llms.txt Examples You Can Copy Today
Contents
A real llms.txt file is a plain Markdown document, usually hosted at yourdomain.com/llms.txt, that opens with an H1 project name, a one or two sentence blockquote summary, and H2 sections of curated links. Stripe, Cloudflare, and Anthropic already publish one. Below are their exact, live file structures, checked in July 2026, plus a template you can adapt today.
If you are still deciding whether llms.txt is worth setting up at all, the definition and background is worth reading first. This article assumes you already know the basics and want to see what real companies actually publish right now, verified directly rather than paraphrased from someone else's screenshot.
What does a real llms.txt file actually look like?
A valid llms.txt file has three required parts: an H1 with your project name, a blockquote summary right below it, and H2 sections grouping curated links with short descriptions.
That structure comes from the specification hosted at llmstxt.org, the same page Google's AI Overview cites when someone searches this exact keyword (source: llmstxt.org, 2026). The spec also defines an optional "## Optional" section: a place for links that are useful but not essential, so a system with a limited context window can skip them without losing the core picture of a site. Structure matters more than length. A 200-word file with three clearly labeled sections is more useful to an AI system than a 2,000-word file with no headings at all.
The specification also defines a second, optional file: llms-full.txt, a complete export of a site's content in one long Markdown document, instead of the curated-links index that plain llms.txt provides (source: llmstxt.org, 2026). It suits a narrower use case, typically a small documentation site where dumping full page content into a single file still fits inside a reasonable context-window budget. For most sites, the short index version covers what an AI system actually needs.
llms.txt is one of several machine-readability signals worth setting up together. It works through a plain-text file, while structured data built for AI systems works through JSON-LD markup instead, but both aim at the same goal: giving machines a faster path to understanding what a page or a site actually contains.
What does Stripe's real llms.txt file contain?
Stripe's live llms.txt opens with the H1 "Stripe Documentation," followed by a blockquote warning LLMs to check npm for the latest package version instead of relying on memorized numbers, then more than twenty H2 sections covering the full product line, from Payments and Billing to Connect, Tax, and Capital.
Verified directly at docs.stripe.com/llms.txt in July 2026, the file's blockquote does something unusual for a company overview: instead of a tagline, it tells any AI system reading it to verify the current package version on npm before trusting a number it might have picked up during training. That is a defense against a documented failure mode, language models confidently citing outdated software versions, and it signals real engineering intent rather than a copy-pasted template. Rather than a single link list, 25 H2 sections follow, starting with "Docs" and running through Payment Methods, Checkout, Payments, Billing, Connect, Issuing, Capital, Tax, Invoicing, and more, down to an "Optional" section for support and changelog links.
What does Cloudflare's real llms.txt file contain?
Cloudflare's llms.txt uses the H1 "Cloudflare Developer Documentation," a blockquote explaining that each product links to its own dedicated llms.txt, and nine H2 sections grouping more than 100 of those product-specific files by category.
Checked live at developers.cloudflare.com/llms.txt in July 2026, Cloudflare takes a federated approach rather than one flat file. Its blockquote tells readers that each product area, from Workers to R2 to DNS, maintains its own separate llms.txt, and the root file mostly points to them. Nine H2 sections, covering more than 100 individual product llms.txt files in total, organize these links by category. That approach makes sense at Cloudflare's scale: a single file trying to cover every product would blow past most practical context-window budgets, so splitting by product keeps each individual file small and genuinely useful.
What does Anthropic's llms.txt contain, and why did the URL change?
Anthropic's llms.txt has moved from docs.claude.com to platform.claude.com, and its current version skips the blockquote summary that the llms.txt spec calls for, showing that even reference implementations drift over time.
Most existing articles about llms.txt examples still point readers to docs.claude.com/llms.txt, because that was the correct address when those articles were published. As of July 2026, that URL returns a 301 redirect straight to platform.claude.com/llms.txt. The file now living there opens with the H1 "Anthropic Developer Documentation" but has no blockquote at all, jumping straight to an H2 labeled "Root URL." That is a genuine deviation from what the spec recommends, and it is a useful reminder on its own: an example captured once and never rechecked can go stale quietly, which is exactly the kind of drift this article was built to catch.
How do Stripe, Cloudflare, and Anthropic's llms.txt files compare?
Side by side, the three verified files reflect three different philosophies: a version-safety warning, a federated multi-file system, and a stripped-down index with no summary at all. The table below lists exactly what each file contains, checked live in July 2026, not pulled from an older screenshot.
| Company | URL | Has a blockquote? | Structure pattern | Notable feature |
|---|---|---|---|---|
| Stripe | docs.stripe.com/llms.txt | Yes | Single index file, 25 H2 sections by product | Warns AI systems to verify the current package version on npm rather than rely on memorized training data |
| Cloudflare | developers.cloudflare.com/llms.txt | Yes | Federated: root index groups more than 100 product-specific llms.txt files into nine H2 categories | Splits content by product line to stay within practical context-window limits |
| Anthropic | platform.claude.com/llms.txt | No | Single index file, no summary | URL migrated from docs.claude.com via a 301 redirect; current version drops the blockquote the spec recommends |
This article, Blue could have written it for you: content optimized for Google + AI, without you writing a single word.
Try mentionLABDoes llms.txt actually help you get cited by AI, or is it mostly ignored?
Most llms.txt files go unread: an analysis of 137,000 domains found that 97% of the roughly 38,000 published files got zero requests in May 2026, and Google's own Search Central guidance says the files aren't needed to appear in Google's generative AI features.
Adoption itself is not the bottleneck. An analysis found that 28% of the 137,000 domains in its sample already publish an llms.txt file (source: an independent analysis, 2026). The gap is what happens after publishing. Among the roughly 38,000 valid files identified in that sample, 97% received zero requests during May 2026. Of the small share that did get requested, AI retrieval bots such as ChatGPT and Perplexity accounted for only about 1% of that traffic, SEO audit tools made up 21%, unidentified bots 14%, and classic crawlers like Googlebot 13% (source: an independent analysis, 2026).
Google has stated its own position clearly. Its Search Central guidance, updated June 29, 2026, says sites do not need new machine-readable files, AI text files, markup, or Markdown to appear in Google Search, including its generative AI features, because Google Search itself does not use them (source: Google Search Central, 2026). John Mueller has compared llms.txt to the old meta keywords tag: no AI service has confirmed using it, and server logs show most of them do not even request the file (source: Search Engine Journal, 2026).
There is one exception worth flagging. Google added a check for llms.txt presence to Chrome Lighthouse's Agentic Browsing audit category on May 5, 2026, then promoted it from experimental to default status in Lighthouse 13.3.0 two days later, on May 7, 2026 (source: Search Engine Land, 2026). That is not Google Search using the file, it is Chrome's agentic-browsing tooling, built for AI agents that navigate a browser on a user's behalf, checking whether the file exists at all.
Structure, sourcing, and freshness move AI citations far more reliably than any single file at your domain root. For the fuller picture of what actually earns citations, see how to get your content cited by AI.
Should you still create an llms.txt file?
Yes, if it takes under an hour: llms.txt costs little to maintain and can't hurt your site, even though no major AI provider has confirmed using it for citations yet. Treat it as a low-cost addition, not a ranking or visibility strategy on its own.
The realistic case for llms.txt is modest but real. It costs almost nothing to maintain, it cannot hurt indexing or rankings, and Chrome's agentic-browsing audit now checks for it. None of that adds up to a visibility strategy by itself. If clear documentation or a short list of priority pages already exists, publishing a file is a reasonable hour of work. If the goal is actually moving AI citations, structure, verifiable sourcing, and topical depth do far more of that job than a text file at the domain root.
If it is worth doing, the step-by-step process for building your own file covers the format rules and common mistakes in more detail than the template below.
What is a good starting template for your own llms.txt file?
A minimal, spec-compliant llms.txt needs four elements: an H1 with the project name, a one-sentence blockquote summary, two or three H2 sections of grouped links, and, optionally, one more H2 for lower-priority links. Here is a structure to copy and adapt directly.
# Your Company Name
> One or two sentences describing what your company or product does,
> written the way you would explain it to someone unfamiliar with it.
## Docs
- [Getting Started](https://yourdomain.com/docs/getting-started): Setup and first steps
- [API Reference](https://yourdomain.com/docs/api): Full endpoint documentation
## Product
- [Pricing](https://yourdomain.com/pricing): Plans and billing details
- [Changelog](https://yourdomain.com/changelog): Recent updates and releases
## Optional
- [Blog](https://yourdomain.com/blog): Articles and long-form content, lower priority for context-limited readers
Keep descriptions short and specific instead of promotional. An AI system reading this file is deciding which link to follow next, not evaluating marketing copy. Update the file whenever a major section of the site changes; a stale llms.txt, as Anthropic's own file shows, is worse than an honest, current one.
Frequently asked questions about llms.txt examples
Is llms.txt an official web standard?
No. It is a proposed convention introduced by Jeremy Howard of Answer.AI, published September 3, 2024, not an official standard ratified by the W3C or any other standards body (source: llmstxt.org, 2026). John Mueller has echoed this, noting that no AI provider has confirmed using it as a ranking or citation signal (source: Search Engine Journal, 2026).
Does having an llms.txt file help SEO or Google rankings?
No, not directly. Google's Search Central guidance, updated June 29, 2026, states plainly that sites do not need new machine-readable files, including llms.txt, to appear in Google Search or its generative AI features, because Google Search does not use them (source: Google Search Central, 2026).
Where should I host my llms.txt file?
At the root of the domain, for example yourdomain.com/llms.txt, exactly like robots.txt and sitemap.xml. That location comes from the original specification (source: llmstxt.org, 2026), and it is where Stripe, Cloudflare, and Anthropic all host theirs.
Do ChatGPT, Claude, or Perplexity actually read llms.txt files?
Rarely, based on available server-log data. An independent analysis found that AI retrieval bots accounted for only about 1% of the already-small volume of requests these files receive, with most of the remaining traffic coming from SEO audit tools and unidentified bots rather than AI assistants (source: an independent analysis, 2026).
The three files above make the same point in different ways: llms.txt is cheap to build and easy to let go stale. Publish one if it is useful, then verify it against the live spec every few months the way this article did, rather than trusting a screenshot from last year.
Blue handles your SEO and your GEO. On autopilot.
You approve, she produces content optimized for Google + AI.
Join the Lab · 5-day trial