llms.txt is a plain Markdown file served at /llms.txt that summarises a site for LLMs and AI agents.
llms.txt guide
What is llms.txt?
llms.txt is a Markdown file at the root of a website that gives LLMs and AI agents a curated, machine-readable map of the site. Think robots.txt, but for what to read instead of what to crawl.
Completely free for 30 days. No credit card required.
01Details
The short definition
It is part of the emerging llms.txt standard proposed in 2024 and adopted across docs sites, SDK repos, and AI-first products through 2025–2026.
Unlike robots.txt, llms.txt does not gate access — it curates what is worth reading and where to find the canonical Markdown version of each page.
02Details
What goes in an llms.txt file
A top-level H1 with the project or product name.
A short blockquote describing the project in one or two sentences.
Optional context paragraphs and a Markdown file index grouped by section, with links to the canonical .md version of each page.
An optional Optional section for nice-to-have material the agent can skip if context is tight.
03Details
Why llms.txt matters in 2026
LLMs are increasingly the first reader of your docs, not the user. A curated entry point reduces hallucinations and dead-end navigation.
Most HTML docs sites carry navigation chrome, ads, and dynamic content that wastes the model's context window. llms.txt points agents at clean Markdown.
Search engines have not (publicly) confirmed weighing llms.txt yet, but adoption among AI-first crawlers is real — expect that signal to grow.
For API teams specifically: an llms.txt that links every endpoint's Markdown reference lets coding agents wire up calls correctly on the first try.
04Details
How llms.txt differs from robots.txt and sitemap.xml
robots.txt is access control: which crawlers may fetch which URLs.
sitemap.xml is discovery: a flat list of every URL the site wants indexed.
llms.txt is curation: the small set of pages that matter most, with one-line context for each.
Ship all three. They solve different problems and good citizens of the AI-readable web carry every one.
05Outputs
How Bloom generates llms.txt automatically
Bloom projects can ship a generated llms.txt at the root of the published docs site (in preview alongside the SDK pipeline).
The file is built from the OpenAPI spec — endpoint summaries, schemas, examples, auth modes — so it stays in sync as the spec evolves.
Bloom also publishes per-page canonical .md files so the links in llms.txt resolve to real Markdown agents can consume directly.
The whole pipeline is part of the 30-day free trial; no credit card required to preview the generated file.