: /llms.txt. Some sites also publish a fuller version at /llms-full.txt.
How often should I update it?
When the structure of your site changes. It is a map, not a feed.
Is this tool free?
Yes, with no account and no limit.
The file is free to build. What it cannot tell you is whether AI engines actually name your brand once they have read it.
This free tool is one slice of the job. The full kit - rank tracking, audits,
competitor gaps, content, and your visibility across ChatGPT and Google's
AI Overviews - runs on credits you only spend when you use them. Sign up and we'll
drop 25 credits in your account to try the paid tools on us.
