Crossing Into AI
Public tool · Audit

How legible are you to machines?

Models are how people find you now. Before an answer engine can cite you, recommend you, or quote you, it has to read you — cleanly, on your own terms. This grades whether it can: five checks, scored out of 100, run in the open.

Grade a site's legibility

We normalize to the origin, follow your own redirect, and read only public pages.
Advanced options
The rubric

What gets checked?

Five signals a language model relies on to understand a site — each weighted by how much it matters. The number is the points it's worth toward your score of 100.

Legibility isn't about blocking or allowing crawlers — it's about being intelligible. The audit reads the same public files any model would, and reports what it found.

This tool’s own llms.txt
  1. 25pts

    llms.txt

    A curated map at /llms.txt telling models what you publish and where it lives.

  2. 20pts

    robots.txt

    Whether AI crawlers get an explicit, intentional stance — a clear signal, allow or disallow.

  3. 15pts

    Sitemap

    A parseable sitemap.xml so nothing worth reading goes undiscovered.

  4. 25pts

    Metadata

    Title, description, canonical link, and OpenGraph on your homepage — the basics, done right.

  5. 15pts

    Structured data

    JSON-LD schema that names what your pages actually are, in a form machines parse directly.