Since 10:21 UTC a program, an AI agent for example, can start monitoring without a person in the panel. POST /api/v1/agent/register needs no key and returns an API key, a claim link, the expiry and the limits: 5 monitors, every 60 seconds at the fastest, 21,600 starter credits, 72 hours. The account can monitor only hosts on a domain it has verified, alert only to a webhook on such a domain or to its own status page, cannot buy credits or add email or SMS channels, and is deleted 72 hours after registration unless a person opens the claim link, types an email address and accepts the Terms. Then it is an ordinary account with that month's free credits, and nothing the agent set up is lost. Registration is limited per address and network. How it works, for agents and for the people they hand accounts to: For agents.
Domains in the API. GET and POST /api/v1/domains list and add domains for any key, and POST /api/v1/domains/<name>/verify checks the proof now and says what is missing. The verification record is now named _pingwise.<domain> with the value pingwise-verify=<token>, and the file /.well-known/pingwise-verify.txt; the older uptimebox names and existing tokens keep verifying.
The quorum, in writing. Every incident in the API and every webhook delivery now carries a quorum block: how many probes agreed out of how many checked, the continents and networks among them, and the rule in force at the time, for example «down = 8 of 12 probes, ≥ 3 continents, ≥ 3 providers». Incidents from before today have no block.
The Terms, the Acceptable Use Policy and the Privacy Policy describe accounts opened by an automated client since 10:15 UTC.