Consequence firewall for machine actions. EMILIA Gate verifies exact authority before money, code, permissions, infrastructure, or regulated state changes; the open protocol makes the evidence independently verifiable.

A vendor-neutral protocol for recording why, by whom, under which policies, scopes, risks, and human-review conditions an AI agent action was authorized, restricted, deferred, or denied.