Agent2Agent
Open protocol for agent-to-agent delegation across vendors and frameworks.
About
Agent2Agent, usually shortened to A2A, is the open standard that fills the gap MCP left behind: model-to-tool communication had a standard, but agent-to-agent communication did not. Google launched A2A in April 2025 with more than 50 partners and donated it to the Linux Foundation two months later, where it is now governed vendor-neutrally under Apache 2.0.\n\nThe protocol is deliberately narrow: an Agent Card published at a well-known URL advertises an agent's capabilities, and a client agent submits tasks with a defined lifecycle (pending, in-progress, completed, failed, canceled) over HTTP with JSON-RPC, streaming progress over SSE. Agents stay opaque, so a LangGraph agent can delegate to a CrewAI agent to a Google ADK agent without any of them knowing each other's internals. By 2026 it had reached a stable 1.0 release with official SDKs in Python, JavaScript, Java, Go, and .NET, and support inside Microsoft Copilot Studio, Azure AI Foundry, and Amazon Bedrock AgentCore.
What it can do
- Agent Card discovery advertising capabilities, skills, and auth requirements
- Structured task lifecycle with delegation across frameworks and vendors
- JSON-RPC transport with SSE streaming for long-running tasks
- OAuth 2.0, API key, and mTLS authentication options
- Opaque collaboration without exposing either agent's internals
Setup
Publish an Agent Card at /.well-known/agent.json and implement the task endpoints; official SDKs exist for Python, JavaScript, Java, Go, and .NET.
Prerequisites
An existing agent implementation that can make and receive HTTP requests.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 100; Trending score 84/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.