Agent-to-Agent Protocol
Google's open standard for agent collaboration—enabling autonomous agents to discover, authenticate, and work together across organizational boundaries.
While MCP addresses agent-to-tool communication, A2A solves the horizontal coordination challenge: how do agents from different vendors, frameworks, or organizations collaborate? The protocol provides a standardized layer that makes multi-agent collaboration as straightforward as single-agent tool use.
Design Principles
Agent Cards
Central to A2A is the Agent Card—a JSON document that serves as a digital identity badge. Agent Cards describe an agent's identity, capabilities, endpoints, and authentication requirements.
Task Lifecycle
A2A defines a managed task lifecycle ensuring reliability for long-running and asynchronous operations:
Communication Modes
A2A Extensions
A2A Extensions add opt-in, domain-specific capabilities on top of the core protocol, letting agents advertise, negotiate, and use extra schemas without forking the standard.
Client Request
Client includes X-A2A-Extensions header with comma-separated list of extension URIs
Agent Processing
Agent identifies supported extensions and performs activation. Unsupported extensions ignored.
Response
Response includes X-A2A-Extensions header listing all successfully activated extensions
Security Architecture
🔑 Authentication
OAuth 2.0, OpenID Connect, API keys, digital signatures. Multi-factor with behavioral and contextual checks.
🔐 Encryption
HTTPS/TLS required. Quantum-resistant schemes supported. Rapid key rotation.
🛡️ Authorization
Granular access controls via Agent Cards. Fine-grained permissions and enterprise policy integration.
📋 Auditing
Full audit trail support. Standard metadata tokens for governance and policy enforcement.
A2A ensures agents can collaborate like services in a network—predictable, secure, and interoperable across organizational and technological boundaries.