Agency Check is a human decision interface built on the same evidence discipline as AI SAFE2. It is not a software merge gate.
The service planned for agency-check.app should translate a bounded evidence
set into one Decision Card:
ready_to_review;clarify_one_thing;pause_and_verify.Every card should preserve:
It must not call an action safe, execute the action, or convert an absence of evidence into approval.
Public reference: https://cyberstrategyinstitute.com/agency-check/