← All initiatives
New · August 20, 2026Watch / conditionCommercial protocol · Working draft

AI Agency Protocol

Purpose-bound, expiring authority for autonomous AI agents

WHY ENGAGE

Why this initiative matters

Separates an agent’s technical capabilities from the authority it may exercise now, using scoped grants, mediated execution, revocation, and attenuated delegation.

HIGHEST-VALUE CONTRIBUTION

Where help can compound

Threat-model the broker, build identity-provider and agent-framework adapters, test delegation invariants, and create independent conformance and abuse cases.

THE FIRST QUESTION TO ASK

Which parts are genuinely interoperable public standards, and what evidence would demonstrate that the broker fails safely?