What is Pane used for?

Running several AI coding agents at once, such as Claude Code, Codex or Aider, each on its own git worktree, then reviewing and shipping what they changed. The use cases here show the most common workflows step by step.

Which use case should I start with?

Parallel features. It shows the core loop the others build on: a pane per feature, a worktree per pane, review, commit and merge.

Do I need several agents to get value from Pane?

No. One agent still gets its own worktree, a diff view and git shortcuts. Pane pays off most once you run two or more tasks at once.

Can I run these workflows on Windows?

Pane runs on Windows, macOS and Linux. Cross-machine workflows have extra requirements and WSL limits; check the individual guide before choosing a destination.

Can I check on agents from my phone?

Yes, two ways: Remote Pane in your phone's browser, or Claude Code's Remote Control in Pane Chat through the Claude app. Pane's own iOS and Android app is coming soon.

Use cases - Pane

Use cases

TL;DR

Pane pays off when you have more work that can run in parallel than you can watch at once. These pages walk through practical workflows step by step: shipping several features, refactoring many packages, reviewing agent work, running agents on a bigger machine, steering everything from your phone, checking another OS and handing work to another machine.

Workflows

Each page has a step-by-step walkthrough and the shortcuts to use.

Coding agents across Mac, Windows and WSL →

Use Pane Workspaces to run commands and read files on your own machines. Learn where Mac, Windows and WSL fit, and when to hand off a task.

Cross-OS QA with AI coding agents →

Check an AI coding change on Mac and Windows with Pane Workspaces. Compare the same commit, run native checks and keep evidence for each OS.

Hand off a coding agent task to another machine →

Continue a coding task on another machine with runpane handoff. Prepare a verified note, preview the destination and start a fresh agent at your code revision.

Ship three features at once →

One agent per feature, each on its own worktree, so nothing collides while they work. About two and a half hours instead of six and a half.

Refactor five packages in parallel →

One spec, one pane per package. Or run several attempts on the same package and keep the best.

Review agent work faster →

The agent and its diff sit in the same pane. Read, ask for a fix, re-check and commit without leaving the window.

Run agents on a bigger machine →

Keep Pane on your laptop while worktrees, terminals and agents run on a desktop, server or cloud VM.

Drive every agent from your phone →

Turn on Claude Code's Remote Control in Pane Chat, and one chat on your phone creates panes, hands out work, unblocks agents and reports back.

Where to start

New to running agents in parallel? Start with parallel features.

Parallel features shows the core loop: a pane per feature, a worktree per pane, review and merge. The others build on it. For background on the idea, read what is an agent manager? or git worktrees for AI agents.

Pane showing Claude Code running in a terminal, parallel panes in the sidebar and the Details panel with git actions — Agents running in parallelEach pane runs the agent's real terminal UI, with several panes going at once.
1.Parallel panes: Each pane is its own worktree. The dots show which are working, finished or waiting on you.
2.The real agent: It's the actual Claude Code terminal UI, so every feature and shortcut works.
3.Git at hand: Fetch, commit, push, pull, rebase and merge, each with a shortcut.
Pane 2.4.155 on Windows 11, repo inside WSL2

来源

勘误

如果这里有任何错误或过时的内容，请告诉我们，我们会修正并更新顶部的日期。在 GitHub 上提交 issue

来源

→ What is an agent manager?
→ Pane docs

常见问题

或

下载*

Windows SmartScreen 警告Pane 尚未签名，直接下载时可能会出现 SmartScreen 警告。 Pane 完全开源，你可以审查代码，也可以自己从源码构建。1. 点击“更多信息”2. 点击“仍要运行”3. 继续安装用 PowerShell 安装会直接下载官方版本，避开大多数浏览器下载限制。npm 全局安装

把它粘贴到 PowerShell 中。Linux

作者 Parsa Khazaeepoul，Pane 联合创始人。最后审查于 2026年10月.