The Agent2Agent (A2A) protocol is an open standard for communication between AI agents. It lets an agent built with one framework or vendor delegate work to an agent built with another.
The Problem It Solves
Organisations increasingly run many agents: a travel agent, an expense agent, a support agent. Without a common protocol, connecting them needs custom integration for every pair.
Core Ideas
- Agent Card: a machine-readable description of an agent — its skills, endpoint and authentication requirements — used for discovery.
- Tasks: units of work with a lifecycle, from submitted through working to completed, failed or cancelled.
- Messages: exchanges between client and remote agent, made of parts such as text, files or structured data.
- Artifacts: the outputs a task produces.
Built on Web Standards
A2A uses HTTP, JSON-RPC and server-sent events for streaming, so it fits existing infrastructure and security practice.
Opaque Agents
Agents collaborate without exposing their internal reasoning, tools or memory — only what they choose to share through the protocol.
Status
A2A was introduced by Google in 2025 and moved to open governance under the Linux Foundation. The specification continues to evolve, so check the current version for details.