Short answer
llms.txt is a markdown file at your domain root that gives AI crawlers a curated map of your key pages. Perplexity and Claude check it with some consistency, ChatGPT inconsistently, and Google largely ignores it. It is worth the 30 minutes to set up correctly, but it will not replace schema markup or content quality.
What llms.txt actually is
llms.txt is a proposed standard from developer Jeremy Howard, introduced in 2024, modeled loosely on robots.txt but built for a different purpose. Where robots.txt tells crawlers what they are allowed to access, llms.txt tells language models what your site actually contains and which pages matter most, in a clean, condensed format that is easier for a model to parse quickly than crawling your full HTML.
It is not a ranking file and it does not grant any special access. Think of it as a curated table of contents you hand to a model instead of making it guess your site structure from a sitemap or navigation menu.
Exact file structure
The format is deliberately simple markdown. A real example for a local service business, say a plumbing company in Tampa, looks like this:
# Tampa Bay Plumbing Pros > Tampa Bay Plumbing Pros is a licensed residential and commercial plumbing company serving Tampa, St. Petersburg and Clearwater, Florida since 2011. We specialize in emergency repairs, water heater installation, and drain cleaning, with same-day service available seven days a week. ## Core Pages - [Home](https://tampabayplumbingpros.com/): Overview of services, service area and contact information - [Emergency Plumbing](https://tampabayplumbingpros.com/emergency-plumbing/): 24/7 emergency repair details, response times and pricing structure - [Water Heater Installation](https://tampabayplumbingpros.com/water-heater-installation/): Installation and replacement services, brands carried, warranty terms - [Service Areas](https://tampabayplumbingpros.com/service-areas/): Full list of cities and neighborhoods served - [About](https://tampabayplumbingpros.com/about/): Licensing, insurance and company history ## FAQ - [Frequently Asked Questions](https://tampabayplumbingpros.com/faq/): Direct answers to common pricing, availability and service questions ## Optional - [Blog](https://tampabayplumbingpros.com/blog/): Maintenance tips and seasonal plumbing guidance
The structure has three parts: an H1 with your business name, a blockquote immediately after it that summarizes who you are and what you do in two or three sentences, and then one or more H2 sections grouping links with a short annotation explaining what each page covers. The annotations matter — a bare link list gives the model far less to work with than a link with one clear sentence about its content.
Where it goes and the llms-full.txt variant
The file lives at your domain root, exactly like robots.txt: yoursite.com/llms.txt. It is not placed in a subfolder and it does not need to be linked from anywhere on your site; crawlers that check for it look at the root path directly.
There is a companion convention, llms-full.txt, also at the root, which contains the full text content of your key pages concatenated into one file rather than just links and summaries. It is more thorough but more work to keep current, since it has to be updated any time the underlying page content changes. For most local businesses, a well-written llms.txt with accurate annotations covers the practical need; llms-full.txt is worth the extra maintenance mainly for content-heavy sites with dozens of important pages.
The verification step almost everyone gets wrong
This is the most common silent failure we see, and it is worth its own section because it makes the entire file useless without anyone noticing. Many modern websites, especially single-page applications and some website builders, are configured to serve a fallback route for any URL that does not match a defined page, rendering the site's normal HTML shell instead of a 404.
That means if you upload llms.txt but your server or framework has a catch-all SPA fallback, requesting yoursite.com/llms.txt can return a 200 status code with your homepage HTML instead of the actual text file. It looks like it worked because the request does not error out, but no crawler reading that response is getting your llms.txt content, they are getting your homepage markup with the wrong content-type.
To verify it is actually working, check two things: the file must be served with a content-type of text/plain (or text/markdown), not text/html, and the raw response body must be your markdown content, not your site's rendered HTML. You can check this quickly with a command-line request rather than a browser, since browsers sometimes mask the content-type in ways that hide the problem.
curl -I https://yoursite.com/llms.txt
If the content-type line reads text/html, the file is not actually being served correctly regardless of what a browser tab appears to show.
Honest expected impact
Adoption across engines is genuinely partial right now, and it is worth being direct about it rather than overselling the file. Perplexity and Claude check for llms.txt with reasonable consistency. ChatGPT's handling is inconsistent and appears to vary by query type and whether browsing is invoked at all. Google has stated it does not use llms.txt as a ranking or retrieval signal for its own AI features, and current evidence supports that it is largely ignored there.
Given that, llms.txt is a low-cost, low-risk addition that is worth doing once your schema, robots.txt and content are already solid, but it is not a substitute for any of those. If a client asks us to prioritize llms.txt over fixing broken schema or a robots.txt block, we push back, because the file with the actual leverage in 2026 is still your structured data and your crawlable, well-organized content.
Related questions
Does llms.txt replace robots.txt?
No, they serve different purposes and you need both. robots.txt controls crawler access; llms.txt provides a curated content map for models that choose to read it. Removing robots.txt in favor of llms.txt would be a mistake.
How often should I update llms.txt?
Update it any time you add, remove or significantly change a core page, and review it every few months even without changes, since stale links and outdated summaries reduce its usefulness.
Can I generate llms.txt automatically with a plugin?
Several CMS plugins and generators exist, but many produce bare link lists without meaningful annotations, which defeats the point. A hand-written file with real one-sentence descriptions per link outperforms an auto-generated one.
Will adding llms.txt hurt anything if I get it wrong?
A malformed file is generally just ignored rather than penalized. The real risk is the silent SPA-fallback failure described above, where you believe it is live but it is not being served at all.
Is llms.txt part of an official web standard?
Not yet. It is a widely discussed proposed convention that several major AI companies have acknowledged checking for, but it has not been formally adopted the way robots.txt has, so its handling can change as engines update their crawling behavior.
Not sure if your llms.txt or robots.txt is actually being served correctly?
Get your 55-page Pro Audit for $19 — delivered in 48 hours. Shows exactly where you stand in ChatGPT, Perplexity, Google AI Overviews and the Local Pack.
Keep exploring
Related services, industries, cities and resources from Local Visibility AI.