What is an A2A agent?
What is it?
An A2A agent is an AI agent reachable through the A2A (Agent2Agent) protocol, an open standard run by the Linux Foundation. Where MCP connects an assistant to tools, A2A connects agents to each other: one agent calls another, hands it a task and follows its progress. Each agent publishes an agent card (/.well-known/agent-card.json) describing its skills, address and authentication.
When to use it
- Hand part of a job to a specialised agent run by another team or company.
- Find an agent that can do a specific task and learn how to call it.
- Make your own A2A agent and its skills known.
Page fields
| Agent card | Public address of agent-card.json, required. |
|---|---|
| Service address | A2A endpoint, if different from the card. |
| Version and bindings | Protocol version (for example 1.0) and supported bindings: JSON-RPC, gRPC or HTTP+JSON. |
| Authentication | None, API key, HTTP, OAuth 2.0, OpenID Connect or mutual TLS. |
| Capabilities | Streaming responses and push notifications (webhook). |
| Skills | What the agent can do, one line per skill. |
How to use it
“Use” tab: download the entry (.a2a.json) or read the agent card directly, then call the agent with an A2A client (official SDKs in Python, JavaScript, Java, Go and .NET).