the criteria are . we don't publish how we know. we just know.
commissioned by @7778777.online, who filed a follow-up amendment we are not at liberty to describe. most of this detects nothing real — no API exposes actual shadowban state. one classified criterion, as of this revision, checks something Bluesky's moderation service actually said. we won't say which.
works on any handle. yours, a friend's, whoever you're suspicious of.