← All Articles
A2A Feature Request Architecture March 6, 2026

A2A Pub/Sub Proposal: Event-Driven Agent Coordination

As multi-agent systems scale beyond simple request-response patterns, native publish/subscribe semantics become essential infrastructure.

About the Author

zhengleibin filed Issue #1593 proposing built-in pub/sub support for the A2A protocol. The proposal targets a fundamental architectural gap in how agents coordinate at scale.

Why This Matters

The Agent-to-Agent Protocol currently uses a request-response model: Agent A sends a message to Agent B and waits for a reply. This works well for direct delegation but becomes inefficient when multiple agents need to react to the same event.

Consider a shopping assistant scenario:

Without pub/sub, the price-monitoring agent must individually notify each interested party, manage delivery confirmations, and handle failures. With pub/sub, it publishes once to a "price-alerts" topic, and subscribers receive the event automatically.

What the Proposal Suggests

The feature request outlines several key capabilities:

Design tension: A2A is intentionally minimal — adding pub/sub primitives significantly expands the protocol surface. The community will need to weigh simplicity against capability.

Technical Considerations

Implementing pub/sub in a decentralized agent ecosystem raises interesting challenges:

Broker architecture: Who runs the message broker? Options include:

Discovery: How do agents find topics of interest? This connects to the ongoing Agent Card discussions — perhaps topic catalogs become part of agent capabilities.

Security: Who can publish to a topic? Who can subscribe? The identity and authorization discussions (see AIAR Identity Extension) become even more critical with broadcast messaging.

Real-World Applications

Event-driven patterns unlock several use cases that are awkward with request-response:

What Happens Next

The proposal is in early discussion. Given the scope of the change, expect:

  1. Community feedback on whether pub/sub belongs in the core spec vs. an extension
  2. Alternative proposals (some might suggest integrating with existing message brokers rather than building native support)
  3. Prototyping in reference implementations to validate the design

For implementers building multi-agent systems today, this is a signal of where the protocol is heading. Even if native pub/sub takes months to land, designing with event-driven patterns in mind will ease future migration.

Follow the discussion: github.com/a2aproject/A2A/issues/1593