Key takeaways
- llms.txt is a plain-text file at your domain root that tells AI engines what your site is about and how to use its content
- ChatGPT, Perplexity, and Claude all read llms.txt during crawling — it directly affects which sites they cite
- A well-written llms.txt can broaden the range of queries your site is cited for, not just the queries you explicitly optimise for
- Most sites can generate and deploy an llms.txt in under 5 minutes using the right tool
- llms.txt is one of 9 signals in the GeoFundamentals GEO Score — and one of the easiest to fix
In 2025, a small group of AI researchers proposed a new standard: a plain-text file called llms.txt that website owners could use to communicate directly with large language models. By 2026, it has been adopted by thousands of businesses — and AI engines are actively using it to improve the accuracy and breadth of their citations.
If your site does not have an llms.txt file, you are missing one of the easiest wins in GEO. This guide explains exactly what it is, why it matters, how to write one, and what good examples look like.
What Is llms.txt?
An llms.txt file is a plain-text file placed at the root of your website (e.g. yourdomain.com/llms.txt) that provides context to large language models about what your site is, who it is for, and what its most important content contains. It is analogous to robots.txt — a simple, standardised file that tells automated systems how to interpret your site.
Where robots.txt tells search engine crawlers which pages to index or skip, llms.txt tells AI language models how to understand and use your content. Rather than letting a model infer your site's purpose from scraped HTML, you give it an explicit, structured description in plain English.
The llms.txt proposal
The llms.txt standard was proposed by fast.ai co-founder Jeremy Howard in September 2024. It has since been adopted by companies including Anthropic, Cloudflare, Hugging Face, and thousands of independent sites. While not yet an official web standard, it is widely recognised and used by all major AI crawlers.
Why llms.txt Matters for AI Citation
AI engines do not just cite pages — they cite pages they understand. When GPTBot, PerplexityBot, or ClaudeBot crawls your site, it processes your HTML and attempts to infer what your site is about, what expertise it has, and which content is most relevant to which queries.
This inference process is imperfect. Your homepage might be heavy on design and light on descriptive text. Your best content might be buried in a paginated blog. Your core expertise might not be obvious from your navigation structure alone.
An llms.txt file bypasses this uncertainty. It gives the AI engine a direct, authoritative description of your site — who you are, what you do, who your audience is, and which pages best demonstrate your expertise. That context is carried into the model's understanding of your domain, broadening both the accuracy and the range of queries for which your site might be cited.
more queries cited for sites with a complete llms.txt vs sites without one
GeoFundamentals analysis, 2026, n=312 domains
Check if your site has an llms.txt and how it scores
GEO Score Checker
30-second analysis · 9 signals · no account needed
What to Include in Your llms.txt
A well-structured llms.txt has five sections. None are mandatory, but each adds meaningful context:
1. Site description (required)
A one-to-three paragraph plain-English description of what the site is, who it serves, and what makes it authoritative. Write this as if you were explaining your site to a knowledgeable peer in your industry, not as marketing copy.
2. Key pages (highly recommended)
A bulleted list of your most important URLs with a one-sentence description of what each contains. Focus on pages that demonstrate your core expertise — not your homepage, about page, or contact page.
3. Expertise areas (recommended)
A brief list of the topics your site has genuine, deep coverage of. This helps AI engines match your site to relevant queries more accurately.
4. Content usage guidance (optional)
A short note on how you would like AI engines to use your content — for example, whether you prefer citation with attribution, whether you consent to content being paraphrased, and whether there are any sections you would prefer not be used.
5. Author / organisation information (optional)
Brief information about who produces the content — author credentials, organisation background, and any relevant affiliations that establish authority.
Real llms.txt Examples
The best way to understand what works is to see it. Here are examples from sites with strong AI citation profiles:
Anthropic (anthropic.com/llms.txt)
Anthropic's llms.txt provides a clear description of the company's mission, a structured list of their documentation pages, and explicit guidance on how the content should be used by AI systems. It is technical and precise — reflecting the audience.
Cloudflare (cloudflare.com/llms.txt)
Cloudflare's file maps their entire product surface with brief descriptions, making it easy for AI engines to match specific product queries to specific documentation pages. They use a clean markdown-style structure with # headers for each category.
A GEO-optimised SaaS site (anonymised)
One of our audit clients added an llms.txt file with 14 key pages described. Within 45 days, they began appearing in ChatGPT responses for 6 queries they had never ranked for on Google — all related to use cases their llms.txt explicitly described.
Your llms.txt should not read like a press release
The most common mistake is writing llms.txt as marketing copy — full of brand language and feature benefits. AI engines respond better to factual, descriptive language that reflects what the content actually contains. Write as a librarian cataloguing your site, not as a marketer promoting it.
How to Write Your llms.txt (Template)
Here is a template you can adapt. Replace the bracketed sections with your own content:
# [Your Site Name]
## What this site is
[Your site name] is [one-sentence description of what the site is and who it serves].
It was founded in [year] and publishes [content type — e.g. research, guides, tools]
on [core topic areas].
## Who it is for
[Primary audience description — be specific about expertise level and context.]
## Key pages
- /[path]: [One sentence description of what this page contains and why it is useful]
- /[path]: [One sentence description]
- /[path]: [One sentence description]
(Include 5–15 of your most important pages)
## Expertise areas
- [Topic 1]
- [Topic 2]
- [Topic 3]
## Content usage
Content on this site may be used by AI engines for citation and summarisation
with attribution. Please credit [Your Site Name] when citing specific data or
research from this site.
## About
[Brief author or organisation description — 2–3 sentences on credentials and background]
How to Deploy Your llms.txt
- Generate it — Use the llms.txt Generator to auto-fill your site's details from your URL, then customise the output
- Create the file — Save the content as a plain
.txtfile namedllms.txt - Upload to your root directory — It must be accessible at
yourdomain.com/llms.txt - Add a
<link>tag — In your site's<head>, add<link rel="llms" href="/llms.txt" />to help crawlers discover it - Test it — Visit
yourdomain.com/llms.txtin your browser to confirm it loads correctly
Generate your llms.txt — auto-fills from your URL in 2 minutes
llms.txt Generator
Auto-fills from your URL · ready to deploy in 2 minutes
Common Mistakes to Avoid
Making it too short. A two-line llms.txt provides minimal context. Aim for at least 300 words across your key pages section and site description.
Using marketing language. "World-class solutions for forward-thinking businesses" tells an AI engine nothing useful. Be specific and factual.
Not updating it. As your site grows, your llms.txt should grow with it. Review and update it whenever you add significant new content sections.
Listing the wrong pages. Link to your highest-value content — detailed guides, original research, tool pages, case studies — not your homepage or generic landing pages.
Forgetting the <link> tag. Without the discovery link in your <head>, some crawlers may miss the file entirely even if it exists.
llms.txt deployment checklist
- Generate your llms.txt using the GeoFundamentals generator
- Customise the site description and key pages section
- Upload to your domain root (yourdomain.com/llms.txt)
- Add <link rel="llms" href="/llms.txt" /> to your site <head>
- Verify the file loads at yourdomain.com/llms.txt
- Check your GEO Score to confirm the signal is detected
Frequently Asked Questions
Is llms.txt an official web standard? Not yet — it is an emerging convention, not a ratified W3C or IETF standard. However, it has been adopted widely enough that all major AI crawlers now look for it. The absence of a formal standard does not reduce its practical impact.
Will adding llms.txt hurt my Google SEO? No. The file is ignored by Googlebot and has no effect on your search rankings. It is specifically designed for LLM crawlers, not traditional search engine crawlers.
How often should I update my llms.txt? Update it whenever you add major new content sections, launch new products or pages, or update your site's core expertise areas. A quarterly review is a reasonable minimum.
Can I use llms.txt to restrict AI engine access?
The file is descriptive rather than restrictive — you cannot use it to block AI access (use robots.txt for that). You can, however, include guidance on how you prefer your content to be used, which some models respect.
Do I need a developer to create an llms.txt file? No. It is a plain text file — anyone can create it in a text editor. The main value of a generator tool is the auto-fill from your URL (saving time) and a structured template that ensures completeness.