dmux-workflows

Coordinate parallel AI agent sessions across dmux in tmux panes.

Updated Apr 21, 2026
One-click install
npx skills add https://github.com/choub00/transformer --skill dmux-workflows-choub00
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: dmux-workflows
Source: https://github.com/choub00/transformer/tree/main/.cursor/skills/dmux-workflows
Command: npx skills add https://github.com/choub00/transformer --skill dmux-workflows-choub00

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

dmux provides a structured way to orchestrate multiple agent sessions simultaneously, reducing context switching and improving collaboration across tools.

Core Features & Use Cases

  • Independent parallel panes: run separate agent sessions concurrently for research, implementation, and review.
  • Cross-harness coordination: works with Claude Code, Codex, OpenCode, and other harnesses to streamline multi-tool workflows.
  • Best-practices workflow: use git worktrees and pane merging to keep work isolated yet easily integrated.

Quick Start

Start a dmux session, then create and manage panes to run parallel agent prompts.

Frequently Asked Questions about dmux-workflows

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run parallel AI agent sessions in tmux?

To run parallel AI agent sessions in tmux, you can use dmux to create independent panes that execute separate agent harnesses concurrently for research, implementation, and review tasks.

What is cross-harness coordination for AI agents?

Cross-harness coordination is the process of managing multiple AI tools like Claude Code, Codex, and OpenCode simultaneously. dmux orchestrates these separate agent sessions to streamline multi-tool workflows and merge outputs.

Can I use dmux to coordinate Claude Code and Codex sessions together?

Yes, dmux supports cross-harness coordination, allowing you to run Claude Code, Codex, and OpenCode agent sessions in separate panes and merge their outputs efficiently within a tmux environment.

How do I start a multi-agent workflow with dmux?

To start a multi-agent workflow, initiate a dmux session, then create and manage panes to run parallel agent prompts across different harnesses, keeping work isolated using git worktrees.

What is the best way to keep parallel AI coding agents isolated but integrated?

The best way to isolate but integrate parallel AI agents is using git worktrees alongside dmux pane management, which keeps separate agent sessions independent while allowing structured workflow merging.