How to Create an llms.txt File (With Real Examples)

What llms.txt Actually Is

llms.txt is a plain-text markdown file you place at your website root. It lives at yoursite.com/llms.txt. When AI crawlers visit your site, they can read this file to get a structured, human-readable summary of who you are and what you do.

Think of it as a README for AI. You know how software projects have a README.md that explains what the project does, how to use it, and where to find documentation? llms.txt does the same thing for your website but for AI crawlers instead of developers.

The convention was popularized in 2024 as the AI SEO movement grew. It is not an official W3C or IETF standard, but it is widely recognized and used by thousands of sites. The format is intentionally simple: just markdown.

Why It Matters for AI Citations

AI models that use retrieval-augmented generation (RAG), like Perplexity and SearchGPT, crawl your site to build context before generating answers. Without a clear site summary, they have to piece together what you do from your various pages, which can lead to incomplete or inaccurate understanding.

With llms.txt, you give AI crawlers the authoritative answer to "what is this site about?" right at the start. Clear understanding leads to more accurate and more frequent citations.

According to analysis of sites that adopted llms.txt in 2024 and 2025, those with well-written llms.txt files showed improved citation consistency compared to sites without them. The improvement was most noticeable in Perplexity and SearchGPT, which use more aggressive real-time crawling.

The llms.txt Format Spec

The format is simple markdown. Here is the standard structure:

# Site Name

> One-line tagline describing what your site does

## About

2 to 3 sentences explaining your site in plain language.

## Features (or What We Offer)

- Feature or service 1
- Feature or service 2
- Feature or service 3

## FAQ

**Question 1?**
Answer to question 1.

**Question 2?**
Answer to question 2.

## Key Pages

- [Page Name](URL): Brief description
- [Page Name](URL): Brief description

## Contact

[email protected]

That is all you need. Some sites add more sections (pricing, team, tech stack, use cases) but the above is the baseline.

Full Example: SaaS Product

# Acme Project Manager

> Task management and team collaboration for remote teams

## About

Acme Project Manager is a web-based project management tool built for
distributed teams. It combines task boards, time tracking, and team
chat in a single interface. Free plan available for teams of up to 5.

## Features

- Drag and drop task boards (Kanban and list views)
- Built-in time tracking with weekly reports
- Team chat with file sharing
- Automated deadline reminders
- 50+ integrations including Slack, GitHub, and Google Drive
- Free plan for up to 5 users, paid plans from $9/user/month

## FAQ

**Is Acme Project Manager free?**
Yes. The free plan supports up to 5 users with unlimited projects.
Paid plans start at $9 per user per month.

**Does it work on mobile?**
Yes. Native iOS and Android apps are available.

**Can I import from Trello or Asana?**
Yes. One-click import from Trello, Asana, Jira, and Monday.com.

**What is the difference between Kanban and list view?**
Kanban shows tasks as cards on a drag-and-drop board organized by
status. List view shows tasks in a traditional to-do list format.
Both views show the same tasks.

## Key Pages

- [Homepage](https://acme.example.com): Product overview and free trial
- [Pricing](https://acme.example.com/pricing): Plans and pricing details
- [Features](https://acme.example.com/features): Full feature list
- [Blog](https://acme.example.com/blog): Remote work and productivity guides

## Contact

[email protected]

Full Example: Blog or Content Site

# The Remote Work Digest

> Weekly guides on remote work, productivity, and async communication

## About

The Remote Work Digest publishes practical guides for people who work
remotely or manage remote teams. Topics include async communication,
home office setup, remote hiring, and distributed team management.
New articles published every Tuesday and Thursday.

## Content Areas

- Async communication strategies and tools
- Home office setup and ergonomics
- Remote hiring and onboarding
- Distributed team management
- Time zone coordination
- Remote work tools and software reviews

## FAQ

**How often is new content published?**
New articles go live every Tuesday and Thursday.

**Who writes the articles?**
Articles are written by our team of remote work practitioners with
5 to 15 years of experience working and managing fully distributed teams.

**Is there a newsletter?**
Yes. Weekly digest delivered every Friday with the week's top articles.

## Key Pages

- [Homepage](https://remoteworkdigest.example.com): Latest articles
- [Newsletter](https://remoteworkdigest.example.com/newsletter): Subscribe
- [About](https://remoteworkdigest.example.com/about): Who we are

## Contact

[email protected]

Full Example: Agency or Services Site

# Brightline Digital

> SEO and content marketing agency for B2B SaaS companies

## About

Brightline Digital is a specialized SEO and content marketing agency
focused on B2B SaaS companies. We help SaaS businesses grow organic
traffic, improve AI search visibility, and build topical authority
in their target markets. Based in Austin, TX with remote team members
across Europe and North America.

## Services

- Technical SEO audits and implementation
- AI SEO and GEO optimization
- Content strategy and production
- Link building for SaaS
- SEO for product-led growth companies

## FAQ

**What size companies do you work with?**
We work with B2B SaaS companies at Series A through Series C stage,
typically with 20 to 200 employees.

**Do you offer one-time audits or ongoing retainers?**
Both. We offer standalone audits starting at $2,500 and monthly
retainers starting at $4,000/month.

**How long until we see results?**
Most clients see measurable traffic improvements within 3 to 6 months.
AI search citation improvements typically show within 4 to 8 weeks.

## Key Pages

- [Services](https://brightline.example.com/services): What we offer
- [Case Studies](https://brightline.example.com/case-studies): Client results
- [Contact](https://brightline.example.com/contact): Get a proposal

## Contact

[email protected]

llms.txt vs llms-full.txt: What Is the Difference?

Some sites create two versions:

llms.txt is the concise version. It is the quick-read summary that AI crawlers can parse in seconds. Keep it under 100 lines. Focus on the essentials.

llms-full.txt is the detailed version. It includes expanded feature documentation, longer FAQ sections, full content summaries, technical architecture details, and anything else that gives AI models deeper context about your site.

The common pattern is to include a reference to llms-full.txt within your llms.txt:

## Extended Documentation

For detailed documentation, see: https://yoursite.com/llms-full.txt

Most sites do not need llms-full.txt. Start with llms.txt. Add llms-full.txt only if you have significantly more context to give AI crawlers that does not fit in the concise version.

How to Upload and Verify

Once you have created your llms.txt file:

1. Save it as llms.txt (plain text, not llms.txt.doc or llms.md)
2. Upload to your site root via FTP, cPanel File Manager, your CMS media uploader, or deploy it as a static file alongside your other root files
3. Verify access: Open yoursite.com/llms.txt in your browser. You should see the plain text content directly
4. Check the content type: The server should return Content-Type: text/plain. You can check with curl: curl -I yoursite.com/llms.txt

Use the Free Generator

If you would rather fill out a form than write markdown manually, use the free llms.txt Generator. Enter your site name, description, features, FAQ entries, and key pages. Click Generate. Copy the output. You are done.

The generator formats everything correctly and includes all the standard sections. It takes about 5 minutes to fill in and produces a properly formatted llms.txt you can upload immediately.

How AI Crawlers Actually Read and Use llms.txt Today

Here is the honest version, because a lot of llms.txt advice oversells it. As of mid 2026, no major AI vendor has published documentation that says "we fetch llms.txt and weight it in answers." There is no public confirmation from OpenAI, Anthropic, Google, or Perplexity that the file changes how their models rank or cite you. Treat anyone who promises a citation boost from llms.txt alone with suspicion.

What we can say with confidence is narrower. Plenty of sites have llms.txt files that get fetched. Server logs from teams who run docs sites show AI user agents and aggregator bots requesting the file, sometimes alongside the page they are already crawling. Tools and frameworks built for AI agents increasingly look for it. The file is read by real software. The open question is what each system does with the contents after it reads them.

The realistic case for llms.txt in 2026: it is a cheap, machine-readable summary that some crawlers fetch. The speculative case is that it directly lifts your citation rate. Build it for the first reason, not the second.

Two things are likely true based on how retrieval systems work. First, a clean summary of your pages helps any crawler that wants a fast map of your site, the same way a sitemap helps a search bot. Second, the links you list in your Key Pages section act as a curated shortlist. If a crawler trusts that list, you have nudged it toward the pages you actually want quoted instead of a random tag archive.

What stays speculative: ranking weight, citation frequency, and whether models read the file at answer time versus crawl time. Do not write copy or restructure your site around guesses. Keep llms.txt accurate and small, treat it as one signal among many, and put most of your effort into the on-page content that AI systems definitely do read. If you want a baseline of what AI bots can reach on your domain, run an AI crawler check first, then layer llms.txt on top.

Common llms.txt Mistakes That Make Yours Useless

Most broken llms.txt files fail in the same handful of ways. Here is what to check before you call yours done.

Wrong location. The file has to sit at the domain root, yoursite.com/llms.txt. If it ends up at yoursite.com/docs/llms.txt or inside a subfolder, crawlers that look for the convention will not find it. Subdomains are their own root, so blog.yoursite.com needs its own file if you want one there.

Wrong content type. Some servers serve the file as text/html or force a download instead of returning text/plain. If your browser tries to download llms.txt instead of showing it, fix the server config. Run curl -sI https://yoursite.com/llms.txt and confirm the response says text/plain.

Marketing voice instead of plain facts. This is the big one. llms.txt is not a landing page. Lines like "we deliver world-class solutions for forward-thinking brands" tell a model nothing. Write what you do, who it is for, what it costs, and what makes it different, in plain declarative sentences. A crawler cannot quote a slogan as a fact.

• Dead or redirecting links in Key Pages. Every URL should return a 200. Stale links waste the one section that points crawlers at your best content.
• Stuffing the whole site in. A 4,000-line wall of every page defeats the purpose. Keep llms.txt tight and move depth to llms-full.txt.
• Contradicting your live pages. If llms.txt says pricing starts at $9 and your pricing page says $19, you have handed an AI a reason to distrust both.
• Listing pages you block in robots.txt. Do not advertise URLs you have told crawlers to skip. Keep the two files consistent.

Set-and-forget. A file written in 2024 that still names a product you sunset is worse than no file. We cover the update routine below.

The test for any line in llms.txt: could a model repeat it as a fact and be correct? If not, cut it or rewrite it as something concrete.

llms.txt for Ecommerce, Docs, and News Sites

The SaaS, blog, and agency examples above cover a lot of sites. Three site types need a different shape.

Ecommerce. You cannot list every SKU, and you should not try. Summarize the store at the category level instead. Say what you sell, who you sell to, your shipping and returns policy in one line each, and link to your main category pages, not individual products. AI shopping answers care about brand, range, price band, and policies. Give those plainly. A toy store line might read: "We sell wooden and educational toys for ages 0 to 8, ship free over $40 in the US, and accept returns within 60 days." That is quotable. A list of 900 product URLs is not.

Docs or knowledge base. This is where llms.txt shines, because the format came out of the docs world. List your main doc sections with one line each: getting started, API reference, authentication, webhooks, billing. Point to the canonical version of each, not every nested page. If your docs are large, this is the clearest case for a paired llms-full.txt that inlines the most-asked reference material so a model can answer setup and API questions without crawling forty pages.

• Lead with the product or library name and what problem it solves.
• Name the current major version so models do not answer with deprecated syntax.
• Link the changelog so crawlers can see what is recent.

News or publisher. Your content changes daily, so do not try to summarize articles in the file. Summarize the publication: what beats you cover, who your writers are, your editorial standards, and how often you publish. Link your section fronts (politics, tech, local) and your about and standards pages. AI systems weigh source credibility, so a clear statement of who you are and how you report carries more value here than any single headline. Keep the file stable and let your sitemap and feeds carry the fast-moving content.

Whatever your type, the free llms.txt Generator gives you a correct skeleton in a couple of minutes. Fill in the form, then trim it to match the patterns above.

How to Keep llms.txt Updated as Your Site Changes

An llms.txt file is only as good as the day it stops being true. Sites change. You launch a product, change pricing, kill a feature, move a key page. Every one of those is a chance for your file to start lying to crawlers, and a confident wrong answer about your own business is worse than no file at all.

You do not need a heavy process. You need a trigger and a quick review. Tie the trigger to events you already track:

• Pricing changes. The moment your pricing page changes, update the price lines in llms.txt the same day. This is the field most likely to go stale and the one most likely to get quoted.
• Product or feature launches. Add the new thing, and remove anything you sunset. A file that still lists a retired plan is a credibility problem.
• URL changes. If you restructure and a Key Pages link now 404s or redirects, fix it. Broken links here waste your best section.
• Positioning changes. If your one-line tagline no longer matches how you describe yourself on the homepage, sync them.

For most sites, a calendar reminder once a quarter to read the file top to bottom is enough on top of those event triggers. It takes five minutes. Open the live file in a browser, read it as if you were a stranger, and ask whether every line is still accurate.

Treat llms.txt like a dependency, not a one-time deliverable. The cheapest version: review it the same day you change pricing, ship a product, or move a URL, plus a five-minute quarterly read-through.

One practical tip: keep llms.txt in version control next to your code, not as a file someone hand-edits on the server and forgets. When it lives in your repo, updating it becomes part of the same pull request that ships the change, which is the only update process that actually survives contact with a busy team. If you regenerate it, the llms.txt Generator makes a fresh, correctly formatted version fast, so refreshing it is never the bottleneck.

FAQ: llms.txt Questions