The Linux Foundation has launched the Agent2Agent (A2A) project, an open protocol created by Google for secure agent-to-agent communication and collaboration.

The protocol was developed by Google to provide a way to enable communication between opaque agentic applications and let them interoperate. The protocol can be used by developers to build agents that seamlessly interoperate, regardless of platform, vendor or framework. The A2A protocol is a collaborative effort that was launched by Google in April and now has support from around 100 other companies.

The problem A2A aims to solve is that gen AI agents, built on diverse frameworks by different companies running on separate servers, need a way to communicate and collaborate as agents. A2A aims to provide a common language for agents, so they can discover each other's capabilities, then negotiate interaction processes. It also aims to let agents collaborate securely on long running tasks, and be able to operate without exposing their internal state, memory, or tools.

The A2A protocol is a collaborative effort that was launched by Google in April and now has support from around 100 other companies. Under the Linux Foundation's governance, A2A will remain vendor neutral. The announcement of the project by the Linux Foundation was made with Amazon Web Services, Cisco, Google, Microsoft, Salesforce, SAP, and ServiceNow. The project will be hosted by the Linux Foundation and will be seeded with Google's transfer of the Agent2Agent (A2A) protocol specification, accompanying SDKs, and developer tools.

The protocol is designed to use standardized communication via JSON-RPC 2.0 over HTTP(S). Agent discovery is made using "Agent Cards" that provide details of the agent's capabilities and connection info. Interaction support includes synchronous request/response, streaming (SSE), and asynchronous push notifications. Data exchange can be made via text, files, and structured JSON data.

The announcement says that future plans start with enhancements to agent discovery with the formalizing of inclusion of authorization schemes and optional credentials directly within the AgentCard.

Agent collaboration improvements include developing a QuerySkill() method for dynamically checking unsupported or unanticipated skills, while task lifecycle improvements include support for dynamic UX negotiation within a task, for example an agent adding audio/video mid-conversation. 
There are also plans to extend support to client-initiated methods, and to improve streaming reliability and push notification mechanisms.

The Agent2Agent Protocol is on GitHub now. 

More Information

Linux Foundation Website

Agent2Agent On GitHub

Related Articles

Linux Foundation Launches Chromium Browser Support

Amazon Donates OpenSearch To Linux Foundation

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info