📄 中文摘要
当前已经存在许多代理工作流，来自不同平台的编码代理变得相当强大，具备记忆、工具调用等多种实用功能。在实际应用中，用户常常需要使用多个平台的工具，如 Claude Code、Copilot、OpenCode 等。每种工具各有优劣，但让这些命令行工具之间进行交互并持续互动仍然面临挑战。在 openclaw 项目下，有一个名为 acpx 的工具，它采用了不同的方向，要求对每个代理进行定制，并通过结构化协议使 AI 代理与编码代理之间能够进行交流。
让 Copilot 管理 Qwen：代理间命令行协作的简单实验
📄 English Summary
Letting Copilot Manage Qwen: A Simple Experiment in Agent-to-Agent CLI Collaboration
There are numerous agent workflows available today, with coding agents from various platforms becoming increasingly robust, equipped with memory, tool calling, and other useful capabilities. In practice, users often utilize tools from multiple platforms, such as Claude Code, Copilot, and OpenCode, each with its own strengths and weaknesses. However, facilitating interaction among these CLI tools and maintaining ongoing communication remains a challenge. The openclaw project introduces a component called acpx, which takes a different approach by requiring customization for each agent and enabling AI agents and orchestrators to communicate with coding agents over a structured protocol.