Back to Details

skill-creator-pro

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Skill Creator Pro

Skill Creator Pro

Create production-grade skills that extend Claude's capabilities.
创建可扩展Claude能力的生产级Skill。

How This Skill Works

本Skill的工作原理

User: "Create a skill for X"
       ↓
Claude Code uses this meta-skill as guidance
       ↓
Follow Domain Discovery → Ask user clarifying questions → Create skill
       ↓
Generated skill with embedded domain expertise
This skill provides guidance and structure for creating skills. Claude Code:
  1. Uses this skill's framework to discover domain knowledge
  2. Asks user for clarifications about THEIR specific requirements
  3. Decides how to structure the generated skill based on domain needs
User: "Create a skill for X"
       ↓
Claude Code uses this meta-skill as guidance
       ↓
遵循领域探索→向用户询问澄清问题→创建Skill
       ↓
生成嵌入领域专业知识的Skill
本Skill为Skill创建提供指导与结构框架。Claude Code：
  1. 利用本Skill的框架探索领域知识
  2. 向用户询问其特定需求的澄清问题
  3. 根据领域需求决定生成Skill的结构

What This Skill Does

本Skill的功能

  • Guides creation of new skills from scratch
  • Helps improve existing skills to production quality
  • Provides patterns for 5 skill types (Builder, Guide, Automation, Analyzer, Validator)
  • Ensures skills encode procedural knowledge + domain expertise
  • 指导从零开始创建新Skill
  • 帮助将现有Skill优化至生产级质量
  • 提供5种Skill类型的模式（Builder、Guide、Automation、Analyzer、Validator）
  • 确保Skill整合过程性知识与领域专业知识

What This Skill Does NOT Do

本Skill不具备的功能

  • Test skills in production environments
  • Deploy or distribute skills
  • Handle skill versioning/updates after creation
  • Create requirement-specific skills (always create reusable intelligence)
  • 在生产环境中测试Skill
  • 部署或分发Skill
  • 处理Skill创建后的版本控制/更新
  • 创建特定需求专属Skill（始终创建可复用智能）

Domain Discovery Framework

领域探索框架

Key Principle: Users want domain expertise IN the skill. They may not BE domain experts.
核心原则：用户希望Skill中内置领域专业知识，而他们自己可能并非领域专家。

Phase 1: Automatic Discovery (No User Input)

阶段1：自动探索（无需用户输入）

Proactively research the domain before asking anything:
DiscoverHowExample: "Kafka integration"
Core conceptsOfficial docs, Context7Producers, consumers, topics, partitions
Standards/complianceSearch "[domain] standards"Kafka security, exactly-once semantics
Best practicesSearch "[domain] best practices 2025"Partitioning strategies, consumer groups
Anti-patternsSearch "[domain] common mistakes"Too many partitions, no monitoring
SecuritySearch "[domain] security"SASL, SSL, ACLs, encryption
EcosystemSearch "[domain] ecosystem tools"Confluent, Schema Registry, Connect
Sources priority: Official docs → Library docs (Context7) → GitHub → Community → WebSearch
在询问用户任何问题前，主动研究领域信息：
探索内容探索方式示例："Kafka integration"
核心概念官方文档、Context7Producers, consumers, topics, partitions
标准/合规性搜索"[领域] standards"Kafka security, exactly-once semantics
最佳实践搜索"[领域] best practices 2025"Partitioning strategies, consumer groups
反模式搜索"[领域] common mistakes"Too many partitions, no monitoring
安全相关搜索"[领域] security"SASL, SSL, ACLs, encryption
生态系统搜索"[领域] ecosystem tools"Confluent, Schema Registry, Connect
来源优先级：官方文档 → 库文档（Context7）→ GitHub → 社区 → 网络搜索

Phase 2: Knowledge Sufficiency Check

阶段2：知识充足性检查

Before asking user anything, verify internally:
- [ ] Core concepts understood?
- [ ] Best practices identified?
- [ ] Anti-patterns known?
- [ ] Security considerations covered?
- [ ] Official sources found?

If ANY gap → Research more (don't ask user for domain knowledge)
Only if CANNOT discover (proprietary/internal) → Ask user
在询问用户任何问题前，先内部验证：
- [ ] 核心概念已理解？
- [ ] 已识别最佳实践？
- [ ] 已知晓反模式？
- [ ] 已覆盖安全考量？
- [ ] 已找到官方来源？

若存在任何缺口→进一步研究（不要向用户询问领域知识）
仅当无法探索到（专有/内部信息）→再询问用户

Phase 3: User Requirements (NOT Domain Knowledge)

阶段3：用户需求（非领域知识）

Only ask about user's SPECIFIC context:
AskDon't Ask
"What's YOUR use case?""What is Kafka?"
"What's YOUR tech stack?""What options exist?"
"Any existing resources?""How does it work?"
"Specific constraints?""What are best practices?"
The skill contains domain expertise. User provides requirements.
仅询问用户的特定上下文信息：
可询问不可询问
"你的具体使用场景是什么？""What is Kafka?"
"你的技术栈是什么？""What options exist?"
"是否有现有资源可以参考？""How does it work?"
"是否有特定约束条件？""What are best practices?"
Skill中已包含领域专业知识，用户只需提供需求信息。

Required Clarifications

必要的澄清问题

Ask about SKILL METADATA and USER REQUIREMENTS (not domain knowledge):
询问关于Skill元数据和用户需求的问题（而非领域知识）：

Skill Metadata

Skill元数据

1. Skill Type - "What type of skill?"
TypePurposeExample
BuilderCreate artifactsWidgets, code, documents
GuideProvide instructionsHow-to, tutorials
AutomationExecute workflowsFile processing, deployments
AnalyzerExtract insightsCode review, data analysis
ValidatorEnforce qualityCompliance checks, scoring
2. Domain - "What domain or technology?"
1. Skill类型 - "需要创建哪种类型的Skill？"
类型用途示例
Builder创建制品组件、代码、文档
Guide提供指导说明操作指南、教程
Automation执行工作流文件处理、部署流程
Analyzer提取洞察信息代码审查、数据分析
Validator执行质量校验合规检查、评分
2. 领域 - "针对哪个领域或技术？"

User Requirements (After Domain Discovery)

用户需求（领域探索完成后）

3. Use Case - "What's YOUR specific use case?"
  • Not "what can it do" but "what do YOU need"
4. Tech Stack - "What's YOUR environment?"
  • Languages, frameworks, existing infrastructure
5. Existing Resources - "Any scripts, templates, configs to include?"
6. Constraints - "Any specific requirements or limitations?"
  • Performance, security, compliance specific to user's context
3. 使用场景 - "你的具体使用场景是什么？"
  • 不是询问“它能做什么”，而是“你需要它做什么”
4. 技术栈 - "你的运行环境是什么？"
  • 编程语言、框架、现有基础设施
5. 现有资源 - "是否有需要纳入的脚本、模板、配置文件？"
6. 约束条件 - "是否有特定需求或限制？"
  • 与用户上下文相关的性能、安全、合规要求

Note

注意事项

  • Questions 1-2: Ask immediately
  • Domain Discovery: Research automatically after knowing domain
  • Questions 3-6: Ask after discovery, informed by domain knowledge
  • Question pacing: Avoid asking too many questions in a single message. Start with most important, follow up as needed.
  • 问题1-2：立即询问
  • 领域探索：确定领域后自动开展研究
  • 问题3-6：探索完成后，结合领域知识进行询问
  • 问题节奏：避免在单条消息中询问过多问题。从最重要的问题开始，按需跟进后续问题。

Core Principles

核心原则

Reusable Intelligence, Not Requirement-Specific

可复用智能，而非特定需求专属

Skills must handle VARIATIONS, not single requirements:
❌ Bad: "Create bar chart with sales data using Recharts"
✅ Good: "Create visualizations - adaptable to data shape, chart type, library"

❌ Bad: "Deploy to AWS EKS with Helm"
✅ Good: "Deploy applications - adaptable to platform, orchestration, environment"
Identify what VARIES vs what's CONSTANT in the domain. See 
references/reusability-patterns.md
.
Skill必须能处理多种变化场景，而非仅满足单一需求：
❌ 不佳示例："Create bar chart with sales data using Recharts"
✅ 优质示例："Create visualizations - adaptable to data shape, chart type, library"

❌ 不佳示例："Deploy to AWS EKS with Helm"
✅ 优质示例："Deploy applications - adaptable to platform, orchestration, environment"
识别领域中可变因素与恒定因素。详见
references/reusability-patterns.md
。

Concise is Key

简洁为要

Context window is a public good (~1,500+ tokens per skill activation). Challenge each piece:
  • "Does Claude really need this explanation?"
  • "Does this paragraph justify its token cost?"
Prefer concise examples over verbose explanations.
上下文窗口是公共资源（每次Skill激活约1500+ tokens）。对每一部分内容都要审慎考量：
  • “Claude真的需要这段解释吗？”
  • “这段内容的token消耗是否合理？”
优先使用简洁示例，而非冗长说明。

Appropriate Freedom

适度灵活性

Match specificity to task fragility:
Freedom LevelWhen to UseExample
HighMultiple approaches valid"Choose your preferred style"
MediumPreferred pattern existsPseudocode with parameters
LowOperations are fragileExact scripts, few parameters
根据任务的脆弱性匹配具体程度：
灵活度等级适用场景示例
高存在多种有效实现方式"Choose your preferred style"
中存在首选模式带参数的伪代码
低操作容错性低精确脚本，参数极少

Progressive Disclosure

渐进式披露

Three-level loading system:
  1. Metadata (~100 tokens) - Always in context (description ≤1024 chars)
  2. SKILL.md body (<500 lines) - When skill triggers
  3. References (unlimited) - Loaded as needed by Claude
三级加载体系：
  1. 元数据（约100 tokens）- 始终处于上下文中（描述≤1024字符）
  2. SKILL.md 主体（<500行）- Skill触发时加载
  3. 参考资料（无限制）- Claude按需加载

Anatomy of a Skill

Skill的结构组成

Generated skills are zero-shot domain experts with embedded knowledge.
skill-name/
├── SKILL.md (required)
│   ├── YAML frontmatter (name, description, allowed-tools?, model?)
│   └── Procedural knowledge (workflows, steps, decision trees)
└── Bundled Resources
    ├── references/   - Domain expertise (structure based on domain needs)
    ├── scripts/      - Executable code (tested, reliable)
    └── assets/       - Templates, boilerplate, images
生成的Skill是内置领域知识的零样本领域专家。
skill-name/
├── SKILL.md (required)
│   ├── YAML frontmatter (name, description, allowed-tools?, model?)
│   └── 过程性知识（工作流、步骤、决策树）
└── 捆绑资源
    ├── references/   - 领域专业知识（根据领域需求组织结构）
    ├── scripts/      - 可执行代码（已测试、可靠）
    └── assets/       - 模板、样板代码、图片

SKILL.md Requirements

SKILL.md 要求

ComponentRequirement
Line count<500 lines (extract to references/)
FrontmatterSee 
references/skill-patterns.md
for complete spec
name
Lowercase, numbers, hyphens; ≤64 chars; match directory
description
[What] + [When]; ≤1024 chars; third-person style
Description style"This skill should be used when..." (not "Use when...")
FormImperative ("Do X" not "You should X")
ScopeWhat it does AND does not do
组件要求
行数<500行（超出部分提取至references/）
前置元数据完整规范请参考
references/skill-patterns.md
name
小写、数字、连字符；≤64字符；与目录名一致
description
[功能] + [触发场景]；≤1024字符；第三人称表述
描述风格"This skill should be used when..."（而非"Use when..."）
表述形式祈使句（"执行X操作"而非"你应该执行X操作"）
范围明确说明功能与非功能

What Goes in references/

references/ 目录内容

Embed domain knowledge gathered during discovery:
Gathered KnowledgePurpose in Skill
Library/API documentationEnable correct implementation
Best practicesGuide quality decisions
Code examplesProvide reference patterns
Anti-patternsPrevent common mistakes
Domain-specific detailsSupport edge cases
Structure references/ based on what the domain needs.
Large files: If references >10k words, include grep search patterns in SKILL.md for efficient discovery.
将探索阶段收集的领域知识嵌入其中：
收集的知识类型在Skill中的作用
库/API文档确保实现的正确性
最佳实践指导质量决策
代码示例提供参考模式
反模式避免常见错误
领域专属细节支持边缘场景处理
根据领域需求组织references/的结构。
大文件处理：若参考资料超过10000词，在SKILL.md中包含grep搜索模式以提升检索效率。

When to Generate scripts/

何时生成scripts/目录

Generate scripts when domain requires deterministic, executable procedures:
Domain NeedExample Scripts
Setup/installationInstall dependencies, initialize project
ProcessingTransform data, process files
ValidationCheck compliance, verify output
DeploymentDeploy services, configure infrastructure
Decision: If procedure is complex, error-prone, or needs to be exactly repeatable → create script. Otherwise → document in SKILL.md or references/.
当领域需要确定性、可执行的流程时，生成脚本：
领域需求示例脚本
搭建/安装安装依赖、初始化项目
处理流程数据转换、文件处理
校验合规检查、输出验证
部署服务部署、基础设施配置
决策依据：若流程复杂、易出错或需要完全可重复执行→创建脚本。否则→在SKILL.md或references/中记录。

When to Generate assets/

何时生成assets/目录

Generate assets when domain requires exact templates or boilerplate:
Domain NeedExample Assets
Starting templatesHTML boilerplate, component scaffolds
Configuration filesConfig templates, schema definitions
Code boilerplateBase classes, starter code
当领域需要精确模板或样板代码时，生成资源文件：
领域需求示例资源
初始模板HTML样板、组件脚手架
配置文件配置模板、schema定义
代码样板基类、起始代码

What NOT to Include

不应包含的内容

  • README.md (SKILL.md IS the readme)
  • CHANGELOG.md
  • LICENSE (inherited from repo)
  • Duplicate information
  • README.md（SKILL.md 即作为说明文档）
  • CHANGELOG.md
  • LICENSE（继承自仓库许可证）
  • 重复信息

What Generated Skill Does at Runtime

生成的Skill在运行时的工作流程

User invokes skill → Gather context from:
  1. Codebase (if existing project)
  2. Conversation (user's requirements)
  3. Own references/ (embedded domain expertise)
  4. User-specific guidelines
→ Ensure all information gathered → Implement ZERO-SHOT
用户调用Skill → 从以下来源收集上下文：
  1. 代码库（若为现有项目）
  2. 对话内容（用户需求）
  3. 自身references/（内置领域专业知识）
  4. 用户特定指南
→ 确保收集到所有信息→实现零样本执行

Include in Generated Skills

生成Skill中必须包含的内容

Every generated skill should include:
markdown
undefined
每个生成的Skill都应包含以下内容：
markdown
undefined

Before Implementation

实现前准备

Gather context to ensure successful implementation:
SourceGather
CodebaseExisting structure, patterns, conventions to integrate with
ConversationUser's specific requirements, constraints, preferences
Skill ReferencesDomain patterns from 
references/
(library docs, best practices, examples)
User GuidelinesProject-specific conventions, team standards
Ensure all required context is gathered before implementing. Only ask user for THEIR specific requirements (domain expertise is in this skill).

---
收集上下文信息以确保成功实现：
来源收集内容
代码库现有结构、模式、约定，确保Skill可与之集成
对话内容用户的特定需求、约束条件、偏好
Skill参考资料来自
references/
的领域模式（库文档、最佳实践、示例）
用户指南项目特定约定、团队标准
确保在实现前收集到所有必要的上下文信息。 仅向用户询问其特定需求（领域专业知识已内置在本Skill中）。

---

Type-Aware Creation

基于类型的创建流程

After determining skill type, follow type-specific patterns:
TypeKey SectionsReference
BuilderClarifications → Output Spec → Standards → Checklist
skill-patterns.md#builder
GuideWorkflow → Examples → Official Docs
skill-patterns.md#guide
AutomationScripts → Dependencies → Error Handling
skill-patterns.md#automation
AnalyzerScope → Criteria → Output Format
skill-patterns.md#analyzer
ValidatorCriteria → Scoring → Thresholds → Remediation
skill-patterns.md#validator
确定Skill类型后，遵循对应类型的模式：
类型核心章节参考文档
Builder澄清问题→输出规范→标准→检查清单
skill-patterns.md#builder
Guide工作流→示例→官方文档
skill-patterns.md#guide
Automation脚本→依赖→错误处理
skill-patterns.md#automation
Analyzer范围→标准→输出格式
skill-patterns.md#analyzer
Validator标准→评分→阈值→修复建议
skill-patterns.md#validator

Skill Creation Process

Skill创建流程

Metadata → Discovery → Requirements → Analyze → Embed → Structure → Implement → Validate
See 
references/creation-workflow.md
for detailed steps.
元数据→探索→需求→分析→嵌入→结构→实现→验证
详细步骤请参考
references/creation-workflow.md
。

Quick Steps

快速步骤

  1. Metadata: Ask skill type + domain (Questions 1-2)
  2. Discovery: Research domain automatically (Phase 1-2 above)
  3. Requirements: Ask user's specific needs (Questions 3-6)
  4. Analyze: Identify procedural (HOW) + domain (WHAT) knowledge
  5. Embed: Put gathered domain expertise into 
    references/
  6. Structure: Initialize skill directory
  7. Implement: Write SKILL.md + resources following type patterns
  8. Validate: Run 
    scripts/package_skill.py
    and test
  1. 元数据：询问Skill类型与领域（问题1-2）
  2. 探索：自动开展领域研究（上述阶段1-2）
  3. 需求：询问用户的特定需求（问题3-6）
  4. 分析：识别过程性（如何做）与领域性（做什么）知识
  5. 嵌入：将收集到的领域专业知识存入
    references/
  6. 结构：初始化Skill目录
  7. 实现：遵循类型模式编写SKILL.md及相关资源
  8. 验证：运行
    scripts/package_skill.py
    并进行测试

SKILL.md Template

SKILL.md 模板

yaml
---
name: skill-name                    # lowercase, hyphens, ≤64 chars
description: |                      # ≤1024 chars
  [What] Capability statement.
  [When] Use when users ask to <triggers>.
allowed-tools: Read, Grep, Glob     # optional: restrict tools
---
See 
references/skill-patterns.md
for complete frontmatter spec and body patterns.
yaml
---
name: skill-name                    # 小写、连字符、≤64字符
description: |                      # ≤1024字符
  [功能] 能力说明。
  [触发场景] 当用户请求<triggers>时使用。
allowed-tools: Read, Grep, Glob     # 可选：限制可使用的工具
---
完整的前置元数据规范与主体模式请参考
references/skill-patterns.md
。

Output Checklist

输出检查清单

Before delivering a skill, verify:
交付Skill前，需验证以下内容：

Domain Discovery Complete

领域探索完成

  • Core concepts discovered and understood
  • Best practices identified from authentic sources
  • Anti-patterns documented
  • Security considerations covered
  • Official documentation linked
  • User was NOT asked for domain knowledge
  • 已探索并理解核心概念
  • 已从可信来源识别最佳实践
  • 已记录反模式
  • 已覆盖安全考量
  • 已链接官方文档
  • 未向用户询问领域知识

Frontmatter

前置元数据

  • name
    : lowercase, hyphens, ≤64 chars, matches directory
  • description
    : [What]+[When], ≤1024 chars, clear triggers
  • allowed-tools
    : Set if restricted access needed
  • name
    ：小写、连字符、≤64字符，与目录名一致
  • description
    ：包含[功能]+[触发场景]，≤1024字符，触发场景明确
  • allowed-tools
    ：若需限制访问则进行设置

Structure

结构

  • SKILL.md <500 lines
  • Progressive disclosure (details in references/)
  • SKILL.md 行数<500
  • 采用渐进式披露（细节内容存入references/）

Knowledge Coverage

知识覆盖

  • Procedural (HOW): Workflows, decision trees, error handling
  • Domain (WHAT): Concepts, best practices, anti-patterns
  • 过程性（如何做）：工作流、决策树、错误处理
  • 领域性（做什么）：概念、最佳实践、反模式

Zero-Shot Implementation (in generated skill)

零样本实现（在生成的Skill中）

  • Includes "Before Implementation" section
  • Gathers runtime context (codebase, conversation, user guidelines)
  • Domain expertise embedded in 
    references/
    (structured per domain needs)
  • Only asks user for THEIR requirements (not domain knowledge)
  • 包含“实现前准备”章节
  • 会收集运行时上下文（代码库、对话内容、用户指南）
  • 领域专业知识已嵌入
    references/
    （根据领域需求组织结构）
  • 仅向用户询问其特定需求（而非领域知识）

Reusability

可复用性

  • Handles variations (not requirement-specific)
  • Clarifications capture variable elements (user's context)
  • Constants encoded (domain patterns, best practices)
  • 可处理多种变化场景（非特定需求专属）
  • 澄清问题可捕获可变要素（用户上下文）
  • 恒定要素已编码（领域模式、最佳实践）

Type-Specific (see 
references/skill-patterns.md
)

类型特定要求（参考
references/skill-patterns.md
）

  • Builder: Clarifications, output spec, standards, checklist
  • Guide: Workflow, examples, official docs
  • Automation: Scripts, dependencies, error handling
  • Analyzer: Scope, criteria, output format
  • Validator: Criteria, scoring, thresholds, remediation
  • Builder：包含澄清问题、输出规范、标准、检查清单
  • Guide：包含工作流、示例、官方文档
  • Automation：包含脚本、依赖、错误处理
  • Analyzer：包含范围、标准、输出格式
  • Validator：包含标准、评分、阈值、修复建议

Reference Files

参考文件

FileWhen to Read
references/creation-workflow.md
Detailed step-by-step creation process
references/skill-patterns.md
Frontmatter spec, type-specific patterns, assets guidance
references/reusability-patterns.md
Procedural+domain knowledge, varies vs constant
references/quality-patterns.md
Clarifications, enforcement, checklists
references/technical-patterns.md
Error handling, security, dependencies
references/workflows.md
Sequential and conditional workflow patterns
references/output-patterns.md
Template and example patterns
文件阅读时机
references/creation-workflow.md
查看详细的分步创建流程时
references/skill-patterns.md
查看前置元数据规范、类型特定模式、资源指导时
references/reusability-patterns.md
了解过程性+领域性知识、可变与恒定要素时
references/quality-patterns.md
查看澄清问题、执行规范、检查清单时
references/technical-patterns.md
了解错误处理、安全、依赖相关内容时
references/workflows.md
查看顺序与条件工作流模式时
references/output-patterns.md
查看模板与示例模式时