---
title: "Agent2Agent protocol (A2A): definition | KYA glossary"
url: https://knowyouragent.ch/en/knowledge/glossary/a2a
lang: en
updated: 2026-10-12
license: © Innopulse Consulting GmbH
---

# Agent2Agent protocol (A2A)

The Agent2Agent protocol (A2A) is an open protocol through which AI agents from different vendors discover each other, hand over tasks and exchange results.

Google introduced A2A in April 2025 and later handed the project to the Linux Foundation. Agents describe their capabilities in an Agent Card, and other agents send them tasks, track their status and receive results, usually over HTTP and JSON. While MCP connects an agent with tools, A2A connects agents with each other.

When agents hand tasks on, a chain arises: one agent acts on behalf of another, which in turn acts on behalf of a person or company. This makes the question of who was allowed to do what on whose behalf especially important with A2A.

## Example

A travel assistant hands the hotel search to a specialized hotel agent via A2A. The hotel agent finds an offer and reports back. It may only carry out the booking if it can prove that the travel assistant passed this part of the instruction on to it with limited scope.

## How it differs

MCP connects an agent with tools and data, A2A connects agents with each other. A [delegation chain](https://knowyouragent.ch/en/knowledge/glossary/delegation-chain) in KYA describes how authority is passed along such a chain in a provable way while only being narrowed.

## In practice

- Hand tasks on via A2A only with limited, provable scope, never with blanket access.
- Record which agent handed which task to whom, so the chain stays traceable later.
- Check actions with consequences at the last agent in the chain against the original mandate.

## Role in KYA

KYA does not verify A2A messages, but it verifies the evidence that should accompany a hand-over: a mandate from the [principal](https://knowyouragent.ch/en/knowledge/glossary/principal) and, when passed on, a delegation chain in which each level only narrows authority.

Related check: [Delegation chain](https://knowyouragent.ch/en/method/checks) (`delegation_provenance`). Is every hop signed, connected without gaps and no broader than the previous one?

[Verify now: open a matching example](https://knowyouragent.ch/en/verify?beispiel=paket)

## Frequently asked questions

### What is the difference between A2A and MCP?

MCP connects an agent with tools and data sources. A2A connects agents with each other so they can hand over tasks and exchange results.

### Who is behind A2A?

Google introduced A2A in April 2025 and handed the project to the Linux Foundation, where it is developed further as an open project.

### How do agents prove their authority with A2A?

The protocol governs communication and authentication. Proof of authority along the chain also requires mandates and delegations with clear limits.

### Can KYA verify an A2A chain?

KYA verifies the evidence accompanying a chain: mandate, delegation levels, decision and receipt. It does not verify the A2A messages themselves.

## Sources

- [Agent2Agent protocol (A2A), repository](https://github.com/a2aproject/A2A)

## Related terms

- [Agent Card](https://knowyouragent.ch/en/knowledge/glossary/agent-card)
- [Model Context Protocol (MCP)](https://knowyouragent.ch/en/knowledge/glossary/mcp)
- [Delegation chain](https://knowyouragent.ch/en/knowledge/glossary/delegation-chain)
- [AI agent](https://knowyouragent.ch/en/knowledge/glossary/ai-agent)

---
Quelle: https://knowyouragent.ch/en/knowledge/glossary/a2a · KYA – Know Your Agent