llms.txt is a Markdown-formatted text file served at yoursite.com/llms.txt that gives AI systems a clean, token-efficient guide to your site: what it is, what you offer, and which pages contain the facts that matter. Think of it as robots.txt's constructive sibling — robots.txt says where crawlers may go; llms.txt says what is worth understanding.
Why it exists
AI systems work under context limits and parse messy HTML imperfectly. A concise, structured summary at a predictable URL lets them ingest your canonical facts cheaply and correctly. The convention started in developer documentation (docs platforms now auto-generate it) and spread to product and business sites as AI answers became a discovery channel.
The format
An H1 with your name, a one-line blockquote summary, then short Markdown sections with the facts and key links. Plain text, no scripts, no tricks:
# Acme > Acme is invoicing software for freelance designers. Plans from $12/month. ## What we do Acme creates, sends, and chases invoices automatically for solo designers and small studios. ## Key pages - Home: https://acme.com/ - Pricing: https://acme.com/pricing - Product tour: https://acme.com/tour - FAQ: https://acme.com/faq ## Frequently asked questions - Q: How much does Acme cost? A: Plans start at $12/month with a 14-day free trial. - Q: Who is Acme for? A: Freelance designers and studios under 10 people. ## Contact - Website: https://acme.com - Email: hello@acme.com
What llms.txt can and cannot do
- Can: give engines a canonical, unambiguous statement of what you are — reducing hallucinated descriptions and stale facts.
- Can: point crawlers at your highest-value pages (pricing, FAQs, comparisons) so they get read first.
- Cannot: force any engine to mention you — it is one signal in the stack, not a ranking hack.
- Cannot: substitute for crawlable HTML, structured data, or third-party corroboration.
Keep it truthful and current. An llms.txt that contradicts your site — old pricing, discontinued products — actively spreads wrong answers about you. Review it whenever facts change.
Checklist for a good llms.txt
- Serve it at the root as plain text (not HTML), reachable without JavaScript.
- Name + one-sentence identity in the first two lines.
- State category, audience, and real pricing — numbers are what engines quote.
- 5–10 key page links, each with a word on why it matters.
- 3–8 FAQs with direct answers.
- Keep it under ~200 lines; it is a guide, not a sitemap dump.
- Update it when facts change; stale llms.txt is worse than none.
Key takeaways
- llms.txt = a plain-text canonical guide to your site for AI systems, served at /llms.txt.
- It reduces wrong AI answers about you by stating identity, pricing, and FAQs unambiguously.
- It is one layer of AEO — it complements structured data and crawlable content, never replaces them.
- Generate it from your brand facts and keep it in sync as they change.
Zero hand-writing
Generate your llms.txt automatically
AnswerScope's Fix Pack builds llms.txt, a facts JSON endpoint, robots rules and FAQ schema from your brand facts.
Frequently asked questions
What is llms.txt?
A plain-text, Markdown-formatted file served at your site root (/llms.txt) that tells AI systems what your site is about, your key facts, and which pages matter most.
Does llms.txt improve AI visibility?
It is one signal: it gives engines a canonical, cheap-to-read statement of your facts, which reduces wrong or stale descriptions. It works alongside crawler access, structured data, and quotable content — not instead of them.
Is llms.txt the same as robots.txt?
No. robots.txt controls where crawlers may go; llms.txt tells AI systems what your site means and which pages carry the canonical facts.
How do I create an llms.txt file?
Write it by hand following the format above, or generate it: AnswerScope's Fix Pack produces llms.txt from your stored brand facts, alongside a facts JSON endpoint, AI-crawler robots rules, and FAQPage schema.