A2A Concepts

Understand the key A2A concepts used to discover, call, and manage Oracle Autonomous AI Database agents using the A2A server.

• Select AI Agent Teams
  A runnable in-database AI agent that perform an agentic workflow. An agent team coordinates responsibilities, shares context, and runs agent-task pairs to complete multi-step interactions reliably.

• A2A Protocol
  A standard REST-based protocol that enables AI agents and applications to discover, call, and coordinate with each other across systems.

• A2A Server (Remote Agent)
  An AI agent or agent system that exposes an HTTPS endpoint implementing the A2A protocol. It receives requests, processes tasks, and returns results or status updates. From the client perspective, the server operates as a black-box system. Your Autonomous AI Database instance can be enabled as an A2A server.

• A2A Client (Client Agent)
  An application, service, or AI agent that acts on behalf of a user. The client initiates communication to other agents using the A2A protocol.

• Agent Card
  Defines the agent’s capabilities, interaction model, and authentication requirements.

• Authentication and Authorization
  A2A uses standard web security practices. Authentication requirements are defined in the agent card. Autonomous AI Database uses OAuth for A2A server authentication and database credentials for user identity.

• Agent Discovery
  The process of identifying available agents and their capabilities using the A2A protocol.

• Artifact
  Represents outputs or results generated by an agent.

• Part
  A structured component of a message.

• Message
  Represents communication exchanged between client and agent during task execution.

• A2A Task
  Represents a unit of work requested by the client and processed by the agent.