llms.txt: a contents page for machines, not a ranking trick

Point at the pages that explain the work

List the homepage, the about page, and the guides that state your actual position. Give each link a one-line description. Leave out campaigns, thin tags, and pages you would not quote yourself.

Keep the file at the root and link it from robots.txt so a careful crawler can find it. Update it when you publish a cornerstone, not on every typo fix.

Do not use the file as a cloaking layer

The descriptions in llms.txt should match the pages. A file that claims services or credentials the site does not show is a second, less honest version of the business.

The file is optional context. Models will still read whatever they crawl. Blocking a crawler in robots.txt while praising the same bot in llms.txt is a confused policy.

• Curated links only.
• Descriptions that match the pages.
• A mention in robots.txt so the file is discoverable.

Keep the HTML strong anyway

If a guide is only summarised in llms.txt and the HTML is empty until JavaScript runs, you have given machines a map to a locked door. Prerender the page.

Review the file when a guide is retired so you are not advertising a 404 as a preferred source.

Does llms.txt improve rankings?

There is no evidence it is a ranking signal. It is a courtesy map for systems that choose to read it. The pages themselves still have to be good.

Where should llms.txt live?

At the root of the host, for example https://www.example.com/llms.txt, alongside robots.txt.