Agent2Agent protocol (A2A)

The Agent2Agent (A2A) protocol is a communication standard that lets AI agents work together smoothly, even if they come from different vendors or use different models. Created by Google in 2025, it sets clear rules for how agents can share tasks, update each other, and coordinate without needing custom code for every interaction.

A key part of A2A is the agent card. This is a structured summary that shows what an agent can do, what it is working on, and what it needs. This card helps AI agents discover each other and decide who should handle which task.

The A2A protocol works alongside the Model Context Protocol (MCP). While A2A handles collaboration between agents, MCP manages connections to external tools and data. Together, they help multi-agent AI systems coordinate, execute tasks, and scale efficiently.