What is the A2A (Agent2Agent) Protocol?
The Agent2Agent (A2A) Protocol is an open standard for letting independent AI agents — built by different teams, vendors or frameworks — discover each other and delegate work, without a bespoke integration for every pair. It is the agent-to-agent counterpart to MCP, which connects a single agent to tools and data.
Where it came from
Google announced A2A in April 2025 and handed it to the Linux Foundation in June 2025, which now governs it as a vendor-neutral project. By 2026 it had 150+ backing organisations, with support in Microsoft Azure AI Foundry, Amazon Bedrock AgentCore and Salesforce Agentforce.
How it works
| Piece | Purpose |
|---|---|
| Agent Card | A JSON document at /.well-known/agent-card.json describing the agent's identity, skills, interfaces and security requirements |
| Transport | JSON-RPC 2.0 over HTTPS; Server-Sent Events for streaming; push notifications for long-running tasks |
| Task model | One agent sends another a task, tracks its state, and receives messages and artifacts back |
An agent publishes what it can do and how to reach it — not its prompts, its tools or its internal code. That boundary is the point: agents cooperate without revealing their implementation.
A2A vs MCP
They are complementary. MCP gives an agent a uniform way to use tools; A2A gives agents a uniform way to hand work to each other. A system can use both at once.