A2A

Expose Agno agents via Google’s Agent-to-Agent protocol for inter-agent communication.

A2A (Agent-to-Agent Protocol) is Google’s open standard for agent interoperability. Expose your Agno agents so other A2A-compatible systems can discover and communicate with them.

a2a_agent.py

from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from agno.os import AgentOS

agent = Agent(
    name="Research Assistant",
    model=OpenAIResponses(id="gpt-5.4"),
    instructions="You help with research tasks.",
)

agent_os = AgentOS(
    agents=[agent],
    a2a_interface=True,
)
agent_os.serve(port=9001)

Install the dependency: uv pip install 'agno[a2a]'

The Protocol Stack

A2A is one layer of a three-protocol stack:

Protocol	Purpose
MCP	Agents ↔ Tools & Data
A2A	Agents ↔ Agents
AG-UI	Agents ↔ Users

Quick Start

1. Run your agent

   python a2a_agent.py
   

   Your agent exposes A2A endpoints at http://localhost:9001/a2a/agents/{agent_id}/.

2. Discover the agent

   Other A2A clients can fetch your agent’s card:

   curl http://localhost:9001/a2a/agents/research-assistant/.well-known/agent-card.json
   
3. Send a message

   curl -X POST http://localhost:9001/a2a/agents/research-assistant/v1/message:send \
     -H "Content-Type: application/json" \
     -d '{"message": {"role": "user", "parts": [{"text": "Hello!"}]}}'
   

Endpoints

Each agent, team, and workflow exposes three A2A endpoints:

Endpoint	Purpose
/.well-known/agent-card.json	Agent discovery card
/v1/message:stream	Streaming responses
/v1/message:send	Non-streaming responses

Developer Resources

• A2A Interface
• A2A Client
• A2A Protocol Docs