What it is
A Markdown file at /llms.txt containing an H1 with your site name, a blockquote summary, and grouped link lists with one-line descriptions. An optional /llms-full.txt carries expanded context for models that can ingest more.
How it differs from robots.txt and sitemap.xml
robots.txt controls access, sitemap.xml lists every indexable URL for crawlers, llms.txt curates and describes a subset for language models. They complement each other; none replaces another.
A minimal working example
Start with '# Site Name', then a blockquote of one or two sentences describing what the site does, then sections like '## Free Tools' and '## Guides' with Markdown links and a short description after each colon. Keep it under a few hundred lines and keep every link live.
What it realistically achieves
No major engine has committed to using it as a ranking or retrieval input. Its practical value today is discoverability of your best pages for agents that do fetch it, plus internal clarity about your content hierarchy. Treat it as low-cost, low-risk hygiene rather than a growth lever.
Maintenance rules
Regenerate it whenever you publish or retire significant pages, never list private or gated routes, and validate that every URL returns 200. A stale llms.txt full of 404s is worse than none.