Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Dmux Multi Agent Workflows

ASecurity

当需要在 tmux 面板里并行跑多个 AI Agent 会话、跨 Claude Code/Codex/OpenCode 等 harness 分而治之时使用;用 dmux 按 n 开面板派任务、按 m 合并产出,配 git worktree 隔离冲突文件、面板数控制在 5-6 以内;不适用于任务彼此依赖输出、需 Agent 间实时通信、或单会话内编程式并行(改用 Task 工具)的场景;触发词:dmux、并行跑 Agent、多 Agent、split this work、tmux 面板编排

3 stars
0 votes
0 copies
1 views
Added 9/19/2026
ai-agentsbashnodeexpressgitapi

Works with

claude codecursorcliapi

Security Analysis

A92/100
mediumInstalls packages at runtime which could introduce malicious dependencies

Pro shows the line behind each finding and how to fix it

Scanned 9/19/2026

$npx -y skills add findscripter/everything-skills --skill dmux-multi-agent-workflows --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Dmux Multi Agent Workflows?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Dmux Multi Agent Workflows
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/findscripter-dmux-multi-agent-workflows/badge)](https://www.skillsdirectory.com/skills/findscripter-dmux-multi-agent-workflows)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
Files
SKILL.md
---
name: dmux-multi-agent-workflows
title: dmux 多 Agent 并行编排(tmux 面板)
description: 当需要在 tmux 面板里并行跑多个 AI Agent 会话、跨 Claude Code/Codex/OpenCode 等 harness 分而治之时使用;用 dmux 按 n 开面板派任务、按 m 合并产出,配 git worktree 隔离冲突文件、面板数控制在 5-6 以内;不适用于任务彼此依赖输出、需 Agent 间实时通信、或单会话内编程式并行(改用 Task 工具)的场景;触发词:dmux、并行跑 Agent、多 Agent、split this work、tmux 面板编排
domain: 智能/agents
triggers: [用 dmux 编排, 并行跑多个 Agent, split this work / 拆分并行, 多 Agent 工作流, tmux 面板管理 Agent, 跨 Claude Code 和 Codex 协作, research 和 implement 分两条轨, test 与 fix 双面板循环, 并行代码评审多视角, git worktree 隔离每个面板]
tags: [多智能体, 并行编排, dmux, tmux, git-worktree, 跨harness, claude-code, codex, divide-and-conquer]
level: 进阶
status: stable
agents: [claude-code, codex, cursor, gemini-cli]
tools: [dmux, tmux, git, claude-code, codex]
requires: []
related: [parallel-agent-hub, multi-agent-orchestrator, multi-agent-system-designer, autonomous-coding-agent-patterns]
combines_with: [git-worktrees-workflow, tmux-session-management, coding-agent-headtohead-eval]
license: MIT
source: affaan-m/ECC
source_license: MIT
---
## 何时使用

当一个任务可以拆成**互相独立、可分而治之**的子任务,且并行推进比单线迭代更划算时,用 dmux 在 tmux 面板里同时跑多个 Agent 会话。典型场景:

- 同时跑多个 Agent 会话(每个面板一个独立 harness 实例)。
- 跨 Claude Code、Codex、OpenCode 等不同 harness 协调同一项目的工作。
- 复杂任务的分而治之并行(多文件功能、并行评审、测试+修复双轨)。
- 用户说「并行跑」「split this work」「用 dmux」「多 Agent」。

**不该用的边界:**
- 子任务**彼此依赖对方输出**——别把有先后依赖的步骤并行(会拿到半成品)。
- 需要 Agent 之间**实时通信/协作**——dmux 面板各自独立,靠人工 `m` 合并单向汇总。
- 改动会**相互冲突的全局重构**——除非每个面板用独立 git worktree 隔离。
- **单会话内的编程式并行**——那是 Claude Code Task 工具的活,不必起 dmux。

## 步骤

dmux 是基于 tmux 的 Agent 面板编排器。安装:`npm install -g dmux`(详见 github.com/standardagents/dmux)。支持 Claude Code、Codex、OpenCode、Cline、Gemini、Qwen。

1. **启动**:终端跑 `dmux` 进入会话。
2. **开面板派任务**:按 `n` 新建面板并输入该面板的 prompt(一个面板 = 一个独立 Agent 会话,负责一块**边界清晰**的工作)。
3. **监控**:各面板各自跑自己的 Agent;面板无响应时按 `m` 读其输出,确认是否在等待输入。
4. **合并产出**:按 `m` 把某面板结果合并回主会话上下文。**合并前先 review** 输出,避免引入冲突。
5. **(冲突高发时)worktree 隔离**:对会触碰重叠文件的任务,先建 git worktree,每个面板在各自 worktree 里 `claude`,完成后回主分支 `git merge`。

## 指令

核心快捷键(在 dmux 会话内):

- `n` —— 新建面板并附带 prompt,启动一个新 Agent 会话。
- `m` —— 把面板输出合并回主会话(也用于读取面板当前输出)。

git worktree 隔离(重叠文件并行的标准做法):

```bash
# 为每条特性建独立 worktree
git worktree add ../feature-auth    feat/auth
git worktree add ../feature-billing feat/billing

# 不同面板进各自 worktree 跑 Agent
# Pane 1: cd ../feature-auth    && claude
# Pane 2: cd ../feature-billing && claude

# 完成后回主分支合并
git merge feat/auth
git merge feat/billing
```

最佳实践(硬约束):

1. **只并行独立任务**——有依赖关系的不要并行。
2. **边界清晰**——每个面板只动各自的文件/关注点。
3. **合并要克制**——合并前 review 面板输出防冲突。
4. **冲突用 worktree**——文件易冲突的工作每面板一个 worktree。
5. **资源意识**——每个面板都是一个完整 Agent 会话、消耗 API token,**总面板数控制在 5-6 以内**。

互补工具选型:

| 工具 | 作用 | 何时用 |
|---|---|---|
| dmux | tmux 面板管理 Agent | 并行 Agent 会话 |
| Superset | 终端 IDE,10+ 并行 Agent | 大规模编排 |
| Claude Code Task 工具 | 进程内派生子 Agent | 单会话内编程式并行 |
| Codex 多 Agent | 内建 Agent 角色 | Codex 专属并行 |

## 示例

常用编排模式(每行一个面板的 prompt):

**模式 1 · 研究 + 实现(research/implement 双轨)**
```
Pane 1 (Research):  "调研 Node.js 限流最佳实践,对比现有库与方案,把结论写到 /tmp/rate-limit-research.md"
Pane 2 (Implement): "为 Express API 实现限流中间件,先做基础 token bucket,待研究完成再精修"
# Pane 1 完成后,把结论合并进 Pane 2 的上下文
```

**模式 2 · 多文件功能(独立文件并行)**
```
Pane 1: "为计费功能创建数据库 schema 与迁移"
Pane 2: "在 src/api/billing/ 实现计费 API 端点"
Pane 3: "创建计费仪表盘 UI 组件"
# 全部合并后,在主面板做集成
```

**模式 3 · 测试 + 修复循环**
```
Pane 1 (Watcher): "watch 模式跑测试套件,失败时汇总失败项"
Pane 2 (Fixer):   "根据 Pane 1 的错误输出修复失败测试"
```

**模式 4 · 跨 harness**
```
Pane 1 (Claude Code): "评审 auth 模块的安全性"
Pane 2 (Codex):       "为性能重构工具函数"
Pane 3 (Claude Code): "为结算流程写 E2E 测试"
```

**模式 5 · 并行评审流水线(多视角)**
```
Pane 1: "评审 src/api/ 的安全漏洞"
Pane 2: "评审 src/api/ 的性能问题"
Pane 3: "评审 src/api/ 的测试覆盖缺口"
# 三份评审合并为一份报告
```

## 注意事项

- **面板无响应**:多半是 Agent 会话在等待输入,按 `m` 读输出确认。
- **合并冲突**:用 git worktree 把各面板的文件改动隔离开。
- **token 高耗**:减少并行面板数——每个面板都是一个完整 Agent 会话。
- **tmux 未安装**:macOS `brew install tmux`,Linux `apt install tmux`。
- dmux 是 tmux 之上的编排层,强依赖本机 tmux 环境,不适用于纯 Windows 原生终端(用 WSL/Linux/macOS)。

## 互见

- related:`parallel-agent-hub` —— 同一任务上多 Agent 竞赛择优(worktree 隔离 + 指标/LLM 评审),比 dmux 的人工合并更自动化。
- related:`multi-agent-workflow-designer` —— 先用它选编排模式(顺序/并行/路由/编排器/评估器)画蓝图,再用 dmux 落到面板执行。
- related:`multi-agent-system-designer`、`autonomous-coding-agent-patterns`、`agent-workflow-pattern-designer`
- combines_with:`git-worktrees-workflow` —— 为高冲突并行任务提供分支/工作树隔离,是模式 2/全局重构的前提。
- combines_with:`crewai-multi-agent` —— 需要进程内、有结构化角色与通信的多 Agent 系统时改用它(dmux 是面板级、无 Agent 间通信)。

---
采编自 affaan-m/everything-claude-code(MIT 许可),适配重写为中文版,非逐字翻译。

Attribution

findscripterfindscripter
View sourceSee grades on GitHubMore from findscripter →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

698621 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

741 votes
View all in ai-agents →