---
title: "llms.txt Validator: Five Checks You Can Run With curl"
url: https://hostmy.blog/llms-txt-validator-wordpress/
date: 2026-10-28
modified: 2026-10-29
lang: en
author: "Aditya Sharma"
description: "There is no official llms.txt validator. Here are five checks you can run with curl, what each proves, and what a passing file still does not tell you."
categories:
  - "RankReady"
image: https://hostmy.blog/wp-content/uploads/2026/10/hmb-card-2630-1024x538.jpg
word_count: 1325
---

# llms.txt Validator: Five Checks You Can Run With curl

To validate llms.txt with curl, you run five checks, and that short list is all an llms.txt validator is. We know of no official validator, and of no search engine or assistant that says it will reject a file that fails. The checks are useful for a plainer reason: a broken llms.txt is a small piece of untidiness you published by mistake, and it takes five minutes to find.

This post gives the five checks, with the command for each, and says what a passing result does and does not prove. It does not argue about whether the file is worth having, since [llms.txt does not work, and we generate it anyway](/llms-txt-does-not-work/) already does that. Nor does it cover how to create one, which is in [how to generate an llms.txt on WordPress](/llms-txt-generator-wordpress/). It is about checking one you already have.

## What the proposal says

The format comes from the llms.txt proposal at [llmstxt.org](https://llmstxt.org/), written by Jeremy Howard of Answer.AI and first published in September 2024. It describes a Markdown file at /llms.txt on your site root, or at a subpath where it covers URLs beneath it. The structure is specific.

| Part | Rule |
| ---- | ---- |
| H1 | The only required section, giving the name of the site or project |
| Blockquote | A short summary |
| Body text | Optional paragraphs or lists with more detail, but no headings |
| H2 sections | File lists, each item a Markdown link in the form [name](url), optionally followed by a colon and a note |
| Optional section | An H2 named Optional, for links an agent can skip when it needs less context |

A small example shows the shape.

`# Example Blog
> A blog about home baking, with tested recipes and equipment reviews.

Posts are written by one author and updated when a recipe changes.

## Recipes
- [Sourdough for beginners](https://example.com/sourdough/): A first loaf, step by step
- [Weeknight focaccia](https://example.com/focaccia/): Ready in three hours

## Optional
- [About the author](https://example.com/about/)`

The proposal also suggests clean Markdown versions of pages, at the same address with .md added, and ways to advertise them. A file called llms-full.txt is a common companion on many sites. It is not part of the proposal as published on llmstxt.org, so do not treat its absence as a fault.

## Check 1: it exists, returns 200, and is text

The first check is the plainest, and it catches more than you would expect. Many sites return a styled 404 page with a 200 status, which makes a missing file look present. Ask for the headers, and then for the first lines.

`curl -sI https://yourdomain.com/llms.txt | grep -iE '^(HTTP|content-type)'
curl -s https://yourdomain.com/llms.txt | head -5`

You want a 200 and a plain text or Markdown content type, and the first line should be a heading, not an HTML tag. If the first line starts with a doctype or a bracket, you are looking at a web page, and the file is not being served. The proposal does not set a content type, so what matters is that the body is text and not HTML.

## Check 2: the structure follows the format

The second check counts the parts. Run these three and compare them with the rules above.

`curl -s https://yourdomain.com/llms.txt | grep -c '^# '
curl -s https://yourdomain.com/llms.txt | grep -c '^> '
curl -s https://yourdomain.com/llms.txt | grep -c '^## '`

The first number should be 1, because there is one H1. The second should be at least 1 if you wrote a summary. The third counts your sections, and it should match what you meant to list. If the first number is 0, the one required part is missing. If it is more than 1, something is repeating the title, which usually means two plugins are writing to the same file.

For reference, this is what checks 1, 2 and 5 return on this site's own file.

![Terminal output of llms.txt curl checks on hostmy.blog: HTTP 200 text/plain, one H1, one blockquote, three H2s, 158 links](https://hostmy.blog/wp-content/uploads/2026/10/llms-txt-validator-curl-output.png)

## Check 3: every link works

A file full of dead links is worse than no file, because it points a reader to nothing. This reads the links out of the file and prints a status code for each one.

`curl -s https://yourdomain.com/llms.txt | grep -oE '\]\(https?://[^)]+\)' | sed -E 's/^\]\(//; s/\)$//' | while read u; do curl -s -o /dev/null -w "%{http_code} $u\n" "$u"; done`

Every line should say 200. A 301 means the link points at an old address, which should be updated in the file. A 404 means the page is gone, and the file should drop it. Redirects in the list are the common fault, and a redirect left behind by an old permalink change or a moved page is the usual source.

## Check 4: crawlers can fetch it

A file that loads for you may not load for a crawler. A security plugin, a host firewall or a CDN can refuse automated visitors while serving people normally. This request uses a crawler's user-agent string, taken from OpenAI's documentation, and shows what comes back.

`curl -sI -A "Mozilla/5.0 (compatible; OAI-SearchBot/1.4; +https://openai.com/searchbot)" https://yourdomain.com/llms.txt | head -5`

You want the same 200 you got before. A 403, or a redirect to a login or challenge page, means a rule is refusing that user agent, and the same rule is probably refusing it everywhere else too. The method and its limits are the same as in [OAI-SearchBot: the crawler behind ChatGPT search](/oai-searchbot-wordpress/). The test comes from your own address, so it cannot show a block aimed at published crawler ranges.

## Check 5: it matches the site today

The last check is about freshness. A file generated once and never updated drifts from the site. Look at what it lists, and compare it with what you have published.

`curl -s https://yourdomain.com/llms.txt | wc -w
curl -s https://yourdomain.com/llms.txt | grep -c '](http'`

The first number is the size in words, and the second is the number of links. Ask whether the links include your newest posts, whether they exclude pages you have removed or marked noindex, and whether the file is a sensible size. A file that is far larger than the useful part of your site, or that lists every tag page, is doing the opposite of what the format is for.

## What passing proves, and what it does not

Passing all five checks proves the file is well formed, reachable and current. It does not prove anything reads it. Google's guide to [AI features and your website](https://developers.google.com/search/docs/appearance/ai-features) says you do not need to create new machine readable files, AI text files or markup to appear in those features. It does not name llms.txt, so read it as a statement about Google Search only. We know of no commitment from the major assistants to use one. Treat the file as a tidy index that costs little to keep, not as a lever.

That is also the right way to decide how much time to spend. Five minutes on a check is proportionate. A day of tuning is not.

## Fixing what the checks find

| Result | Likely cause | Fix |
| ------ | ------------ | --- |
| 404 or an HTML page | Nothing is serving the file, or a rule intercepts it | Check the generating plugin, and the permalink and server rules |
| No H1, or two | The file is hand-edited or written by two sources | Make one source the owner |
| Dead links | The file is stale | Regenerate it, or remove the dead entries |
| 403 for a crawler | A firewall or security rule | See [your host may be blocking AI crawlers](/host-blocking-ai-crawlers/) |
| Far too large | It lists everything | Limit it to the pages worth reading |

## Where RankReady fits

RankReady generates an llms.txt and an llms-full.txt, with a section structure and a summary for each post, and it keeps them current as you publish. It also says it cannot promise rankings, citations or traffic, which is the honest position. It is free on WordPress.org and runs alongside Yoast, Rank Math, AIOSEO and SEOPress, including where Yoast writes an llms.txt of its own. The free [AI Readiness Checker](https://hostmy.blog/plugins/rankready/ai-readiness-checker/) runs 41 checks on your live site, and the five checks above are the manual way to look at one file.

If you ran the first command right now, would it return a text file, or a page that happens to say 200?