omx

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

omx (oh-my-codex) — Multi-Agent Orchestration for Codex CLI

omx (oh-my-codex) — 面向Codex CLI的多Agent编排工具

When to use this skill

何时使用该技能

You want a stronger Codex CLI runtime with reusable role prompts and workflow skills
You need coordinated parallel agents via tmux team workers
Complex tasks requiring persistent execution (
```
$ralph
```
) or consensus planning (
```
$ralplan
```
)
You need repository exploration (
```
omx explore
```
) or bounded shell inspection (
```
omx sparkshell
```
)
Cross-model consultation with
```
$ask-claude
```
or
```
$ask-gemini
```

你需要功能更强大的Codex CLI运行时，支持可复用的角色提示词和工作流技能
你需要通过tmux团队工作器实现协同并行Agent调度
需要持久执行（
```
$ralph
```
）或共识规划（
```
$ralplan
```
）的复杂任务
你需要仓库探索功能（
```
omx explore
```
）或受限shell检查功能（
```
omx sparkshell
```
）
通过
```
$ask-claude
```
或
```
$ask-gemini
```
实现跨模型咨询

1. Quick Start

1. 快速开始

bash

npm install -g @openai/codex oh-my-codex
omx setup
omx --madmax --high

Inside Codex:

text

/prompts:architect "analyze the authentication flow"
$plan "map the safest implementation path"

Mental model: Codex does the agent work. OMX adds better prompts, reusable workflow skills, and a runtime layer around it. Start with
omx --madmax --high
and let the agent pull in
$team
,
$ralph
, or other workflows when the task needs them.

bash

npm install -g @openai/codex oh-my-codex
omx setup
omx --madmax --high

在Codex内使用：

text

/prompts:architect "analyze the authentication flow"
$plan "map the safest implementation path"

核心逻辑：Codex负责执行Agent任务，OMX为其提供更优质的提示词、可复用的工作流技能以及运行时层。推荐先使用
omx --madmax --high
启动，当任务需要时Agent会自动调用
$team
、
$ralph
或其他工作流。

2. Agent Roles (30+)

2. Agent角色（30+个）

Tier	Agents
Core Development	`architect` · `planner` · `executor` · `debugger` · `verifier` · `explore`
Quality Assurance	`style-reviewer` · `quality-reviewer` · `api-reviewer` · `security-reviewer` · `performance-reviewer`
Domain Experts	`dependency-expert` · `test-engineer` · `build-fixer` · `designer` · `writer` · `qa-tester`
Product Strategy	`product-manager` · `ux-researcher` · `product-analyst` · `information-architect`

Invoke via:

/prompts:<agent-name> "task"

层级	包含Agent
核心开发	`architect` · `planner` · `executor` · `debugger` · `verifier` · `explore`
质量保障	`style-reviewer` · `quality-reviewer` · `api-reviewer` · `security-reviewer` · `performance-reviewer`
领域专家	`dependency-expert` · `test-engineer` · `build-fixer` · `designer` · `writer` · `qa-tester`
产品策略	`product-manager` · `ux-researcher` · `product-analyst` · `information-architect`

调用方式：

/prompts:<agent-name> "task"

3. Workflow Skills (35+)

3. 工作流技能（35+个）

Skill	Trigger	Description
`autopilot`	`$autopilot`	Full autonomous pipeline — idea to working code
`ralph`	`$ralph`	Persistent execution loop until verified complete
`ultrawork`	`$ulw`	Maximum parallelism across parallel agents
`team`	`$team`	N coordinated agents with tmux/worktree isolation
`plan`	`$plan`	Strategic planning before implementation
`ralplan`	`$ralplan`	Consensus-based iterative planning
`deep-interview`	`$deep-interview`	Socratic clarification before planning or execution
`deepsearch`	`$deepsearch`	Codebase-focused deep search
`analyze`	`$analyze`	Evidence-driven causal analysis (trace methodology)
`tdd`	`$tdd`	Test-first development (red-green-refactor)
`build-fix`	`$build-fix`	Fix build errors, type errors, toolchain failures
`code-review`	`$code-review`	Comprehensive multi-dimension code review
`security-review`	`$security-review`	Security audit — vulnerabilities, trust boundaries
`ultraqa`	`$ultraqa`	Maximum parallelism QA pass
`visual-verdict`	`$visual-verdict`	UI/UX visual review with verdict
`frontend-ui-ux`	`$frontend-ui-ux`	Frontend UI/UX design and review
`ai-slop-cleaner`	`$ai-slop-cleaner`	Clean AI expression patterns from code/docs
`ask-claude`	`$ask-claude`	Cross-model consultation with Claude
`ask-gemini`	`$ask-gemini`	Cross-model consultation with Gemini
`git-master`	`$git-master`	Atomic commits, rebasing, history management
`ecomode`	`$ecomode`	Token-efficient execution mode
`web-clone`	`$web-clone`	Clone and adapt web pages or UI patterns
`cancel`	`$cancel`	Stop active execution modes

技能	触发词	描述
`autopilot`	`$autopilot`	全自动化流水线 — 从想法到可运行代码
`ralph`	`$ralph`	持久执行循环，直到验证完成
`ultrawork`	`$ulw`	跨并行Agent的最大并发调度
`team`	`$team`	N个协同Agent，支持tmux/worktree隔离
`plan`	`$plan`	开发前的战略规划
`ralplan`	`$ralplan`	基于共识的迭代规划
`deep-interview`	`$deep-interview`	规划或执行前的苏格拉底式澄清
`deepsearch`	`$deepsearch`	面向代码库的深度搜索
`analyze`	`$analyze`	基于证据的因果分析（跟踪方法论）
`tdd`	`$tdd`	测试驱动开发（红-绿-重构）
`build-fix`	`$build-fix`	修复构建错误、类型错误、工具链故障
`code-review`	`$code-review`	多维度综合代码评审
`security-review`	`$security-review`	安全审计 — 漏洞、信任边界检查
`ultraqa`	`$ultraqa`	最大并发QA测试
`visual-verdict`	`$visual-verdict`	带结论的UI/UX视觉评审
`frontend-ui-ux`	`$frontend-ui-ux`	前端UI/UX设计与评审
`ai-slop-cleaner`	`$ai-slop-cleaner`	清理代码/文档中的AI生成冗余表达
`ask-claude`	`$ask-claude`	调用Claude进行跨模型咨询
`ask-gemini`	`$ask-gemini`	调用Gemini进行跨模型咨询
`git-master`	`$git-master`	原子提交、变基、历史管理
`ecomode`	`$ecomode`	Token高效执行模式
`web-clone`	`$web-clone`	克隆与适配网页或UI模式
`cancel`	`$cancel`	停止正在运行的执行模式

4. Team Runtime

4. 团队运行时

bash

undefined

bash

undefined

Start parallel team workers

启动并行团队工作器

omx team 3:executor "fix all TypeScript errors" omx team 2:security-reviewer,1:architect "audit auth module"

Team management

团队管理

omx team status <team-name> omx team resume <team-name> omx team shutdown <team-name>


Inside Codex:
```text
$team 3:executor "parallelize this refactor"

Requirements:

Platform	Dependency
macOS	`brew install tmux`
Ubuntu/Debian	`sudo apt install tmux`
Fedora	`sudo dnf install tmux`
Windows (native)	`winget install psmux`
Windows (WSL2)	`sudo apt install tmux`

omx team status <team-name> omx team resume <team-name> omx team shutdown <team-name>


在Codex内使用：
```text
$team 3:executor "parallelize this refactor"

依赖要求：

平台	依赖
macOS	`brew install tmux`
Ubuntu/Debian	`sudo apt install tmux`
Fedora	`sudo dnf install tmux`
Windows (原生)	`winget install psmux`
Windows (WSL2)	`sudo apt install tmux`

5. CLI Commands

5. CLI命令

Core launch

核心启动命令

bash

omx                   # Launch Codex with OMX runtime
omx --madmax --high   # Recommended default (bypass + high reasoning)
omx --xhigh           # Extra-high reasoning effort

bash

omx                   # 搭载OMX运行时启动Codex
omx --madmax --high   # 推荐默认配置（绕过审批 + 高推理能力）
omx --xhigh           # 超高推理能力模式

Launch flags

启动参数

Flag	Description
`--yolo`	Fast execution, minimal verification
`--high`	High reasoning effort
`--xhigh`	Extra-high reasoning effort
`--madmax`	Bypass approvals/sandbox ( `--dangerously-bypass-approvals-and-sandbox` )
`--dry-run`	Preview without execution
`--verbose`	Verbose output

参数	描述
`--yolo`	快速执行，最小化验证
`--high`	高推理算力模式
`--xhigh`	超高推理算力模式
`--madmax`	绕过审批/沙箱（等价于 `--dangerously-bypass-approvals-and-sandbox` ）
`--dry-run`	仅预览不执行
`--verbose`	输出详细日志

Operator commands

操作命令

bash

omx setup             # Install prompts, skills, config, AGENTS scaffolding
omx doctor            # Verify installation
omx doctor --team     # Team/swarm diagnostics
omx status            # Show active modes
omx cancel            # Cancel execution modes
omx hud --watch       # Live HUD status monitor
omx reasoning <mode>  # Set: low/medium/high/xhigh
omx help              # Show help

bash

omx setup             # 安装提示词、技能、配置、Agent脚手架
omx doctor            # 验证安装是否正常
omx doctor --team     # 团队/集群诊断
omx status            # 展示当前激活的模式
omx cancel            # 取消执行模式
omx hud --watch       # 实时HUD状态监控
omx reasoning <mode>  # 设置推理等级：low/medium/high/xhigh
omx help              # 展示帮助信息

Explore & inspect

探索与检查命令

bash

omx explore --prompt "find where team state is written"   # Read-only repo lookup
omx sparkshell git status                                  # Bounded shell inspection
omx sparkshell --tmux-pane %12 --tail-lines 400           # Tail tmux pane output

bash

omx explore --prompt "find where team state is written"   # 只读仓库查询
omx sparkshell git status                                  # 受限shell检查
omx sparkshell --tmux-pane %12 --tail-lines 400           # 跟踪tmux面板输出

Instructions

使用说明

Identify intent — match request to the best workflow skill (
```
$plan
```
,
```
$ralph
```
,
```
$team
```
,
```
$deep-interview
```
)
Start with the recommended launch —
```
omx --madmax --high
```
for most sessions
Use role prompts for analysis —
```
/prompts:architect "..."
```
before implementation
Escalate when needed — let the agent pull in heavier workflows; don't force
```
$team
```
for simple tasks
Use
$deep-interview
for vague requests — when intent, scope, or boundaries are unclear
Point to references — link to the relevant references/ file for deep-dive details

明确需求 — 将请求匹配到最合适的工作流技能（
```
$plan
```
、
```
$ralph
```
、
```
$team
```
、
```
$deep-interview
```
）
使用推荐启动方式 — 大多数场景使用
```
omx --madmax --high
```
启动会话
使用角色提示词做分析 — 开发前先执行
```
/prompts:architect "..."
```
做架构分析
按需升级工作流 — 让Agent自动调用更重量级的工作流，简单任务不要强制使用
```
$team
```
模糊请求使用
$deep-interview
— 当需求、范围或边界不明确时使用
提供参考信息 — 附上相关参考资料/文件路径方便深度调研

Examples

示例

Example 1: Standard session

示例1：标准会话

bash

omx --madmax --high

text

/prompts:architect "analyze the current auth flow"
$plan "implement OAuth safely"

bash

omx --madmax --high

text

/prompts:architect "analyze the current auth flow"
$plan "implement OAuth safely"

Example 2: Persistent execution

示例2：持久执行

text

$ralph "fix all failing tests and make them pass"

Loop: execute → verify → fix, until all tests pass.

text

$ralph "fix all failing tests and make them pass"

循环执行：执行 → 验证 → 修复，直到所有测试通过。

Example 3: Parallel team work

示例3：并行团队协作

bash

omx team 3:executor "refactor the API layer with proper tests"

bash

omx team 3:executor "refactor the API layer with proper tests"

Example 4: Clarify before coding

示例4：编码前澄清需求

text

$deep-interview "I want to add a payment system"

One-question-at-a-time Socratic loop until intent is clear, then hands off to

$plan

$ralph

text

$deep-interview "I want to add a payment system"

采用单次提问的苏格拉底式循环直到需求明确，之后自动转交

$plan

或

$ralph

执行。

Best practices

最佳实践

Codex-first — OMX is a workflow layer, not a replacement for Codex; start with
```
omx --madmax --high
```
Use
$plan
before
$ralph
— plan first when scope is unclear; ralph for guaranteed execution
$ralplan
for consensus — use when multiple planning perspectives are needed before committing
$deep-interview
gates execution — always use for vague greenfield or brownfield work
Team mode is optional — only escalate to
```
$team
```
when the task genuinely needs parallel workers
Cross-model consultation —
```
$ask-claude
```
and
```
$ask-gemini
```
for second opinions on design or architecture
omx explore
for safe lookups — read-only; won't modify anything
omx doctor
when stuck — diagnose installation issues before manual debugging

优先使用Codex能力 — OMX是工作流层，不是Codex的替代品，优先使用
```
omx --madmax --high
```
启动
$ralph
执行前先调用
$plan
— 范围不明确时先做规划，ralph用于保障执行完成
需要共识规划时用
$ralplan
— 当提交前需要多个规划视角时使用
$deep-interview
作为执行前置检查 — 所有模糊的新开发或存量改造工作都应该先使用
团队模式按需使用 — 只有任务确实需要并行工作器时才升级到
```
$team
```
模式
跨模型咨询 — 设计或架构需要第二意见时使用
```
$ask-claude
```
和
```
$ask-gemini
```
安全查询使用
omx explore
— 只读模式，不会修改任何内容
遇到问题先执行
omx doctor
— 手动调试前先诊断安装问题

Quick Reference

快速参考

Command/Skill	Action
`omx --madmax --high`	Recommended launch
`/prompts:architect "..."`	Analysis and design review
`$plan "..."`	Strategic planning
`$deep-interview "..."`	Clarify vague requests
`$ralph "..."`	Persistent loop until done
`$ralplan "..."`	Consensus planning
`$team N:role "..."`	Parallel coordinated agents
`$ulw "..."`	Maximum parallelism
`omx explore --prompt "..."`	Read-only repo exploration
`omx sparkshell <cmd>`	Bounded shell inspection
`omx setup`	Install/re-install OMX
`omx doctor`	Diagnose issues
`omx cancel` / `$cancel`	Cancel active modes

→ Website · GitHub · npm

命令/技能	作用
`omx --madmax --high`	推荐启动方式
`/prompts:architect "..."`	分析与设计评审
`$plan "..."`	战略规划
`$deep-interview "..."`	澄清模糊需求
`$ralph "..."`	持久循环直到完成
`$ralplan "..."`	共识规划
`$team N:role "..."`	并行协同Agent
`$ulw "..."`	最大并发模式
`omx explore --prompt "..."`	只读仓库探索
`omx sparkshell <cmd>`	受限shell检查
`omx setup`	安装/重装OMX
`omx doctor`	问题诊断
`omx cancel` / `$cancel`	取消激活的模式

→ 官网 · GitHub · npm

References

参考资料

File	Contents
references/agents-catalog.md	All 30+ agent profiles and use cases
references/skills-reference.md	Full skill list with triggers and descriptions
references/cli-reference.md	Complete CLI: omx team/explore/sparkshell/reasoning

文件	内容
references/agents-catalog.md	全部30+个Agent的介绍与使用场景
references/skills-reference.md	完整技能列表，包含触发词与描述
references/cli-reference.md	完整CLI文档：omx team/explore/sparkshell/reasoning