Advanced10 min
A2A Protocol
Google's Agent-to-Agent protocol for inter-agent communication. Integration with LangGraph Agent Server. When to use A2A vs direct handoffs.
Quick Reference
- →A2A is an open protocol by Google that standardizes how agents discover, authenticate, and communicate with each other across services
- →Each A2A agent publishes an Agent Card describing its capabilities, endpoint, and supported interaction modes
- →A2A uses Tasks as the communication primitive — a client agent creates a task, the remote agent processes it and returns results
- →LangGraph Agent Server can expose any graph as an A2A-compatible endpoint, enabling cross-framework agent interop
- →Use A2A for cross-organization or cross-service agent communication; use direct handoffs for agents within the same graph
What is A2A?
A2A = inter-service agent protocol
Agent-to-Agent (A2A) is an open protocol by Google that standardizes how agents discover, authenticate, and communicate with each other across services and frameworks.
A2A solves the interoperability problem. When agents are built with different frameworks (LangGraph, CrewAI, AutoGen) or deployed in different organizations, they need a standard way to discover capabilities and exchange tasks. A2A provides this via Agent Cards, Tasks, and streaming updates over HTTP.
- ▸Discovery: agents publish Agent Cards at /.well-known/agent.json describing their capabilities
- ▸Communication: client agents create Tasks on remote agents via HTTP POST
- ▸Streaming: long-running tasks return partial results via Server-Sent Events (SSE)
- ▸Authentication: standard HTTP auth (OAuth2, API keys) at the transport layer
- ▸Framework-agnostic: any agent that speaks HTTP can participate, regardless of internal implementation
A2A protocol: discover agent cards, send tasks, stream results over HTTP