Short answer: no, your site doesn't need an llms.txt file, and skipping it won't hurt your Google rankings or break anything. But if you run a docs-heavy site, a knowledge base, or a product you'd genuinely like AI tools to summarise correctly, a well-made llms.txt is a cheap, low-risk way to point them at the pages that matter.
The catch is that llms.txt has picked up a lot of hype in the last year, and now Lighthouse has started auditing it — with one rule that trips up almost everyone: your links must be Markdown links, not bare URLs. Get that wrong and the audit fails even though your file looks fine.
Here's what the file actually does, when it's worth your time, and a template you can paste in and edit in five minutes.
What is an llms.txt file, really?
An llms.txt file is a plain-text file at the root of your domain that lists your most important pages in a clean, Markdown-formatted menu, so language models can find and read them without wading through your whole site.
Think of it as a curated table of contents written for machines that read text, not HTML. It lives at yourdomain.com/llms.txt, the same way robots.txt does. The format is deliberately simple: an H1 with your site name, an optional blockquote summary, then sections of Markdown links to your best pages.
It is not a standard anyone is forced to obey. No search engine ranks you higher for having one. No AI company has promised to read it. It's a proposal — a convention some tools have chosen to support — and that honest framing matters when you decide whether to bother.
Does it actually do anything in 2026?
Right now llms.txt helps in narrow, specific cases and does nothing measurable in most others — so treat it as a small optimisation, not a ranking lever.
The signals that made people pay attention this year are real but modest. Lighthouse added an Agentic Browsing audit that checks for the file. Some developer-doc platforms and AI coding assistants now look for it to pull accurate reference material. And there's an ongoing debate about a companion "LLMs-Author.txt" idea for attribution. None of that adds up to "you're falling behind without one."
Where it genuinely earns its keep:
- Documentation and API references — you want an AI assistant quoting the correct endpoint, not a two-year-old blog post.
- Products with a lot of similar-looking pages — the file steers tools to your canonical explanation instead of a landing page.
- Sites where getting the facts right matters — pricing, policies, support hours, licence terms.
Where it's mostly pointless: a five-page brochure site, a small local business, or a blog that's already easy to crawl. If a model can understand your site in one pass, curating a menu for it changes little.
The Lighthouse markdown-link gotcha
The single most common reason an llms.txt file fails Lighthouse's audit is that the links are written as plain URLs instead of Markdown links — and the fix takes two seconds once you know it.
Lighthouse's Agentic Browsing check doesn't just want links present; it wants them in Markdown syntax with descriptive anchor text. A bare URL on its own line is invisible to the audit even though a human can read it perfectly well.
Here's the difference that decides pass or fail:
| Fails the audit | Passes the audit |
|---|---|
| https://yoursite.com/docs/setup | [Setup guide](https://yoursite.com/docs/setup) |
| See our pricing: yoursite.com/pricing | [Pricing and plans](https://yoursite.com/pricing) |
The rule of thumb: every link is [human-readable label](full-URL), on its own bullet line, ideally with a short note after it. Use absolute URLs, not relative paths — machines fetching the file may not know your base domain. And don't pad the file with a link dump; a focused list of 10–30 genuinely useful pages beats a mirror of your whole sitemap.
A copy-paste llms.txt template
The safest starting point is the standard structure: an H1 title, a one-line summary in a blockquote, then grouped sections of Markdown links with short descriptions.
Copy this, swap in your own pages, and save it as llms.txt in your web root:
- # Your Site Name
- > One sentence on what your site does and who it's for.
- ## Core pages
- - [What we do](https://yoursite.com/): plain overview of the product or service.
- - [Pricing](https://yoursite.com/pricing): current plans and what's included.
- - [Contact and support](https://yoursite.com/support): how to reach a human.
- ## Documentation
- - [Getting started](https://yoursite.com/docs/start): first-run setup steps.
- - [API reference](https://yoursite.com/docs/api): endpoints and auth.
- ## Optional
- - [Changelog](https://yoursite.com/changelog): recent updates.
A few things worth getting right: keep descriptions to one line, list your most important sections first, and only include the "## Optional" heading for links that are nice-to-have — some tools deliberately skip that section to stay focused. Once it's live, open yourdomain.com/llms.txt in a browser to confirm it serves as plain text and isn't being redirected or gzipped into something odd.
If you host with TPC, you can drop the file straight into your public_html root via File Manager or SFTP — no config change, no plugin. It's served like any other static file.
Should you spend time on this at all?
Spend 15 minutes on llms.txt if you have documentation or facts you want AI tools to get right; skip it if you run a small brochure site, because the payoff won't be there yet.
Be honest about the trade-off. This is speculative optimisation — you're betting that more tools will read the file over the next year, and the evidence points that way, but nobody's guaranteed you traffic from it. The good news is the downside is basically zero: a static text file can't slow your site, leak data, or confuse Google. So the decision is purely about whether the upside is relevant to you.
Quick gut-check before you write one:
- Do you have pages where being quoted accurately actually matters (docs, pricing, policy)? → Worth it.
- Is your site small and already easy to read? → Skip it, revisit in six months.
- Are you tempted to list every page? → Don't; curation is the whole point.
One more thing people forget: an llms.txt file only controls what you point to. It doesn't hide anything or block crawlers — that's still robots.txt's job. If you want certain pages left out of AI training or crawling, handle that separately and don't rely on omitting them here.
Keeping it maintained without the busywork
The one habit that keeps llms.txt useful is updating it whenever a URL in it changes, so a stale file with dead links isn't worse than no file at all.
You don't need automation for a handful of links. Add a line to whatever checklist you use when you rename or retire a page, and re-open the file in a browser once a quarter to click through. If you generate docs from a static site builder, some plugins can produce llms.txt from your existing navigation — handy, but check the output, because most of them dump every page and blow past the "keep it focused" principle.
If a page you listed moves, fix the URL in llms.txt at the same time you set up the redirect. Broken links in a file whose entire purpose is accuracy send exactly the wrong signal. On TPC's plans your real engineers are on support 24/7 if you get stuck placing or serving the file, and free migration means moving in doesn't mean re-doing this kind of housekeeping from scratch.
FAQ
Where does the llms.txt file go?
It goes in your website's root directory, so it's reachable at yourdomain.com/llms.txt. That's the same location as robots.txt — upload it to public_html (or your web root) and confirm it loads in a browser as plain text.
Will llms.txt improve my Google rankings?
No, llms.txt has no effect on search rankings. Google's crawlers don't use it to rank pages; its purpose is to help AI tools and agents find and read your key content accurately, which is a separate goal from SEO.
Why does Lighthouse fail my llms.txt file?
The most common cause is using bare URLs instead of Markdown links. Lighthouse's Agentic Browsing audit expects the format [descriptive label](https://full-url), with each link on its own line — plain URLs won't pass even though they look fine to a person.
Is llms.txt an official standard?
No, it's a proposed convention, not a mandated standard. No search engine or AI company is required to read it, but support is growing among documentation platforms and AI assistants, which is why it's a low-risk file to add if accurate machine-reading matters to you.
Does llms.txt block AI crawlers from my site?
No, it only points tools toward the pages you list — it doesn't hide or restrict anything. To control crawling or AI training access, use robots.txt and any provider-specific opt-out mechanisms instead.
