Is there a limit to how many agents I can run at once?

While the tool supports many panes, it is recommended to keep parallel sessions under 5-6 agents to avoid excessive API token consumption and to maintain manageable oversight of the changes.

When should I merge agent outputs?

You should use the merge feature (typically triggered by the 'm' key in dmux) once a sub-agent has completed a specific sub-task, such as generating a research report or finishing a component, to bring that context back to your primary session.

How does this skill prevent merge conflicts between agents?

The skill provides patterns for using Git worktrees, which isolate each agent into its own temporary directory and branch, preventing them from overwriting the same files concurrently.

Can I use this with AI tools other than Claude Code?

Yes, dmux-workflows is designed to be cross-compatible with Claude Code, Codex, OpenCode, Cline, Gemini, and Qwen CLI tools.

What is dmux in the context of AI agents?

dmux is a tmux-based pane manager specifically designed to allow you to run and coordinate multiple AI agent sessions (like Claude Code) simultaneously within a single terminal window.

DMUX Multi-Agent Workflows

Name: DMUX Multi-Agent Workflows
Author: affaan-m

byaffaan-m

•

156,033

•

생산성 및 워크플로

Orchestrates parallel AI agent sessions using dmux to manage complex multi-step development tasks efficiently.

DMUX Multi-Agent Workflows is a specialized orchestration system designed for power users of Claude Code and other AI coding agents. By leveraging the dmux tmux-based pane manager, it enables developers to split complex projects into independent, parallel tracks—such as simultaneous research and implementation or concurrent testing and bug fixing. It facilitates seamless coordination between multiple agents, provides specialized patterns for merging outputs, and integrates with Git worktrees to prevent file conflicts, significantly increasing development velocity for large-scale engineering tasks.

주요 기능

01Pre-defined workflow patterns for Research/Implementation and Test/Fix loops

02Parallel agent orchestration via tmux-based pane management

03156,033 GitHub stars

04Git worktree integration for isolated, conflict-free parallel development

05Seamless merging of multiple agent session outputs into a master context

06Cross-suite support for Claude Code, Codex, OpenCode, and other CLI agents

사용 사례

01Automating a 'Monitor + Repair' cycle where one agent watches test logs and another applies fixes

02Running a dedicated security audit agent while a second agent refactors core logic

03Splitting a large feature into parallel UI, API, and database implementation tasks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add affaan-m/everything-claude-code dmux-workflows

For use in Claude.ai and ChatGPT

주요 기능

01Pre-defined workflow patterns for Research/Implementation and Test/Fix loops

02Parallel agent orchestration via tmux-based pane management

03156,033 GitHub stars

04Git worktree integration for isolated, conflict-free parallel development

05Seamless merging of multiple agent session outputs into a master context

06Cross-suite support for Claude Code, Codex, OpenCode, and other CLI agents

사용 사례

01Automating a 'Monitor + Repair' cycle where one agent watches test logs and another applies fixes

02Running a dedicated security audit agent while a second agent refactors core logic

03Splitting a large feature into parallel UI, API, and database implementation tasks