Observability and control layer for Agent-to-Agent protocol traffic

Route A2A traffic through AI Gateway with protocol-aware metrics, tracing, and agent card rewriting

Route A2A traffic through AI Gateway

The Agent-to-Agent (A2A) protocol defines how AI agents communicate with each other over HTTP using JSON-RPC and REST bindings. As agent-to-agent communication moves into production, teams need visibility into A2A traffic and control over how it flows.

AI Gateway acts as a control and observability layer for A2A traffic, enabling you to route agent-to-agent requests, extract task metadata, and rewrite agent card URLs so clients route through the Gateway. This gives you a single point of control over how agents discover and call each other, with structured metrics fed into the Konnect analytics pipeline. Configure A2A traffic using AI Agents and attach AI Policies for authentication, access control, and observability.

Overview of A2A traffic flowing through AI Gateway

Proxy and observe A2A traffic

Create an AI Agent entity to proxy A2A traffic, then attach an OpenTelemetry AI Policy to export traces and metrics for that traffic.

Secure A2A endpoints

Reference an AI Auth Strategy on your AI Agent entities to authenticate A2A clients, and attach the Request Size Limiting Policy to limit request size before traffic reaches your agents.

Create and monitor an AI Agent

Set up an AI Agent to proxy A2A traffic and observe it with OpenTelemetry.

Observe A2A traffic

AI Gateway records A2A protocol traffic data so you can analyze how agent-to-agent requests are processed and resolved.

  • Audit logs capture task IDs, JSON-RPC method calls, payloads, latencies, and errors.
  • OpenTelemetry spans record task state, context IDs, TTFB, SSE event counts, and response sizes.
  • Metrics track A2A-specific signals and performance indicators over time.

Manage A2A traffic

AI Gateway records A2A protocol traffic data so you can apply rate limits, authenticate users, and restrict request types.

Help us make these docs great!

Kong Developer docs are open source. If you find these useful and want to make them better, contribute today!