See the pricing

Free agent-readiness self-check

Ten checks, weighted by measured evidence rather than fashion. You get a grade out of 100 and the fixes in the order worth doing them. It runs entirely in your browser.

Last reviewed 2026-09-04 · AgentProof, Harmony Future Holdings Limited, Dublin

The same weights as the paid report

The scoring below uses the same weights and grade bands as our audit. The difference is that this one never looks at your site — it scores what you tell it. That makes it instant and private, and it makes it only as accurate as your answers. The guide explains every weight.

Ten checks. Answer yes only where it is true of every page a customer might land on — the audit takes the worst page, and so should you. Partly scores half.

Labelled form inputs (14 points)

Every form input has a label or aria-label.

Valid JSON-LD (15 points)

Valid schema.org JSON-LD is present and parseable.

Real buttons and links (12 points)

Interactive elements are real <button> and <a> elements, not bare clickable <div>s.

Organization or Product schema (10 points)

The JSON-LD declares an Organization, Product, Service or LocalBusiness with real fields.

An explicit AI-crawler policy (8 points)

robots.txt takes a deliberate position on GPTBot, ClaudeBot and Google-Extended rather than staying silent.

Title and meta description (8 points)

A real <title> and meta description exist.

A machine-readable contact route (8 points)

A mailto:, tel: or schema contactPoint exists.

Exactly one h1 (6 points)

One <h1> states what the page is.

An llms.txt file (6 points)

An /llms.txt file exists and describes the site for language models.

Declared language (4 points)

<html lang> is declared.

WebMCP tools (not scored)

The page registers WebMCP tools. Absence is normal today and is never counted against you — this is recorded, not scored.

Nothing leaves your browser. This check never looks at your site — it scores what you tell it, on your device.

Answer for your worst page, not your best

A real audit takes each check’s worst result across the pages it sees. Most people answering this form think about their homepage, which is usually the most cared-for page on the site. The page that decides whether an agent can buy from you is the checkout, and it is usually worse.

Why the weights are not evenly spread

Because the evidence is not evenly spread. Labelled inputs carry 14 points because there is a measured task-completion gap behind them (roughly 78% versus 42%). llms.txt carries 6 because Google says it does nothing for Search and most files see no bot traffic at all.

We weighted down the thing that is easiest to sell

llms.txt is the artefact vendors lead with, because it is a text file anyone can produce in five minutes. We scored it 6 out of 91 after a market scan in August 2026, which was commercially inconvenient and factually right. If someone is selling you agent-readiness and leading with llms.txt, ask them what it scores.

Why WebMCP is not scored

It is a W3C Web Machine Learning Community Group draft, in Chrome origin trial through roughly Q1 2027. Absence is normal today, so counting it against a site would be manufacturing a failure. It is recorded as information and excluded from the denominator entirely — exactly as our engine excludes it.

Privacy

No analytics, no cookies, no network requests once the page has loaded, and no fetch of your site. Nothing you answer reaches us.

Next

  • The full guide — all ten signals, their weights, and the evidence behind each.
  • FAQ — do I need llms.txt? Should I block GPTBot?

The audit on your real pages, and the fix pack

The self-check below scores what you can see by eye. The audit runs the same checks against your actual pages, and the fix pack gives you paste-ready markup plus a retest.

See the pricing — from €49