llms.txt Generator — Make Any Website AI-Readable

Crawl any website and generate standards-compliant llms.txt and llms-full.txt files following the llms.txt specification. Point it at a domain and get back AI-ready files plus a structured per-page dataset — no setup, no code.

What is llms.txt?

llms.txt is the emerging standard (proposed by Jeremy Howard, adopted by Anthropic, Mintlify, Zapier and hundreds of doc sites) for telling AI models what your website contains. Like robots.txt for crawlers or sitemap.xml for search engines, llms.txt gives LLMs a curated, token-efficient map of your content:

• llms.txt — a concise index: site title, summary, and a linked list of your pages with one-line descriptions.
• llms-full.txt — the full extracted content of every page in one clean file, ready to drop into a context window or a RAG pipeline.

If you want ChatGPT, Claude, or Perplexity to represent your product accurately, this is the file they're looking for.

What it does

1. Crawls your site breadth-first from the start URL(s), respecting your page limit and URL include/exclude patterns.
2. Extracts the main content of each page with trafilatura — navigation, cookie banners, and boilerplate are stripped.
3. Writes llms.txt and llms-full.txt to the run's key-value store (direct download URLs).
4. Pushes a per-page dataset with URL, title, description, extracted content, and word count.

Input

{
  "startUrls": [{ "url": "https://docs.yourproduct.com" }],
  "maxPages": 50,
  "globs": ["/docs/**", "/guides/**"],
  "excludes": ["/blog/**", "/changelog/**"],
  "outputFormat": "markdown"
}

Parameter	Type	Default	Description
startUrls	array	required	Where to start crawling
maxPages	integer	10	Maximum pages to crawl
globs	array	[] (all)	Only crawl URLs matching these glob patterns
excludes	array	[]	Skip URLs matching these patterns
outputFormat	string	markdown	markdown or plaintext content extraction
proxyConfiguration	object	Apify proxy	Proxy settings

Output

Key-value store (downloadable files):

# Your Product Docs

> Documentation for Your Product — deployment, API reference, and guides.

- [Quickstart](https://docs.yourproduct.com/quickstart): Install and run in 5 minutes
- [API Reference](https://docs.yourproduct.com/api): REST endpoints and auth
- …

Dataset — one record per crawled page:

{
  "url": "https://docs.yourproduct.com/quickstart",
  "title": "Quickstart",
  "description": "Install and run in 5 minutes",
  "content": "## Quickstart\n\nInstall the CLI…",
  "wordCount": 742
}

Use cases

• AI SEO / GEO (generative engine optimisation) — make sure AI assistants describe your product from your own content, not third-party guesses.
• Documentation portals — publish llms.txt alongside your docs so AI coding assistants answer from the current version.
• RAG ingestion — llms-full.txt is a clean, deduplicated corpus of a whole site in one file; skip the scraping-and-cleaning step entirely.
• Content audits — the per-page dataset (titles, descriptions, word counts) doubles as a crawl report.
• Agency deliverable — generate llms.txt files for client sites as a productised service.

Run it via API

curl "https://api.apify.com/v2/acts/darknezz~llms-txt-generator/run-sync?token=YOUR_TOKEN" \
  -X POST -H 'Content-Type: application/json' \
  -d '{"startUrls": [{"url": "https://example.com"}], "maxPages": 50}'

Then download llms.txt / llms-full.txt from the run's key-value store.

Pricing

Pay-per-event: $0.01 per page crawled and processed. A 50-page documentation site costs about $0.50 per full regeneration. Re-run on a schedule to keep the files current.

FAQ

Where do I put the generated file? Upload llms.txt to your site root (https://yoursite.com/llms.txt), exactly like robots.txt. Serve llms-full.txt next to it if you want to expose full content.

Does it work on JavaScript-heavy sites? It fetches server-rendered HTML. Sites that render all content client-side may yield thin extractions — most marketing sites and doc platforms work well.

How is this different from a generic website scraper? Output is formatted to the llms.txt spec out of the box — index file plus full-content file — not raw HTML or generic JSON you still have to assemble.

Can I limit it to one section of the site? Yes — use globs (e.g. /docs/**) and excludes to fence the crawl precisely.

How often should I regenerate? Whenever content changes; most users schedule weekly. With per-page pricing, regeneration stays cheap.