Do you need an llms.txt file?

Start with useful pages that can be read and maintained. An llms.txt file can provide a concise route to them, but it is not a substitute for the content or a promise of citations. Decide what the file is meant to help with, then keep it consistent with the pages it lists.

What does the file do?

llms.txt is a proposed Markdown overview linking to selected material for agents. It complements access rules and sitemaps; the linked pages carry the content.

A list can help a reader or a compatible tool find the relevant document. It cannot supply an answer missing from that document. If you provide Markdown versions, check that they preserve the visible meaning, qualifications and source links of the corresponding pages.

Choose a useful selection and write descriptions that identify what each linked page covers. Avoid adding claims to the index that the page does not support. When a page moves or its scope changes, update the link and description together.

What do Google’s own products say?

Google Search Central says no special AI text files or markup are needed for its AI features. That guidance concerns Google Search.

Chrome’s Lighthouse documents an agentic-browsing audit for llms.txt. It flags server errors; an absent file returning 404 is marked Not Applicable (N/A) because the file is currently optional.

These documents address different uses. An optional audit does not establish a requirement to appear in search answers. Check the guidance for the product you are actually trying to support, instead of treating all products from one company as the same system.

How can you find out whether it is used?

If you examine server logs, define the period, paths and method for identifying requests. A request to llms.txt tells you the file was requested. It does not by itself show that a page was used in an answer, that your business was cited or that a visit became an enquiry.

Equally, a small sample of requests cannot establish that no agent ever uses the file. Keep observations about file access, citations and referrals in separate columns. Use those records to decide what maintenance is worthwhile for your own site.

An index with stale links is a concrete problem you can fix. A forecast of citation growth from adding the filename alone needs evidence of its own. Ask which result a proposed piece of work will demonstrate and how it will be checked.

What should you maintain on the site?

Keep the underlying pages readable, accurate and clear about their scope.
Check that every listed address returns the intended document and language.
Keep HTML and Markdown wording aligned where both representations exist.
Update the index when titles, routes or content change.
Review generated files before publishing so private notes and unsupported claims stay out.

ErgoStrat uses an llms.txt index alongside separate Markdown versions of service pages and articles. The index points to the content; it does not replace it. Maintaining the two together matters more than producing a file once and forgetting the links behind it.

Questions

Do Markdown page versions replace llms.txt?

They serve different purposes: a page version contains the answer, while the index helps locate material. You can evaluate the value of each without assuming either is a universal platform requirement. Keep an alternate version accurate if you choose to provide one.

Should we commission only the index file?

Ask first whether the important pages are useful, accessible and current. A separate index task may make sense when a specific workflow needs it. It should come with a defined scope, link checks and an owner for updates, rather than a promised visibility increase from the filename.

Can a plugin generate the file?

Automation can help maintain a list. Review what it includes, whether exclusions and language versions are handled as intended, and whether it updates when pages change. A plugin’s popularity or rating does not establish a citation benefit. Test the generated output rather than accepting an unchecked list.

Sources

Check the content and access before the index

In an AI Visibility Report we check whether AI platforms’ bots can read your website at all, before we get to files like llms.txt.

Ask about AI search visibility

or write to contact@ergostrat.com