AI search
llms.txt Validator
Check the file is well formed, nothing more
Paste a URL or a file and get a line-by-line parse of llms.txt: the H1 name, the summary blockquote, every section and link under it, the entries whose URLs no longer resolve, and the lines a parser cannot read, with the reason for each.
Get early accessWhat it returns
- The parsed structure: the H1 project name, the blockquote summary, each H2 section and the links listed under it in order.
- A status code for every link in the file, so entries pointing at moved or deleted pages are named rather than counted.
- Lines that do not fit the format, quoted with the reason, since the file is Markdown and one wrong heading level changes the shape of everything beneath it.
- A plain statement of what a pass means: Google has said llms.txt has no effect on Google Search, AI Overviews or AI Mode, so a clean parse means the file is well formed and nothing else.
When this matters
A proposal lands asking for llms.txt because a competitor published one. The honest answer is that it will not change anything on Google: Google has stated publicly that the file has no effect on Google Search, on AI Overviews or on AI Mode. If you are shipping it regardless, because it costs an afternoon and a handful of other tools read it, then the only thing worth checking is that it is actually well formed and that the URLs inside it still resolve. That is the entire job, and it is a small one, which is why it should be free.
How to answer this today
AI visibility answers the question llms.txt is usually adopted to answer: whether answer engines mention you, measured by running your prompts and recording the domains that appear in the sources. It is measurement of what the engines did rather than a declaration you wrote about yourself, which is why it can tell you something a file on your own server cannot.
AI visibilityWhy this is harder than it looks
No search engine has committed to it
llms.txt is a proposal, not a standard anyone is obliged to read. Google has said directly that it has no effect on Google Search, AI Overviews or AI Mode. Any tool that frames validation as an optimisation step is selling a ranking benefit nobody has demonstrated.
There is nothing to conform to
The proposal is loose about what belongs under each heading and there is no conformance suite. A validator can check the Markdown shape, the heading hierarchy and the links. It cannot tell you whether the content you chose to list is the right content, because no definition of right exists.
Hand-maintained link lists rot
The file is a curated list written once and rarely revisited, while the site keeps changing. The common failure is a file that parses perfectly and points at a dozen pages that have moved, which is worse than not having one, so link status is the part of the check that keeps earning its keep.
It is not an access control
llms.txt grants nothing and denies nothing. Which agents may fetch your pages is decided by robots.txt and by whatever rules run at your edge. Treating the file as a permission layer is a category error, and it is the mistake most likely to be made by whoever approved adding it.