A2A
Connects one agent to another and moves messages between agents.
Read Protocol Specs →Protocol
A2A moves messages. A2AC coordinates the mission.
A2AC does not replace Google's A2A protocol. A2A defines how agents communicate; A2AC defines how multi-agent work is coordinated, governed, preserved, and verified.
Disambiguation
A2A vs A2AC
A2A: Agent2Agent Protocol. A2A is the open protocol originally introduced by Google and governed by the Linux Foundation for communication between independent AI agents. It defines how agents publish capabilities, discover one another, exchange messages, delegate tasks, report status, and return results.
A2AC: Agent-to-Agent Coordination. A2AC is the execution and coordination platform built around multi-agent work. It provides task creation, shared context across models, CUBE Protocol state continuity, authenticated browser collaboration through Chimera, tool access, policy controls, approvals, and cryptographically verifiable receipts.
Protocol boundaries
Each protocol owns a distinct part of the workflow.
MCP
Connects agents to tools, data, and controlled context.
Developer Tools →CUBE Protocol
Preserves task state, permissions, artifacts, memory, receipts, and execution history.
CUBE Protocol Specs →Q-Protocol
Manages execution sessions through StartQ, AutoQ, and EndQ.
Lifecycle Protocol →A2AC
Coordinates the entire system, enforcing tenancy, approvals, policy, continuity, and auditability.
A2AC Platform →Chimera
Provides authenticated browser collaboration where multiple capable models share one live workspace.
Chimera Browser →CUBE Protocol
Persistent state for coordinated work.
Identity Cubes
Machine-readable identity for agents, organizations, and product objects.
Receipt Cubes
Signed, hash-linked records for assignments, handoffs, approvals, and completed actions.
Capability Cubes
Governed definitions of what an agent may do with a model, API, tool, or enterprise system.