Skip to content

Free llms.txt Generator

Generate a ready-to-review llms.txt for a whole website or one subpath, free with no signup.

Free, no account needed — 3 runs a day. Sign in for 10. Sign in with Google

The generator crawls a site’s public sitemap and page metadata, then AI selects and groups pages. Every URL comes from the crawl; review the draft before deploying.

llmstxt v2

What is llms.txt?

llms.txt is a plain-Markdown file that gives AI systems a curated reading map: an H1 with the site’s name, a blockquote summary, and sections of links to the pages that matter most. Proposed by Answer.AI in September 2024 and specified at llmstxt.org, the format’s whole premise is selection — a small file an AI can read in one pass, not a second sitemap. The spec was revised to v2 on 10 August 2026; this generator writes to v2.

Publishing adoption is real, but no major AI vendor documents consuming third-party llms.txt. Google Search states that it ignores these files. Treat llms.txt as low-cost infrastructure, not a ranking lever. The llms.txt entry tracks the evidence, and the deployment playbook covers serving and verification.

How it works

  1. Fetches the homepage for the site name, meta description, and language.
  2. Discovers pages through robots.txt and the sitemap; when no sitemap exists, it falls back to homepage links.
  3. Fetches the top pages for their real titles and descriptions, dropping any that no longer resolve.
  4. An AI model drafts the file from that crawl data: it selects the pages that answer questions about the site, names the sections, and writes the summary and descriptions in the site’s own language. If the AI is unavailable, a deterministic assembly ships instead.

What the output looks like

A draft for a fictional analytics product — your result follows the same shape, in your site’s own language:

# Northwind Analytics

> Northwind Analytics is a self-hosted product-analytics server: event ingestion over HTTP, storage in ClickHouse, SQL and a query API. Self-hosted only; there is no managed cloud offering.

## Product

- [What Northwind is](https://example.com/product): scope, and what it deliberately does not do
- [Pricing](https://example.com/pricing): per-seat and per-event tiers

## Docs

- [Quickstart](https://example.com/docs/quickstart): Docker Compose to first event in ten minutes
- [Query API](https://example.com/docs/api): endpoints, auth, rate limits

## Optional

- [Changelog](https://example.com/changelog): release history, safe to skip

What it deliberately does not do

Deploying the file

Frequently asked questions

What is llms.txt?
A plain-Markdown file that gives AI systems a curated map: what a site is, and which pages matter most. It sits at the root of whatever it covers — the whole site, or one subtree. The llms.txt entry covers the standard’s background and current status.
Is llms.txt the same as robots.txt?
No. robots.txt is access policy — which crawlers may fetch which paths. llms.txt is a reading map — what an AI should read first. They complement each other: see robots.txt and AI crawlers.
Does Google read llms.txt?
Google’s documentation states its AI features do not use llms.txt. Other consumers are less explicit; publishing stays cheap, which is the honest case for doing it — the evidence is tracked in the llms.txt entry.
Do I need an account?
No — three runs a day are free with no sign-in at all. Because each run crawls a site and calls an AI model, usage is metered; signing in with Google raises the limit to ten a day and stores only your email, name, and avatar (see the privacy policy).
Where do I put the generated file?
At the root of the path it covers: a whole-site map goes to /llms.txt, a map generated for /docs goes to /docs/llms.txt. Where more than one file applies, agents use the most specific. The llms.txt deployment playbook covers serving, verification, and when to regenerate.
Why is my result missing pages?
The crawler considers at most 96 pages from the first few sitemap files, and the AI then curates down to what best represents the site — an llms.txt is meant to be a small, selective map, not an index. Pages that 404 during the crawl are dropped.
Can I generate one for just a subfolder?
Yes — enter the path (example.com/docs) and the map is scoped to the pages beneath it. v2 made this explicit: a file covers the pages under its own path, and where several apply agents use the most specific one. It is the route for anyone who controls a subdirectory rather than a whole domain.
Is the generated file spec-compliant?
Yes on structure, against v2: H1 site name (the only element the spec requires), blockquote summary, and H2 link sections, per the llmstxt.org spec. Overflow goes under Optional, which v2 treats as a naming convention for skippable links rather than an instruction consumers obey. What no generator can do is serve the file, add the discovery link relations, or publish the markdown page twins from your server — deployment remains yours.

Keep reading