This week I ran curl on our agency site's /llms.txt, mostly out of curiosity. I didn't expect anything to be there, because none of us ever made one.
There was one. This is all of it:
# 8 Bit Enterprise ## Posts - [Hello world!](https://8bitenterprise.com/2026/11/03/hello-world/): Welcome to WordPress. This is your first post. Edit or delete it, then start writing! Our hosting plugin had generated it on its own. It uses our old brand spelling, and it lists exactly one post: the default WordPress "Hello world!" that we deleted weeks ago. None of our actual articles are in it.
So if any AI tool does read llms.txt, the one thing our site was "telling" it was wrong. That made me want to know whether anything reads these files at all.
What I found:
- Ahrefs looked at 137,210 sites for a study published in June 2026. About 28% published an llms.txt. Of roughly 38,000 valid files, 97% got zero requests in May 2026. The AI bot that showed up most in the few requests that did arrive was GPTBot.
- Google's position is blunt. Its May 2026 guide to AI search says: "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search." llms.txt sits on a short list of things Google says you don't need to do (I went through the rest of that myth list here).
- John Mueller compared llms.txt to the keywords meta tag: a claim a site makes about itself that crawlers have no reason to trust over the actual page.
Where I've landed for now: an llms.txt probably does nothing for search visibility, but a stale one is worse than none. If an agent ever does read it, it'll trust whatever's in there. I'm going to fix ours either way, because I'd rather it be accurate than wrong.
The part I can't figure out is the gap between how many people publish these files and how few requests they get. Either it's early, or it's cargo cult.
So, I'm curious:
- Do you ship an llms.txt? Did you write it, or did your host, CMS or plugin add one without telling you?
- Has anyone actually seen GPTBot, ClaudeBot or another agent fetch it in your server logs?
- If you work on docs sites: is it genuinely useful for coding assistants, even if search ignores it?
If you've never checked, run curl yoursite.com/llms.txt first. You might be as surprised as I was.


