legible

Can AI agents understand your website?

Legible scans your site, shows what AI agents can and cannot understand, and generates the files and fixes to make it Agent-Ready.

If agents can't parse your site, they may skip it, misread it — or recommend someone else.

View sample report (PDF)

20 named checks · published scoring model · every failure maps to a fix

How agents see your site

When someone asks ChatGPT, Claude, or Perplexity to compare businesses like yours, the assistant reads your site — not your design. The page on the left is what you shipped. The structure on the right is everything the assistant gets.

What humans see

What agents read

Page metadata

title"Home"[generic]
desc[no description]

Landmarks

main[none found]
nav[none found]

Headings

h1[none — the name is in an image]
h3"Our Rooms"[skipped h2]

Interactive elements

img[no alt — hero photo invisible]
input[no label — check-in date]
input[no label — check-out date]
button[icon only — the booking button]
link"click here"[weak label]
link"Dining"
link"Spa & Wellness"
Example scan — a boutique hotel homepageLegibility Score 54 — Needs Work

Scan. Score. Fix.

Find what agents miss. Generate the files and fixes. Ship an Agent-Ready site.

Scan

Paste a URL. Legible reads the page the way an agent does — structure first, pixels never.

Score

Twenty checks across interaction clarity, structure, and meaning roll up into one Legibility Score.

Fix

Every failing check comes with a fix — and the artifacts are generated for you, ready to ship.

20 checks. Five grades. No mystery.

Interaction clarity, structure, and meaning — every check is named, every failure comes with the fix.

  • Ghost Buttons— critical severity
  • Icon-Only Actions— critical severity
  • Duplicate Labels— high severity
  • Orphaned Inputs— critical severity
  • Div Links— high severity
  • Ambiguous Links— high severity
  • Identical Links, Different Targets— high severity
  • State Blindness— critical severity
  • Dynamic IDs— medium severity
  • Missing Landmarks— high severity
  • Heading Chaos— medium severity
  • Semantic Soup— medium severity
  • Table Misuse— medium severity
  • Invisible Content— medium severity
  • Structured Data Depth— high severity
  • Unlabeled Images— medium severity
  • Title & Description Quality— medium severity
  • JS-Gated Content— critical severity
  • Missing agents.md— medium severity
  • Missing llms.txt— medium severity
Critical & HighMedium

Don't just score it. Ship the fix.

Other scanners hand you a list of problems. Legible generates the finished artifacts — files you can put on your site today.

Free scan

See where you stand

  • Legibility Score and grade
  • All 20 checks, pass or fail
  • Agent Vision: human view vs. agent view
  • Top fixes, named and prioritized
  • Embeddable score badge

Agent-Ready Fix Pack

Get it fixed

  • Every failing element, located
  • Generated agents.md, ready to ship
  • JSON-LD entity schema for your content
  • Agent policy card + MCP server scaffold
  • Step-by-step fixes for your platform — Webflow, Shopify, WordPress, Squarespace
  • Plain-language fix plan, client-ready

Founder batch open: 10 Fix Packs at $149, delivered within 48 hours — reserve from your scan results. Every free scan already includes the generated artifacts.

Built for sites that need to be understood

  • Agencies shipping client sites
  • Shopify & Webflow businesses
  • SaaS landing pages
  • Consultants & service providers
  • Local businesses compared in AI search

About Legible

“Websites had to be responsive, then accessible, then SEO-friendly. Now they have to be legible to agents.”

Legible shows whether AI agents can understand your website — then generates the fixes. It reads the page the way an agent does — the structure, the labels, the meaning — and tells you, in plain language, what agents can and cannot understand. AI discovery is becoming more structured: sites that are easier for agents to parse are easier to summarize, compare, and recommend.