Home 零对冲(ZeroHedge)每日HackerNews

用于隔离、并行代理开发的桌面应用程序。
A desktop app for isolated, parallel agentic development

原始链接: https://github.com/coder/mux

## mux：并行代理开发 mux 是一款桌面应用程序，专为并行代理工作流程设计，旨在提升开发者效率。它允许用户同时运行多个 AI 代理——利用如 GPT-5-Pro、Grok 和本地 Ollama LLM 等模型——来处理代码审查、重构和新功能开发等任务。 主要功能包括跨变更的上下文连续性、自动恢复的后台处理，以及不同方案的 A/B 测试。工作区可以通过 Git 工作树本地隔离，或通过 SSH 远程隔离。VS Code 扩展提供了无缝集成。 mux 提供了一种类似于 Claude Code 的用户体验，具有计划/执行模式等功能，以及用于管理代理、跟踪成本和审查复杂输出（包括 Mermaid 图表）的独特工具。目前处于预览阶段，对于那些能够容忍潜在错误并寻求增强的代理辅助开发的用户来说，它是一个强大的工具。它是开源的，适用于 macOS 和 Linux。

## AI 编码代理工作流挑战与解决方案 这次 Hacker News 讨论围绕着使用 AI 编码代理进行开发时面临的挑战。一个关键问题是，大多数代理直接修改本地文件，在与 Git 协作和同时编辑时会产生冲突。用户希望获得更精简的“内存中”解决方案，代理以补丁形式提出更改，允许受控应用。 几种解决方案被提出：**Git 工作树**是一种常见的解决方法，但当代理数量很多时可能会变得繁琐。**Docker/虚拟机**提供更强的隔离性，但通常被认为过于笨重，尤其是在 macOS 上。**Isomorphic Git** 和 **Cursor 的“Shadow workspace”** 等工具旨在进行内存中操作，但引发了对测试和编译的担忧。 讨论中提到了几个项目，包括 **workmux**、**sandvault/clodpod**、**OrbStack** 和 **conductor.build**。对话强调了对更好的异步处理、代理通信以及潜在的更细粒度的项目结构的需求，以便有效地利用并行代理能力。最终，用户希望工具能够自动化隔离和工作流管理，同时不牺牲简单性和控制力。
相关文章
原文

mux product screenshot

A desktop application for parallel agentic development.

Why parallelize?

Here are some specific use cases we enable:

  • Contextual continuity between relevant changes:
    • e.g. create a workspace for code-review, refactor, and new-feature
  • GPT-5-Pro: use the slow but powerful GPT-5-Pro for complex issues
    • Run in the background for hours on end
    • The stream will automatically resume after restarts or intermittent connection issues. If the model completes early we will show an indicator.
  • A/B testing: run multiple workspaces in parallel on the same problem but different approaches, abandon the bad ones.
  • Tangent exploration: launch tangents in mux away from main work
  • Isolated workspaces with central view on git divergence
    • Local: git worktrees on your local machine (docs)
    • SSH: regular git clones on a remote server (docs)
  • Multi-model (sonnet-4-*, grok-*, gpt-5-*, opus-4-*)
    • Ollama supported for local LLMs (docs)
    • OpenRouter supported for long-tail of LLMs (docs)
  • VS Code Extension: Jump into mux workspaces directly from VS Code (docs)
  • Supporting UI and keybinds for efficiently managing a suite of agents
  • Rich markdown outputs (mermaid diagrams, LaTeX, etc.)

mux has a custom agent loop but much of the core UX is inspired by Claude Code. You'll find familiar features like Plan/Exec mode, vim inputs, /compact and new ones like opportunistic compaction and mode prompts.

Read the full documentation →

Warning

mux is in a Preview state. You will encounter bugs and performance issues. It's still possible to be highly productive. We are using it almost exclusively for our own development.

Download pre-built binaries from the releases page for macOS and Linux.

More on installation →

Integrated code-review for faster iteration:

Screenshot of code review

Agents report their status through the sidebar:

Screenshot of agent status

Git divergence UI keeps you looped in on changes and potential conflicts:

Screenshot of git status

Mermaid diagrams make it easier to review complex proposals from the Agent:

Screenshot of mermaid diagram

Project secrets help split your Human and Agent identities:

Screenshot of project secrets

Stay looped in on costs and token consumption:

Screenshot of costs table

Opportunistic compaction helps keep context small:

Screenshot of opportunistic compaction

See the documentation for more details.

See AGENTS.md for development setup and guidelines.

Copyright (C) 2025 Coder Technologies, Inc.

This program is free software: you can redistribute it and/or modify it under the terms of the GNU Affero General Public License as published by the Free Software Foundation, version 3 of the License.

See LICENSE for details.

联系我们 contact @ memedata.com