Sigma Logic AI Lead with AI. Thrive with Innovation.

What should our llms.txt contain?

Enter the name, one summary sentence, and the pages that matter grouped into sections. The file comes out in the exact format the llms.txt proposal specifies: an H1, a blockquote, and H2 sections of links with a one-line note each. Copy it to the root of your site as /llms.txt. Nothing you type leaves your browser.

Your site

Sections

One link per line, in the form Title | https://url | one-line note. The note is optional. Put secondary pages in a section called Optional, which the spec reserves for links an agent may skip.

Waiting for a name

The H1 is the only required part of the file. Everything else is optional, and the sections are what make it useful.

Links 0 across 0 sections
Lines needing attention 0 not absolute URLs, or malformed
File size 0 bytes

Your llms.txt

# 
Download llms.txt

The output follows the llms.txt proposal: H1, blockquote, free text, then H2 sections of - [name](url): notes. Your entries are kept in this browser so a reload does not lose them, and are never sent anywhere.

What the file is, and what it is not

llms.txt is a proposal from September 2024 for a markdown file at the root of a site that gives a language model a curated index of the pages that matter. It is not a standard, and as of September 2026 Google has said it does not use it and no major assistant provider has committed to it as a signal. Some developer tools and agents do fetch it, mostly from documentation sites. The honest reading is that it is cheap, harmless and unproven, which is why this generator exists and why nobody should pay for the file as a project.

Where to put it

At /llms.txt on your domain, served as plain text. A file at a subpath such as /docs/llms.txt covers the URLs beneath it, and an agent should use the most specific one that applies.

What makes one useful rather than decorative

  • Curate. Ten links with a note each beat eighty links with none. The note says what a reader gets on that page.
  • Absolute URLs. The file is read away from your site, so a relative path means nothing. The generator flags any line without https://.
  • Use the Optional section. An agent with a small context window is allowed to skip it. Press, careers and history go there.
  • Keep it true. A file that drifts from the site misdescribes you. If the site is built from a content system, generate the file from that instead, and put a review date on this one if you cannot.

llms-full.txt

A companion convention inlines the full text of every linked page into one file. Coding assistants use it for API documentation. For a services or product site it is a duplicate of the site in a second format and nothing reads it, so this generator does not produce one.

The evidence on what llms.txt does and does not do, and where it sits against structured data and crawlability, is in llms.txt and structured data: what actually matters.

Let's talk

The file is the easy part

Whether an assistant cites you depends on crawlability, page structure and third-party presence far more than on this file. The ai search visibility (geo) engagement measures where you stand before changing anything.