tmux-ide is an open-source CLI tool (npm i -g tmux-ide) that turns any coding project into a full terminal IDE using a single YAML config file. It manages tmux sessions, pane layouts, dev servers, and Claude agent teams automatically from one command.

How do Claude agent teams work in tmux-ide?

A lead Claude instance runs in one tmux pane and can spawn additional Claude teammates into their own panes on the fly. Agents coordinate through a shared task list — the lead assigns work, teammates claim tasks and report back, all in parallel.

Do I need tmux installed to use tmux-ide?

Yes — tmux-ide manages tmux sessions under the hood for split-pane mode. The Claude Code agent teams feature requires tmux or iTerm2. It does not work in VS Code's integrated terminal or Windows Terminal.

What stacks does tmux-ide support?

tmux-ide auto-detects Next.js, Vite, Python, Go, and more — spinning up the appropriate dev server alongside your agent team. The layout is defined in ide.yml and is fully reproducible across machines.

What is the Claude Code skill that installs automatically?

When you install tmux-ide, it registers a Claude Code skill so Claude can configure your workspace automatically — detecting your stack, generating the ide.yml layout, and starting the agent team without manual setup.

Is tmux-ide the same as Claude Code's built-in agent teams?

No — Claude Code has its own agent teams feature (research preview) that also uses tmux panes. tmux-ide is an independent open-source layer on top that adds declarative YAML config, self-organizing layouts, and a purpose-built IDE experience around the same tmux+Claude pattern.

tmux-ide: Declarative Terminal IDE with Native Claude Agent Teams

By Prahlad Menon Published 2026-03-13 3 min read

The IDE has always assumed you want a GUI. tmux-ide doesn’t.

tmux-ide — built by Thijs Verreck — is a fully open-source CLI that turns any project into a structured terminal IDE with one YAML file and one command. The genuinely interesting part: Claude agent teams are baked in from the start, not bolted on after the fact.

What it actually does

npm i -g tmux-ide
tmux-ide

That’s the install and the run command. What happens next: tmux-ide reads your ide.yml, spins up a tmux session with your configured pane layout, starts whatever dev server your stack needs (Next.js, Vite, Python, Go — it auto-detects), and launches Claude instances in their designated panes.

The ide.yml is declarative:

layout: lead-plus-teammates
teammates: 2
devServer: true
tasks:
  - "Implement the auth flow in /src/auth"
  - "Write tests for the user model"

Edit the YAML, the layout updates live. The same config file works identically on any machine — your terminal IDE is now version-controllable and reproducible.

The agent teams part

This is where it gets genuinely different from a fancy tmux script.

The lead Claude doesn’t just sit in a pane and wait for input. It analyzes the project, decides it needs help, and spawns teammates dynamically — each in its own pane. Tasks are distributed from a shared task list. Teammates claim tasks, do the work, and report back. The lead can reassign work, spawn new teammates mid-session, and restart the layout from within its own session.

That’s a self-organizing agent team running entirely in your terminal, coordinated through shared state, without any cloud orchestration layer.

The install script also registers a Claude Code skill automatically — so if you ask Claude to set up your workspace, it handles stack detection, layout generation, and config without you touching the YAML.

How it relates to Claude Code’s native agent teams

Claude Code (Anthropic’s coding agent) has its own agent teams feature in research preview, which also uses tmux panes. tmux-ide is a separate, independent open-source project that builds on the same tmux + Claude pattern but adds:

Declarative YAML config — reproducible layouts defined as code
IDE-level structure — dev server, shell pane, agent panes all managed together
Self-organizing lead — the lead spawns teammates on demand, not just on startup
One-command setup — no manual tmux session management

The two aren’t competing — tmux-ide is closer to infrastructure for running Claude agent teams than a fork of Claude Code’s feature.

Who this is for

Terminal-native developers who already live in tmux will get this immediately. If you’re building with Claude Code and you’re tired of managing session splits by hand, ide.yml is a meaningful quality-of-life upgrade.

It’s also worth paying attention to as a pattern: declarative, reproducible, version-controlled dev environments — except the “environment” now includes the AI agents working inside it, not just the toolchain. That’s a genuinely new thing.

If tmux still feels like archaeology to you and you need VSCode’s file tree to function, this isn’t for you. That’s fine.

For everyone else: npx tmux-ide works without installing anything.