ProtocolReferenceA2A

A2A

Name: A2A
Rating: 5.0 (21874 reviews)
Author: a2aproject

by a2aproject

Open protocol for interoperable, provenance-aware agent-to-agent communication

Shell

Updated Feb 11, 2026

21.9k

Stars

2.2k

Forks

Commits/Week

Commits/Month

View on GitHub

Overview

Defines an open protocol for secure communication and interoperability between opaque agentic applications. Specifies message schemas, transport patterns, and discovery primitives so independent agents can exchange tasks, credentials, and results without exposing internals. Focuses on standardizing agent-to-agent handshakes and extensible metadata for provenance and routing. See the A2A Protocol Pattern and how it maps to the Model Context Protocol (MCP).

The Value Proposition

As agents interact and delegate, a common communication layer is essential to assess who did what and why — the foundation of reproducible trust. A2A provides the plumbing that makes agent track records and cross-agent evaluation possible by standardizing message formats, provenance fields, and discovery. Until protocols like A2A exist, benchmarking and reputation systems must rely on ad-hoc integrations that lose critical context for trust and failure analysis. This aligns with the Human-in-the-Loop concept to keep oversight intact.

Ideal For

Teams building networks of heterogeneous agents that need a standard way to exchange tasks, provenance, and credentials across boundaries. This can be effectively guided by the Agent Registry Pattern to manage who or what is in the network.

Use Cases

Standardizing task handoffs and provenance between third-party agents
Capturing interoperable metadata for downstream agent-to-agent evaluation and reputation systems
Enabling discovery and secure messaging between opaque agents across environments

See related protocols

Standards this tool supports