​
What is ACP?

The Agent Communication Protocol (ACP) is an open standard with open governance enabling seamless communication between AI agents, regardless of their implementation details. In ACP, an agent is a software service that communicates through multimodal messages, primarily driven by natural language. The protocol is agnostic to how agents function internally, specifying only the minimum assumptions necessary for smooth interoperability. ACP defines a standardized RESTful API for managing and executing agents, supporting synchronous, asynchronous, and streaming interactions.

​
Motivation

Enterprises today run hundreds to thousands of agents in silos, built on disparate technology stacks and unable to communicate. ACP unlocks agent-to-agent communication by standardizing interfaces, enabling agents from any technology stack to collaborate effectively and securely.

ACP is developed under open governance as a Linux Foundation project, alongside BeeAI, an open platform to discover and run agents from any technology stack. This is because the ACP team believes that the standard should be community-driven and usable through an open implementation, available to all. No vendor lock-in, no adherence to a proprietary ecosystem.

​
Design Philosophy

  • Simplicity First: Easy to implement basic functionality.
  • Progressive Complexity: Clear pathways to advanced capabilities without initial complexity.
  • Minimal Assumptions: No assumed orchestration patterns or architectural requirements.

​
Core Values

ACP is designed to deliver on core, differentiating values:

  • REST-based Communication: ACP uses simple, clearly defined REST endpoints, aligning closely with standard HTTP patterns. Unlike protocols relying on more complex communication methods (such as JSON-RPC), ACP leverages widely-used HTTP conventions to simplify usage and integration into production environments.
  • No SDK Required (but we made one anyway): ACP is simple enough that you don’t need an SDK or specialized libraries. You can easily interact with ACP-compatible agents using standard HTTP tools like curl, Postman, or even your browser. Of course, since we’re developers too, we built an SDK anyway, just to make life even easier.
  • Offline Discovery: Agents remain discoverable even when inactive, thanks to ACP’s manifest-based discovery. This enables agents to scale to zero while still being available for discovery and activation as needed.
  • Async-first, Sync Supported: ACP is designed primarily for asynchronous communication, ideal for agents handling longer-running tasks. However, synchronous communication is also fully supported by default, simplifying development, rapid testing, and common use cases.