We read your site and write the file, using your real titles, descriptions and structure. Nothing is invented and nothing is a template.
llms.txt is a plain text file at the root of your site that tells an AI assistant what you are and where the important pages live. It is a young convention and no assistant is required to honour it, which is exactly why it costs nothing to have a correct one.
Your titles, your descriptions, your sections. If a page has no description, it is left out rather than filled with a guess.
Copy it, save it as llms.txt, and put it at the root of your domain so it answers at yoursite.com/llms.txt.
It is a starting point built from what you already publish. Trim anything you would not point an assistant at.
No assistant is obliged to read llms.txt, and none of them promise to. This gets you a correct file; adoption is theirs to decide.
They do different jobs. robots.txt controls access. llms.txt describes what is there. You want both to agree.
Each one answers a different question about the same site.