Agent Communication Protocol
Also called ACP
An open protocol that lets AI agents built by different teams send each other messages over standard web requests. Not to be confused with the Agentic Commerce Protocol, which shares the abbreviation.
What is the Agent Communication Protocol?
The Agent Communication Protocol (ACP) is an open standard for communication between AI agents. It defines how one agent can discover another and exchange messages and results using ordinary web requests, so agents written in different frameworks can work together.
How does it work?
Each agent exposes a standard interface. Another agent or app can list what it offers, send it a task and receive the result, including streamed updates. It was developed as an open standard under the Linux Foundation, with BeeAI as its reference implementation, and uses simple REST endpoints. Its goals overlapped with the Agent2Agent protocol, and ACP is now part of A2A, so check the current A2A specification before building on it.
What is a simple example?
A research agent built in one framework asks a separate summarising agent, built in another, to condense a set of documents. Because both speak the same protocol, no custom integration is needed.
Why does it matter?
Shared protocols stop agents from being locked inside one vendor's ecosystem. Note that "ACP" is also used for the Agentic Commerce Protocol, which covers a different problem.