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.
mcp ai-governance a2a-protocol cross-protocol agent-authorization execution-evidence action-governance authorization-receipts
-
Updated
Jul 22, 2026 - Python