Coverage

Is a domain observed?

Enter a domain to see whether this archive observes it, from which sealed day onward, and which machine-readable files it looks at. The answer is yes or no plus dates. It says nothing about what those files contain.

How to read the answer

• First sealed observation is the earliest day on which an observation of this domain was included in a published daily root. Observations made before a day is sealed are not counted here, because a sealed observation is the only kind this archive can prove.
• Files observed lists the kinds of resource we request for that domain. It does not say whether any of them exist, what they returned, or what they say. A domain with no robots.txt at all still shows robots.txt here: we look, and the looking is what is recorded.
• Not observed means exactly that, and nothing more. It is not a statement about the domain, its operator, or its files. Most of the web is not observed by this archive.

What this check is not

It is not a search engine and not a directory. There is no way to list the observed domains through it: each lookup is a single request for one domain you already know, and the domain is hashed in your browser before anything is sent, so the name you typed never leaves this page. That is a deliberate property, not a side effect — a neutral witness should not accumulate a record of who asked about whom.

It also does not publish content. What a file said on a given day is part of an evidence extract, which is issued on request under the conditions set out in the terms, not served from a public page. The reasons are set out in what a record contains.

If a domain is not observed

You can suggest it. The form appears with a negative answer above, and it is also available below. Suggesting a domain is a request, not an instruction: what is observed follows documented criteria (a connection to the European Union, publicly served machine-readable files), and the decision stays with the operator of this archive.

A suggestion creates no obligation to observe the domain, no timeline, and no assurance that it will be added. That is not a formality. An archive that promised to observe on request would owe availability it cannot guarantee, and evidence that someone relied on would be missing on the day a crawl failed. The honest position is the one stated here: the suggestion is recorded, it is reviewed, and it is answered.

Suggest a domain

Three fields, all required. The reason matters most: it is what a person reads when deciding, and it is the part a machine cannot fake cheaply. Your address is the reply channel and is used for nothing else — see the privacy notice.

Unlike the check above, a suggestion sends the domain name in the clear. It has to: someone has to read it. If you would rather not have a domain name leave your machine, do not suggest it.

For machines

Both operations on this page are available to software without going through it. An assistant that speaks the Model Context Protocol can connect to https://machinewitness.eu/mcp and use two tools, check_coverage and suggest_domain; no key and no registration. Anything else can call GET /coverage/{sha256(domain)} and POST /request directly. The details are in llms.txt.

This is deliberate rather than decorative. An archive of what the web tells machines that could only be questioned by a human filling in a form would be contradicting its own subject.