defend
Check whether a page can be read, indexed and quoted by a machine. 15 rules, every one citing the specification it comes from, so you can disagree with any of them.
We check pages people ask us to check. Nothing is crawled, nothing is listed, and a report is not indexed unless its owner asks for it.
What it looks at
| Area | From |
|---|---|
| Open Graph metadata | The Open Graph protocol |
| Document title and language | HTML Standard |
| Alt text, heading order, link purpose | WCAG 2.2, Level A |
| Structured data | JSON-LD 1.1 and schema.org |
| Canonical link and robots directives | RFC 6596, RFC 9309 |
| Whether an answer is extractable at all | Our own notes, marked as such |
Findings drawn from our own notes are labelled and never counted against a page we do not own.
When a rule changes
Rules get added, and one gets removed when it turns out to cite a specification that does not say what it was read as saying. That list goes out when it happens, and at no other time.
Double opt in. The confirmation link is the only thing that adds you, every message carries a one-click removal link, and the address is never passed on. Stored: the address, the time you asked, and the time you confirmed.