← ALL EXPERIMENTS

LAB · TOOL · V1.1 · 12 CHECKS · AUGUST 2026

MACHINE READABILITY CHECK

Search engines index you. AI engines have to read you first — and most of them never run a line of your JavaScript. This checks twelve things a machine needs from a page, answers each one yes or no, and tells you how to verify every answer yourself. It is a snapshot of one URL, not an audit, and it stores nothing.

One page, plus its robots.txt and llms.txt. Nothing is stored, no email, no account.

WHAT IT CHECKS

Twelve binary checks · v1.1 · 12 checks · August 2026

Reachability 4

Can a machine get to the page at all?

  • HTTPS, at most two redirects
  • Answers a non-browser client
  • robots.txt lets the search crawlers through — OAI-SearchBot, Claude-SearchBot, PerplexityBot, Googlebot
  • Sitemap declared in robots.txt

Substance 3

Is there anything to read without running JavaScript?

  • At least 500 characters of text in the raw HTML
  • Exactly one H1
  • No skipped heading level

Identity 3

Does the page say what it is and who publishes it?

  • Title 15–65 and meta description 50–160 characters
  • Canonical link pointing at this page
  • Valid JSON-LD naming an organisation or person

Quotability 2

Can a machine hand the page on?

  • Open Graph title, description and image
  • llms.txt reachable

Training agents — GPTBot, ClaudeBot, Google-Extended — are read from the same file and reported with the result, but they count neither for nor against the page: opting out of AI training is a decision, not a defect, and Google-Extended governs Gemini rather than Google Search. And a robots.txt permission shows one thing only — whether that crawler is explicitly allowed or blocked in that file. It does not prove crawling, indexing, retrieval, citation or recommendation; each of those stays the respective platform’s decision.

WHAT IT DOES NOT CHECK

Because a check that pretends to be an audit is worth neither.

  • What any of it costs you in visitors, rankings or citations
  • Which fix comes first, and which finding is safe to ignore
  • Conversion friction, offer hierarchy, the copy itself
  • Accessibility, security headers, Core Web Vitals on real hardware
  • Every page beyond the one URL you enter

This tool measures one dimension: whether a machine can read the page. A Website Audit measures five, on real hardware, read by a human, and ranked by what to fix first.

Ran it and want to know what the red lines are costing you — or want the same standard built into your own site?

START A PROJECT → SEE THE AUDIT

Automated snapshot of a single URL, taken with ordinary GET requests — the submitted page plus the two standard machine-readable files at that origin, robots.txt and llms.txt, the same requests any browser or crawler makes. No advice, no security assessment, no guarantee of completeness or accuracy. Results reflect the page at the moment of the check and the rule set named above. The full terms for this tool are in clause 8 of our Terms of Use; what happens to the address you enter is described in our Privacy Policy.