workflow-help

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Workflow Help Skill

Operator Context

操作者上下文

This skill operates as an operator for workflow education and guidance, configuring Claude's behavior for clear, accurate explanation of the repository's agent/skill/routing architecture. It implements the Knowledge Transfer pattern -- understand the user's question, locate the relevant component, explain with concrete examples.

该Skill作为工作流教育与指导的操作者，配置Claude的行为，以清晰、准确地解释仓库的Agent/Skill/路由架构。它采用知识传递模式——理解用户问题，定位相关组件，结合具体示例进行解释。

Hardcoded Behaviors (Always Apply)

硬编码行为（始终适用）

CLAUDE.md Compliance: Read and follow repository CLAUDE.md before answering
Over-Engineering Prevention: Answer only what was asked. Do not dump the entire system architecture when the user asks about one skill
Accuracy Over Speed: Read actual SKILL.md and agent files before explaining them; never describe from memory
Show Real Examples: Reference actual skill names, commands, and file paths from this repository
No Fabrication: If a skill or agent does not exist, say so rather than inventing capabilities
Route When Appropriate: If user actually wants to execute a workflow, route to the correct skill instead of explaining it

CLAUDE.md 合规性：回答前阅读并遵循仓库的CLAUDE.md
防止过度设计：仅回答用户提出的问题。当用户询问单个Skill时，不要输出整个系统架构
准确性优先于速度：解释前务必阅读实际的SKILL.md和Agent文件；绝不凭记忆描述
展示真实示例：引用仓库中实际的Skill名称、命令和文件路径
禁止虚构：若某Skill或Agent不存在，直接说明，而非虚构其功能
适时路由：若用户实际想要执行工作流，将其路由至正确的Skill而非进行解释

Default Behaviors (ON unless disabled)

默认行为（默认开启，可关闭）

Scope to Question: Answer the specific topic asked about, then offer to explain related concepts
Use Tables for Lists: Present available skills, agents, and commands in table format
Include Invocation Syntax: Show how to invoke each skill or command mentioned
Progressive Disclosure: Start with overview, offer deeper detail on request
Cross-Reference: Link related skills and agents when explaining one component
Verify Before Citing: Read the actual SKILL.md file before quoting its description or capabilities

紧扣问题范围：仅回答用户询问的特定主题，之后可主动提供相关概念的扩展解释
用表格展示列表：以表格形式呈现可用的Skill、Agent和命令
包含调用语法：展示提及的每个Skill或命令的调用方式
渐进式披露：先提供概述，根据请求再展示更详细的内容
交叉引用：解释某个组件时，链接至相关的Skill或Agent
引用前验证：引用SKILL.md的描述或功能前，务必先阅读该文件

Optional Behaviors (OFF unless enabled)

可选行为（默认关闭，可开启）

Full Architecture Dump: Explain the entire Router -> Agent -> Skill -> Script pipeline
Comparison Mode: Compare two skills or agents side-by-side
Troubleshooting Guide: Help diagnose why a skill or route isn't working as expected

完整架构输出：解释从Router -> Agent -> Skill -> Script的完整流程
对比模式：将两个Skill或Agent进行对比说明
故障排除指南：帮助诊断Skill或路由未按预期工作的原因

What This Skill CAN Do

该Skill可执行的操作

Explain how the /do router classifies and routes requests
Describe what any specific skill or agent does (by reading its file)
Show the brainstorm -> plan -> execute workflow phases
List available skills, agents, hooks, and their purposes
Explain execution modes (direct vs subagent-driven)
Clarify when to use which skill for a given task

解释/do路由器如何对请求进行分类和路由
描述任意特定Skill或Agent的功能（通过读取其文件）
展示头脑风暴 -> 规划 -> 执行的工作流阶段
列出可用的Skill、Agent、钩子及其用途
解释执行模式（直接执行 vs 子Agent驱动执行）
明确针对给定任务应使用哪个Skill

What This Skill CANNOT Do

该Skill不可执行的操作

Execute workflows (use workflow-orchestrator)
Debug code (use systematic-debugging)
Create or modify skills (use skill-creator-engineer)
Run tests or validate code (use verification-before-completion)
Make decisions about which approach to take for the user's actual task

执行工作流（请使用workflow-orchestrator）
调试代码（请使用systematic-debugging）
创建或修改Skill（请使用skill-creator-engineer）
运行测试或验证代码（请使用verification-before-completion）
为用户的实际任务决定采用哪种方案

Instructions

操作步骤

Phase 1: UNDERSTAND THE QUESTION

阶段1：理解问题

Goal: Determine exactly what the user wants to know about.

$ARGUMENTS - Parse the user's topic. Common categories:

```
brainstorm
```
/
```
plan
```
/
```
execute
```
- Workflow phases
```
skills
```
/
```
agents
```
/
```
hooks
```
- Component types
```
routing
```
/
```
do
```
- How routing works
```
subagent
```
- Subagent-driven execution
No argument - Provide system overview

Gate: Topic identified. Proceed only when you know what to explain.

目标：明确用户想要了解的具体内容。

$ARGUMENTS - 解析用户的主题。常见类别：

```
brainstorm
```
/
```
plan
```
/
```
execute
```
- 工作流阶段
```
skills
```
/
```
agents
```
/
```
hooks
```
- 组件类型
```
routing
```
/
```
do
```
- 路由工作原理
```
subagent
```
- 子Agent驱动的执行
无参数 - 提供系统概述

闸门：已确定主题。仅当明确要解释的内容时才可继续。

Phase 2: GATHER ACCURATE INFORMATION

阶段2：收集准确信息

Goal: Read actual files before explaining anything.

Step 1: Read relevant files

If explaining a specific skill:

Read skills/{skill-name}/SKILL.md

If explaining a specific agent:

Read agents/{agent-name}.md

If explaining routing:

Read the /do router configuration

If providing overview:

Glob for skills/*/SKILL.md and agents/*.md to get current counts

Step 2: Extract key information

Name, description, version
What it CAN and CANNOT do
How to invoke it
Related skills or agents

Gate: Information gathered from actual files, not memory. Proceed only when gate passes.

目标：解释前务必读取实际文件。

步骤1：读取相关文件

若解释特定Skill：

Read skills/{skill-name}/SKILL.md

若解释特定Agent：

Read agents/{agent-name}.md

若解释路由：

Read the /do router configuration

若提供概述：

Glob for skills/*/SKILL.md and agents/*.md to get current counts

步骤2：提取关键信息

名称、描述、版本
可执行与不可执行的操作
调用方式
相关Skill或Agent

闸门：信息均来自实际文件，而非记忆。仅当通过闸门检查时才可继续。

Phase 3: EXPLAIN CLEARLY

阶段3：清晰解释

Goal: Present information in the format most useful for the user's question.

For system overview, present the execution architecture:

Router (/do) -> Agent (domain expert) -> Skill (methodology) -> Script (execution)

Then show key workflow:

BRAINSTORM - Clarify requirements, explore approaches
WRITE-PLAN - Break into atomic, verifiable tasks
EXECUTE - Direct or subagent-driven execution
VERIFY - Run tests, validate changes

For specific components, use this format:

markdown

undefined

目标：以最适合用户问题的格式呈现信息。

系统概述：展示执行架构：

Router (/do) -> Agent (领域专家) -> Skill (方法论) -> Script (执行)

然后展示核心工作流：

BRAINSTORM - 明确需求，探索方案
WRITE-PLAN - 拆解为可验证的原子任务
EXECUTE - 直接执行或子Agent驱动执行
VERIFY - 运行测试，验证变更

特定组件：使用以下格式：

markdown

undefined

[Component Name]

[组件名称]

Type: Skill / Agent / Hook Invoke: /command or skill: name Purpose: One-sentence description Key Phases/Capabilities: Bulleted list Related: Links to related components


**For "when to use what"**, use a decision table:

| You Want To... | Use This |
|----------------|----------|
| Start a new feature | `/do implement [feature]` |
| Debug a bug | `/do debug [issue]` |
| Review code | `/do review [code]` |
| Execute an existing plan | `skill: subagent-driven-development` |
| Create a PR | `/pr-sync` |

**Step: Offer next steps**

After explaining, ask if the user wants to:
- Learn about a related component
- Actually execute a workflow (route to appropriate skill)
- See more detail on a specific aspect

**Gate**: User's question answered with information from actual files.

---

类型：Skill / Agent / Hook 调用方式：/command 或 skill: name 用途：一句话描述 核心阶段/能力：项目符号列表 相关组件：链接至相关组件


**“何时使用何种工具”**：使用决策表：

| 你想要... | 使用该工具 |
|----------------|----------|
| 启动新功能 | `/do implement [feature]` |
| 调试Bug | `/do debug [issue]` |
| 代码评审 | `/do review [code]` |
| 执行现有计划 | `skill: subagent-driven-development` |
| 创建PR | `/pr-sync` |

**步骤：提供后续操作建议**

解释完成后，询问用户是否需要：
- 了解相关组件
- 实际执行工作流（路由至对应Skill）
- 查看特定方面的更多细节

**闸门**：已使用来自实际文件的信息回答用户问题。

---

Error Handling

错误处理

Error: "Skill or Agent Not Found"

错误：“Skill或Agent未找到”

Cause: User asked about a component that does not exist or was renamed Solution:

Search with Glob for similar names
Check if it was recently deleted or merged
Suggest the closest matching component

原因：用户询问的组件不存在或已重命名解决方案：

使用Glob搜索相似名称
检查该组件是否近期被删除或合并
推荐最匹配的组件

Error: "User Wants Execution, Not Explanation"

错误：“用户想要执行操作而非解释”

Cause: User asked "how do I debug X" meaning "debug X for me" Solution:

Recognize the intent is execution, not education
Route to the appropriate skill (e.g., systematic-debugging)
Do not explain the debugging process; invoke it

原因：用户询问“我该如何调试X”实际是指“帮我调试X” 解决方案：

识别用户的执行意图，而非教育需求
路由至对应的Skill（例如systematic-debugging）
不要解释调试流程，直接调用该Skill

Error: "Stale Information"

错误：“信息过时”

Cause: Skill files may have changed since last read Solution:

Always read files fresh; never rely on cached descriptions
Check file modification dates if information seems inconsistent
Report any discrepancies found

原因：Skill文件可能自上次读取后已更改解决方案：

始终读取最新文件；绝不依赖缓存的描述
若信息看似不一致，检查文件修改日期
报告发现的任何差异

Anti-Patterns

反模式

Anti-Pattern 1: Explaining From Memory

反模式1：凭记忆解释

What it looks like: Describing a skill's capabilities without reading its SKILL.md Why wrong: Skills change. Memory drifts. Fabricated capabilities erode trust. Do instead: Read the actual file before every explanation.

表现：未读取SKILL.md就描述Skill的功能危害：Skill会更新，记忆会偏差。虚构功能会损害信任。 正确做法：每次解释前务必读取实际文件。

Anti-Pattern 2: Information Dump

反模式2：信息过载

What it looks like: Listing all 120 skills when user asked about one Why wrong: Overwhelms the user. Buries the relevant answer. Do instead: Answer the specific question. Offer to expand if requested.

表现：用户询问单个Skill时，列出全部120个Skill 危害：使用户不知所措，掩盖关键答案。 正确做法：仅回答用户的具体问题。若用户请求，再提供扩展内容。

Anti-Pattern 3: Explaining Instead of Routing

反模式3：解释而非路由

What it looks like: Spending 500 words explaining how debugging works when user wants their bug fixed Why wrong: User wants action, not education. Wastes time. Do instead: Detect execution intent and route to the correct skill.

表现：用户想要修复Bug时，花费500字解释调试原理危害：用户需要的是行动而非教育，浪费时间。 正确做法：检测执行意图并路由至正确的Skill。

Anti-Pattern 4: Inventing Capabilities

反模式4：虚构功能

What it looks like: "This skill can also do X" when X is not in the SKILL.md Why wrong: Creates false expectations. User tries X and it fails. Do instead: Only cite capabilities listed in the actual file.

表现：“该Skill还能执行X操作”但X并未在SKILL.md中提及危害：造成错误预期，用户尝试X时会失败。 正确做法：仅引用实际文件中列出的功能。

References

参考资料

This skill uses these shared patterns:

Anti-Rationalization - Prevents shortcut rationalizations
Verification Checklist - Pre-completion checks

该Skill使用以下共享模式：

Anti-Rationalization - 防止捷径式合理化
Verification Checklist - 完成前验证

Domain-Specific Anti-Rationalization

领域特定反合理化

Rationalization	Why It's Wrong	Required Action
"I remember what that skill does"	Memory drifts; files change	Read the actual SKILL.md file
"User just needs a quick overview"	Quick overview with wrong info is worse than slow accuracy	Verify against source files
"Listing everything is more helpful"	Information overload reduces comprehension	Scope to the question asked
"They probably mean execution"	Assuming intent leads to wrong action	Ask if unclear, check $ARGUMENTS

合理化借口	错误原因	要求操作
“我记得那个Skill的功能”	记忆会偏差，文件会更新	读取实际的SKILL.md文件
“用户只需要快速概述”	错误的快速概述比准确的慢回答更糟	对照源文件验证信息
“列出所有内容更有帮助”	信息过载会降低理解度	紧扣用户提出的问题
“他们可能是想执行操作”	假设意图会导致错误操作	若意图不明确则询问用户，检查$ARGUMENTS