An open protocol enabling communication and interoperability between opaque agentic applications.
Agent Card
Facilitates the discovery of agent capabilities and requirements through a standardized metadata file.
A2A Server
Enables agents to expose HTTP endpoints for managing tasks and requests.
Streaming Support
Allows real-time updates for long-running tasks through Server-Sent Events.
Push Notifications
Enables proactive task updates sent directly to a client-provided webhook URL.
The Agent2Agent (A2A) protocol is designed to enhance collaboration among independent AI agents by providing a common language for communication, regardless of their underlying frameworks or vendors. This initiative by Google addresses the challenge of integrating various agents, allowing them to showcase capabilities and negotiate interactions with users effectively. The protocol supports different modes of communication, ensuring that agents can work together securely and efficiently.
Includes core components such as Agents, Tasks, and Messages, with defined states for task lifecycle management. The protocol's JSON specification outlines the structures and methods for task execution and interaction.
Integration of AI agents from different vendors in enterprise applications.
Facilitating communication between AI agents in multi-agent systems.
Supporting the development of collaborative AI applications across varied frameworks.
It is an open protocol that allows different AI agents to communicate and collaborate effectively.
By providing a common framework for communication, agents can work together regardless of their underlying technology.
Yes, contributions are welcome! You can refer to the contributing guide on the GitHub repository.