llms.txt and llms-full.txt are two files from the same proposal that serve different needs. llms.txt is a concise Markdown map — site name, summary, and sections of links with short descriptions — meant as a lightweight index. llms-full.txt follows the same structure but inlines fuller descriptions and, for content sites, much more of the actual page text, so a model can ingest substantial context in a single request instead of following every link. Publish llms.txt for almost any site; add llms-full.txt when your content is text-heavy (docs, guides, references) and you want models to absorb it without extra fetches. Neither file is a proven ranking lever — both are optional content hygiene.
What does each file contain?
Both files come from the llmstxt.org proposal and share the same Markdown skeleton: an H1 site name, an optional summary, and H2 sections containing lists of links. The difference is depth. llms.txt is the index — short, scannable, and primarily a set of pointers with one-line context per link.
llms-full.txt is the expanded edition. It keeps the same organization but inlines significantly more content: fuller descriptions, and for documentation or article sites, the substantive body text of the listed pages. The goal is that a model can read llms-full.txt once and have meaningful context without making a request for every individual page.
- llms.txt: concise index — name, summary, sections of described links.
- llms-full.txt: same structure, but with fuller descriptions and inlined page content.
- Both are Markdown and both live at the site root (/llms.txt, /llms-full.txt).
When should you publish llms-full.txt?
The larger file earns its keep when your value is in the text itself and there is a lot of it. Documentation sites, API references, knowledge bases and long-form guide libraries are the natural fit: a model reading the full text in one pass can answer detailed questions without round-tripping to dozens of pages.
For a small marketing site or a portfolio, llms-full.txt has limited upside — there simply isn't enough content to make inlining worthwhile, and the concise llms.txt already conveys what the site is. In that case, publishing only llms.txt is a reasonable, lower-maintenance choice.
Watch the size. Because llms-full.txt can inline a lot of text, it can grow large; keep it focused on genuinely high-value pages rather than dumping the entire site, both to stay maintainable and to respect the limited context a model will actually use.
How do you generate both files?
The generation steps are identical: discover your key URLs, read each page, group them into logical sections, and render Markdown. The only difference is how much per-page content you include — pointers for llms.txt, fuller text for llms-full.txt.
Our generator produces both at once from a single URL: it discovers up to 20 pages, groups them semantically, and outputs the concise map and the fuller edition together, with deployment instructions. You then review and trim before publishing — generation is a strong first draft, not a final answer.
Frequently asked questions
What is the difference between llms.txt and llms-full.txt?+
llms.txt is a concise Markdown index: site name, summary, and sections of links with short descriptions. llms-full.txt uses the same structure but inlines fuller descriptions and, for content sites, much more page text, so a model can ingest context in a single fetch.
Do I need to publish both files?+
No. llms.txt alone is fine for most sites. Add llms-full.txt when your content is text-heavy — documentation, references, guide libraries — and you want models to absorb the substance without fetching every page individually.
Can llms-full.txt get too big?+
Yes. Because it inlines page text it can grow large, so keep it focused on genuinely high-value pages rather than the whole site. That keeps it maintainable and respects the limited context a model will actually use.