An open protocol developed by Google to enable communication and interoperability between AI agents.

Agent2Agent (A2A) Protocol Project Details

Project Overview

The Agent2Agent (A2A) protocol is an open standard developed by Google, designed to address a key challenge in the AI field: enabling generative AI agents built on different frameworks, by different companies, and running on different servers to communicate and collaborate effectively - as agents, not just tools.

Core Features

Through the A2A protocol, agents can:

• Discover each other's capabilities - Agents can understand the functions that other agents possess.
• Negotiate interaction methods - Supports multiple interaction modes such as text, forms, and media.
• Securely collaborate on long-term tasks - Conduct continuous task collaboration while ensuring security.
• Keep internal states private - Achieve collaboration without exposing internal states, memory, or tool implementation details.

Project Goals

As AI agents become increasingly popular, their interoperability is crucial for building complex, multi-functional applications. The A2A protocol aims to:

Break Down Silos

Connect agents in different ecosystems to achieve cross-platform collaboration.

Enable Complex Collaboration

Allow specialized agents to work together to handle tasks that a single agent cannot complete independently.

Promote Open Standards

Advance agent communication standards in a community-driven manner, encouraging innovation and widespread adoption.

Protect Privacy and Security

Allow agents to collaborate without sharing internal memory, proprietary logic, or specific tool implementations, enhancing security and protecting intellectual property.

Technical Features

Standardized Communication

• Based on HTTP(S) JSON-RPC 2.0 protocol
• Unified communication standards ensure cross-platform compatibility

Agent Discovery

• "Agent Cards" provide detailed descriptions of agent capabilities and connection information
• Supports dynamic discovery and capability querying

Flexible Interaction

• Supports synchronous request/response mode
• Supports streaming (Server-Sent Events)
• Supports asynchronous push notifications

Rich Data Exchange

• Handles text, files, and structured JSON data
• Supports multimedia content transmission

Enterprise-Grade Features

• Designed for security, authentication, and observability
• Meets enterprise application requirements

Project Resources

Documentation and Specifications

• Full Documentation: Agent2Agent Protocol Documentation Site
• Protocol Specification: A2A Protocol Specification

SDKs and Tools

• Python SDK: A2A Python SDK

  pip install a2a-sdk
  

• JavaScript SDK: A2A JavaScript SDK
• Sample Code: A2A Samples

Technical Roadmap

Agent Discovery

• Formally include authorization schemes and optional credentials directly in AgentCard

Agent Collaboration

• Investigate the QuerySkill() method for dynamically checking unsupported or unexpected skills

Task Lifecycle and User Experience

• Support dynamic UX negotiation within tasks (e.g., agents adding audio/video mid-conversation)

Client Methods and Transport

• Explore extending support for client-initiated methods (beyond task management)
• Improve streaming reliability and push notification mechanisms

Demo Video

Watch the demo video to see how A2A enables seamless communication between different agent frameworks.

Get Started

1. Install the SDK:

   pip install a2a-sdk
   

2. Clone the Sample Code:

   git clone https://github.com/google-a2a/a2a-samples.git
   cd a2a-samples/samples/python/agents/helloworld
   uv run .
   

3. Validate Your Agent: Validate your agent using the agent inspector. Follow the instructions in the a2a-inspector repository.

The A2A protocol represents a significant advancement in AI agent interoperability, laying the foundation for building a more collaborative and powerful AI ecosystem.