Agent Client Protocol vs Model Context ProtocolRemoved
Blog post from Credal
In 2025, the introduction of the Agent Client Protocol (ACP), developed by JetBrains and Zed Industries, marked a significant advancement in the integration of coding agents with code editors, addressing the fragmentation problem in IDEs caused by diverse agent integrations. Inspired by the Language Server Protocol (LSP), ACP provides a standardized, editor-agnostic framework that allows agents to control UI components, run subprocesses, and initiate multi-buffer reviews within editors, enabling seamless communication between local and remote environments without the need for custom integrations. ACP complements the Model Context Protocol (MCP), which manages communication between agents and various servers and tools, together enhancing interoperability across agents, editors, and tools in AI-enabled ecosystems. While ACP facilitates agent flexibility and autonomy within different code editors, MCP allows agents to connect to essential tools and data for their functions, enabling developers to forego tedious one-off integrations and allowing users to select optimal tools for their workflows.