llms.txt

llms.txt deserves a clear-eyed entry because the hype around it routinely outruns the facts. The idea is sensible: models work within limited context windows, so a concise, curated manifest of what matters on your site could help them use it well, the way robots.txt and sitemap.xml became conventions for crawlers. Adoption tells a more modest story: documentation tools and some frameworks generate the file, thousands of sites publish one, but none of the major engines documents reading it, and no vendor promises it affects answers. Our honest position: it costs minutes, follows a plausible trajectory, and might help agents and tools that do read it, so adding one is a reasonable cheap bet, provided nobody books it as a win. Anyone selling llms.txt optimisation as a visibility breakthrough is selling ahead of the evidence. Spend the saved budget on what engines demonstrably read today: clean HTML and pages that answer the question.