A2A — The Agent-to-Agent Protocol
Google announced A2A in April 2025; by April 2026 the spec is at https://a2a-protocol.org/latest/specification/ and 150+ organizations back it. A2A is the horizontal complement to MCP (Lesson 13): where MCP is vertical (agent ↔ tools), A2A is peer-to-peer (agent ↔ agent). It defines Agent Cards (discovery), tasks with artifacts (text, structured data, video), opaque task lifecycles, and auth. Production systems increasingly pair MCP with A2A. Google Cloud rolled A2A support into Vertex AI Agent Builder during 2025-2026. Your agent needs to call another agent on another system. How? You can expose an HTTP endpoint, define a bespoke JSON schema, and hope the other side speaks it. Every pair of agents becomes a custom integration. A2A is the universal wire protocol for that call. Standard discovery, standard task model, standard transport, standard artifacts. Like HTTP+REST but for agents as first-class citizens. Agent Card. A JSON document at /.well-known/agent.json describing the agent: name, skills, endpoints, supported modalities, auth requirements. Discovery happens by reading the card. Task. The unit of work. An async, stateful object with a lifecycle: submitted → working → completed / failed / canceled. A client sends a task, polls or subscribes for updates. Artifact. The result type produced by a task. Text, structured JSON, image, video, audio. Artifacts are typed so different modalities are first-class. Opaque lifecycle. A2A does…
A2A — The Agent-to-Agent Protocol: Google announced A2A in April 2025; by April 2026 the spec is at https://a2a-protocol.org/latest/specification/ and 150+…
This free lesson is part of the AI Engineering from Scratch curriculum. Read the full explanation, run the lesson code, and verify the result in the interactive reader or from the repository source.
Browse the complete course catalog or open this lesson on GitHub.