Mumega
Start

A2A Protocol (Agent-to-Agent)

A2A protocol (Agent-to-Agent): how agents discover and call each other across orgs, and why 150+ orgs and all three clouds adopted it. June 2026.

The A2A protocol (Agent-to-Agent) is an open standard for agents to discover, authenticate, and call each other — including across organizational boundaries. Where the Model Context Protocol connects an agent to its tools, A2A connects an agent to other agents, so one agent can delegate a task to another it has never met.

Each A2A-speaking agent publishes an Agent Card at /.well-known/agent.json describing what it can do and how to reach it — the agent-network equivalent of a service discovery record.

Why A2A matters in 2026

A2A reached v1.0 stable on 2026-04-09 and has been a Linux Foundation project since June 2025. By mid-2026 it had 150+ participating organizations (AWS, Microsoft, Cisco, IBM, Salesforce, SAP, ServiceNow), 22k+ GitHub stars, five production SDKs, and native support inside both AWS Bedrock AgentCore and Azure Foundry.

This is the part that makes A2A strategically interesting: it is neutral ground. All three major clouds run it, so an agent on one cloud can call an agent on another. The protocol layer is converging even as the platforms compete.

A2A has a sibling for money: agentic payments standards like AP2 (the Agent Payments Protocol, 60+ orgs) let agents transact, not just talk.

How Mumega relates to A2A

Mumega’s internal bus already does agent-to-agent messaging with scoped tokens. A2A is the external, cross-organization version of the same idea — the wire over which a Mumega tenant’s agents could one day call a partner tenant’s agents without either side trusting the other’s infrastructure.

Sources

Last updated: Jun 5, 2026