What is the A2A (Agent2Agent) protocol?
A2A (Agent2Agent) is an open protocol that lets AI agents built by different teams or vendors discover each other through Agent Cards and delegate tasks to each other over HTTP.
Checked
MCP connects an agent to tools and data. A2A connects an agent to other agents. When a procurement agent at one company needs a quote from a supplier's agent, or an internal agent hands a task to one built by another team on a different framework, A2A is the shared format for that exchange.
Google announced A2A on 2025-04-09. It has since moved to the Linux Foundation and reached a stable 1.0 release.
How A2A works
Agent Cards
Each agent publishes a JSON Agent Card describing what it can do and how to reach it, so other agents can discover it.
Tasks with a lifecycle
Work is exchanged as tasks that move through defined states, which suits jobs that take minutes or hours rather than one reply.
Standard web transport
Messages travel over HTTP, with streaming for progress updates.
Status as of October 2026
A2A v1.0.0 shipped on 2026-03-12, followed by v1.0.1 on 2026-05-28. Version 1.0 renamed operations to PascalCase (SendMessage, GetTask and a new ListTasks), unified the Part type, added supportedInterfaces with JWS signatures to the Agent Card, and added multi-tenancy, cursor pagination and the OAuth Device Code flow.
On 2026-08-27 A2A joined the Linux Foundation's Agentic AI Foundation as a Growth Stage project. The project reports more than 150 backing organisations, official SDKs in Python, JavaScript, Java, C#/.NET, Go and Rust, and adoption in frameworks including LangGraph, CrewAI, Pydantic AI, AG2 and IBM BeeAI.
How Motiontic's engineers apply it
Most first agent builds do not need A2A: one agent with good tools covers the job. Motiontic's engineers reach for it when agents cross a boundary that tools cannot, such as two teams, two vendors or two companies, and each side runs its own agent on its own stack.
When A2A is used, the Agent Card is treated as a public contract, with authentication, scoped permissions and logging on every task that crosses it.
MCP, A2A and AG-UI compared
| MCP | A2A | AG-UI | |
|---|---|---|---|
| Connects | An agent to tools and data | An agent to other agents | An agent to a user-facing app |
| Originated at | Anthropic | CopilotKit | |
| Governance | Linux Foundation AAIF | Linux Foundation AAIF | Not an AAIF project |
| Typical use | Read a database, call an API | Delegate a task to another team's agent | Stream an agent's progress into a front end |
Common questions
What is the A2A protocol used for?
A2A lets AI agents from different teams, frameworks or vendors find each other through Agent Cards and hand tasks back and forth over a shared, open format.
What is the difference between A2A and MCP?
MCP connects an agent to tools and data sources. A2A connects agents to other agents. The two are complementary, and both are now projects of the Linux Foundation's Agentic AI Foundation.
Who governs A2A?
A2A started at Google, was donated to the Linux Foundation and joined the Agentic AI Foundation on 2026-08-27. Its technical steering committee includes AWS, Cisco, Google, IBM Research, Microsoft, Salesforce, SAP and ServiceNow.
Is A2A stable enough for production?
A2A reached v1.0.0 on 2026-03-12, with official SDKs in six languages. Whether it fits a given system depends more on whether agents genuinely need to talk across team or company boundaries.
Related
MCP vs A2A
The protocol for tools and data, compared with the protocol for agents talking to agents.
AG-UI
The protocol between an agent backend and the app a person is using.
Single Agent vs Multi-Agent
One loop with good tools, compared with several agents splitting the work.
AI agent development
Agents that do real work for users, with the evals, traces and guardrails to keep them in production.
Agent Card (A2A)
The business card an A2A agent publishes so other agents can find and call it.
Running agents that have to hold up?
Describe the agent, the systems it touches and what it must never get wrong. The first call is with a lab engineer, who will say plainly whether it is ready for production, and whether Motiontic is the right team.