skill-creator

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Skill Creator

Create, evaluate, and iterate on high-quality agent skills. This skill guides the entire lifecycle: planning what the skill should do, writing SKILL.md and reference files, scoring quality against a rubric, and iterating until the skill meets production standards.

Philosophy: A great skill is not a long skill. It is a precise skill: exhaustive triggers, explicit defaults, clear steps with exit gates, deferred complexity via reference files, and a structured output template.

Core rule — always dynamic, never static: Skills MUST detect what tools, libraries, and auth are available at runtime and adapt their behavior accordingly. Never hardcode a single method. Always provide a detection flow with a decision tree and fallback paths. See

references/dynamic-calling.md

for the complete pattern catalog.

创建、评估并迭代高质量的Agent Skill。本Skill指导完整生命周期：规划Skill应实现的功能、编写SKILL.md和参考文件、根据评分标准评估质量，直至Skill达到生产标准。

理念：优秀的Skill不在于篇幅长，而在于精准：涵盖全面的触发词、明确的默认值、带退出节点的清晰步骤、通过参考文件延迟复杂内容，以及结构化的输出模板。

核心规则——始终动态，绝不静态：Skill必须在运行时检测可用的工具、库和授权信息，并相应调整行为。绝不要硬编码单一方法。始终提供带有决策树和备选路径的检测流程。完整的模式目录请参见

references/dynamic-calling.md

。

Step 1: Understand What the User Wants

步骤1：理解用户需求

Classify the request into one of these modes:

User Intent	Mode	Jump To
Create a brand-new skill	Create	Step 2
Improve / fix an existing skill	Improve	Step 6
Evaluate / score a skill's quality	Evaluate	Step 7

If ambiguous, ask: "Do you want to create a new skill, improve an existing one, or evaluate one?"

将请求归类为以下模式之一：

用户意图	模式	跳转至
创建全新Skill	Create	步骤2
改进/修复现有Skill	Improve	步骤6
评估/评分Skill质量	Evaluate	步骤7

若意图不明确，请询问："你想要创建新Skill、改进现有Skill还是评估某个Skill？"

Gather Requirements (for Create mode)

收集需求（Create模式）

Before writing anything, answer these questions (ask the user if unclear):

Question	Why it matters
What task does the skill automate?	Defines the core workflow
Who is the target user?	Determines complexity and terminology level
What tools/APIs/CLIs does it use?	Determines dependencies and platform restrictions
What does the user provide as input?	Defines parameters and defaults
What should the output look like?	Defines the response template
Does it need API keys or credentials?	Determines `required_environment_variables`
Should it work on Claude.ai or only CLI?	Determines platform field and dynamic commands

在开始编写前，请先回答以下问题（若不清楚则询问用户）：

问题	重要性
该Skill要自动化什么任务？	定义核心工作流
目标用户是谁？	决定复杂度和术语级别
它使用哪些工具/API/CLI？	决定依赖项和平台限制
用户需要提供什么输入？	定义参数和默认值
输出应是什么样的？	定义响应模板
是否需要API密钥或凭据？	决定 `required_environment_variables`
它需要在Claude.ai上运行还是仅支持CLI？	决定平台字段和动态命令

Step 2: Plan the Skill Architecture

步骤2：规划Skill架构

Before writing SKILL.md, plan the structure. Read

references/architecture-patterns.md

for detailed guidance on each pattern.

在编写SKILL.md之前，先规划结构。每种模式的详细指南请阅读

references/architecture-patterns.md

。

Choose a Structural Pattern

选择结构模式

Pattern	When to use	Steps	Example
Linear	Single workflow, no branching	5-7	earnings-preview, etf-premium
Router	Multiple sub-tasks under one umbrella	3 + sub-skills	stock-correlation (4 sub-skills)
Methodology	Complex domain framework with sequential gates	7-9	sepa-strategy (9-step trading methodology)
Widget	Generates interactive UI output	4-5	options-payoff (extract + compute + render)
API Wrapper	Wraps an external API with many endpoints	3-5 + heavy references	funda-data (5 steps, 8 reference files)

模式	使用场景	步骤数量	示例
Linear（线性）	单一工作流，无分支	5-7	earnings-preview, etf-premium
Router（路由）	一个主任务下包含多个子任务	3 + 子Skill	stock-correlation（4个子Skill）
Methodology（方法论）	带有顺序节点的复杂领域框架	7-9	sepa-strategy（9步交易方法论）
Widget（组件）	生成交互式UI输出	4-5	options-payoff（提取+计算+渲染）
API Wrapper（API包装器）	封装带有多个端点的外部API	3-5 + 大量参考文件	funda-data（5步，8个参考文件）

Plan the Step Outline

规划步骤大纲

Write out the step names before writing content. Every skill should have:

Detection flow (Step 1) -- dynamically detect available tools, auth state, and runtime environment; build a decision tree for which method to use
Core methodology (Steps 2-N) -- the actual work, with pass/fail gates; each step that calls an external tool should have method alternatives based on what Step 1 detected
Respond to user (Final step) -- structured output template

Target 5-9 steps total. More than 9 means the skill should be split or use a router pattern.

在编写内容前先列出步骤名称。每个Skill都应包含：

检测流程（步骤1）——动态检测可用工具、授权状态和运行时环境；构建决策树以选择使用哪种方法
核心方法论（步骤2-N）——实际执行的工作，带有通过/失败节点；每个调用外部工具的步骤都应根据步骤1的检测结果提供备选方法
响应用户（最后一步）——结构化输出模板

目标总步骤数为5-9步。超过9步意味着该Skill应拆分或使用路由模式。

Plan the Detection Flow

规划检测流程

Every skill that touches external tools MUST start with a runtime detection flow. Read

references/dynamic-calling.md

for all patterns. The detection flow answers:

Question	How to detect	Decision
Is the CLI tool installed?	`command -v tool`	CLI path vs Python fallback
Is the user authenticated?	`tool auth status` / `echo $API_KEY`	Skip auth setup vs guide through it
Which runtime has the library?	`import lib` in terminal vs execute_code	Route to correct runtime
Is a richer tool available?	`gh --version` vs `git --version`	Rich path vs minimal path
Is live data reachable?	`curl -s endpoint`	Live data vs cached/default

The detection output feeds into a decision tree that the rest of the skill follows. Never assume — always check.

所有涉及外部工具的Skill都必须从运行时检测流程开始。所有模式请阅读

references/dynamic-calling.md

。检测流程需回答以下问题：

问题	检测方式	决策
CLI工具是否已安装？	`command -v tool`	使用CLI路径还是Python备选方案
用户是否已授权？	`tool auth status` / `echo $API_KEY`	跳过授权设置还是引导用户完成
哪个运行环境包含所需库？	在终端执行 `import lib` 或 `execute_code`	路由到正确的运行环境
是否有更丰富的工具可用？	`gh --version` vs `git --version`	使用丰富功能路径还是极简路径
是否可获取实时数据？	`curl -s endpoint`	使用实时数据还是缓存/默认数据

检测结果将生成一个决策树，Skill的后续流程将遵循该决策树。绝不假设——始终检查。

Plan Reference Files

规划参考文件

Decide what goes in SKILL.md vs references/:

In SKILL.md (under ~250 lines)	In references/
Step-by-step workflow	Detailed API documentation
Routing/decision tables	Code templates (>20 lines)
Parameter defaults table	Formulas and edge cases
Output format template	Troubleshooting database
Quick examples (1-3)	Comprehensive examples (4+)

决定哪些内容放在SKILL.md中，哪些放在references/目录下：

放在SKILL.md中（少于250行）	放在references/中
分步工作流	详细API文档
路由/决策表	代码模板（超过20行）
参数默认值表	公式和边缘情况
输出格式模板	故障排查数据库
快速示例（1-3个）	综合示例（4个及以上）

Step 3: Write the SKILL.md

步骤3：编写SKILL.md

Read

references/writing-guide.md

for detailed instructions on writing each section. Read

references/frontmatter-guide.md

for the complete YAML field reference.

编写各部分的详细说明请阅读

references/writing-guide.md

。完整的YAML字段参考请阅读

references/frontmatter-guide.md

。

Key Rules

关键规则

Frontmatter first:
```
name
```
(lowercase-hyphenated, max 64 chars) and
```
description
```
(exhaustive trigger list, max 1024 chars) are required. Description needs 5+ triggers including sideways entry points.
Step 1 = detection flow: Use
```
!
```
command`` with fallbacks to detect available tools, auth state, and runtime. Build a decision tree with multiple method paths (e.g., CLI preferred, Python fallback, built-in tools last resort). Never hardcode a single tool — always detect and adapt. See
```
references/dynamic-calling.md
```
.
Core steps with method alternatives: Each step that calls an external tool should offer at least 2 paths based on what Step 1 detected. Use pattern: "If
```
TOOL_A
```
detected → Method 1, otherwise → Method 2." Each step gets
```
## Step N: [Verb] [Object]
```
, a decision table if routing, a pass/fail gate if evaluative, and a reference pointer for deep content.
Defaults table: Every parameter MUST have an explicit default. No skill should ever stall waiting for input.
Final step = output template: Number every output section. Specify exactly what data goes in each. Include a verdict/grade system if evaluative.

See

references/skill-examples.md

for annotated examples of each pattern.

前置元数据优先：
```
name
```
（小写连字符格式，最多64字符）和
```
description
```
（全面的触发词列表，最多1024字符）为必填项。描述需包含5个以上触发词，包括间接触发场景。
步骤1=检测流程：使用
```
!
```
command``并提供备选方案来检测可用工具、授权状态和运行时环境。构建包含多种方法路径的决策树（例如，优先使用CLI，其次是Python备选方案，最后是内置工具）。绝不要硬编码单一工具——始终检测并适配。参见
```
references/dynamic-calling.md
```
。
带有备选方法的核心步骤：每个调用外部工具的步骤应根据步骤1的检测结果提供至少2种路径。使用模式："如果检测到
```
TOOL_A
```
→方法1，否则→方法2"。每个步骤采用
```
## Step N: [动词] [对象]
```
格式，若涉及路由则包含决策表，若涉及评估则包含通过/失败节点，并提供指向深度内容的参考指针。
默认值表：每个参数必须有明确的默认值。任何Skill都不应因等待输入而停滞。
最后一步=输出模板：为每个输出部分编号。明确指定每个部分应包含的数据。若为评估类Skill，需包含结论/评分系统。

每种模式的带注释示例请参见

references/skill-examples.md

。

Step 4: Write Reference Files

步骤4：编写参考文件

Read

references/writing-guide.md

for the full reference file authoring guide.

完整的参考文件编写指南请阅读

references/writing-guide.md

。

Key Rules

关键规则

Naming:
```
lowercase-hyphenated.md
```
, one file per concept-cluster
Size: Quick lookup 50-150 lines, deep guide 150-400 lines, catalog 400-900 lines
Structure: H1 title, H2 sections, code blocks, tables, edge cases section at end
Linking: Use backtick paths in SKILL.md steps and a
```
## Reference Files
```
section at the end

命名：
```
小写连字符格式.md
```
，每个概念集群对应一个文件
篇幅：快速查询类50-150行，深度指南类150-400行，目录类400-900行
结构：H1标题、H2章节、代码块、表格，末尾包含边缘情况章节
链接：在SKILL.md步骤中使用反引号路径，并在末尾添加
```
## Reference Files
```
章节

Step 5: Quality Check Before Delivery

步骤5：交付前的质量检查

Run the skill through the quality rubric in

references/quality-rubric.md

. Score each dimension.

使用

references/quality-rubric.md

中的质量评分标准对Skill进行检查。为每个维度打分。

Quick Checklist

快速检查清单

Frontmatter has
```
name
```
and
```
description
```
(both required)
Description has 5+ distinct trigger phrases
Description includes sideways entry points
SKILL.md is under 300 lines (ideally under 250)
Every parameter has an explicit default
Steps are numbered (## Step N: ...)
Each step has a clear exit condition or deliverable
Final step specifies exact output structure with numbered sections
Complex content is in reference files, not inline
Reference file pointers use backtick paths
Step 1 has a detection flow with
```
!
```
command`` checks and fallbacks (
```
|| echo "..."
```
)
Detection flow produces a decision tree with 2+ method paths
Core steps adapt behavior based on detection results (not hardcoded to one tool)
Separate runtimes treated as separate environments (terminal vs execute_code)
Legal/ethical disclaimers included where appropriate
No hardcoded ticker lists, tool paths, or static data that will go stale

If any item fails, fix it before delivering to the user.

前置元数据包含
```
name
```
和
```
description
```
（均为必填项）
描述包含5个以上不同的触发短语
描述包含间接触发场景
SKILL.md少于300行（理想情况下少于250行）
每个参数都有明确的默认值
步骤已编号（## Step N: ...）
每个步骤都有明确的退出条件或交付成果
最后一步指定了带编号的精确输出结构
复杂内容放在参考文件中，而非内联
参考文件指针使用反引号路径
步骤1包含使用
```
!
```
command``检查的检测流程和备选方案（
```
|| echo "..."
```
）
检测流程生成包含2种以上方法路径的决策树
核心步骤根据检测结果调整行为（而非硬编码为单一工具）
不同运行环境被视为独立环境（终端vs execute_code）
适当包含法律/伦理声明
无硬编码的代码列表、工具路径或会过时的静态数据

若任何一项未通过，请在交付给用户前修复。

Step 6: Improve an Existing Skill

步骤6：改进现有Skill

When the user asks to improve a skill:

当用户要求改进Skill时：

6a: Read the Current Skill

6a：阅读当前Skill

Load the skill with

skill_view(name)

or read the SKILL.md directly. Also read all reference files.

使用

skill_view(name)

加载Skill或直接读取SKILL.md。同时阅读所有参考文件。

6b: Score It Against the Rubric

6b：根据评分标准打分

Use the quality rubric from

references/quality-rubric.md

. Present the score breakdown to the user:

Dimension	Score	Issue
Trigger quality	6/10	Missing beginner phrasing
Defaults coverage	3/10	No defaults table
Step structure	8/10	Good, but Step 3 lacks exit gate
Output template	4/10	Vague "summarize results"
Reference usage	7/10	Good split, but missing troubleshooting

使用

references/quality-rubric.md

中的质量评分标准。向用户展示评分明细：

维度	得分	问题
触发词质量	6/10	缺少入门级表述
默认值覆盖	3/10	无默认值表
步骤结构	8/10	良好，但步骤3缺少退出节点
输出模板	4/10	"总结结果"表述模糊
参考文件使用	7/10	拆分合理，但缺少故障排查内容

6c: Propose Specific Improvements

6c：提出具体改进建议

List concrete changes ranked by impact:

[Highest impact] Add defaults table with 8+ parameters
[High impact] Rewrite description with 10+ trigger phrases
[Medium impact] Add structured output template to final step
...

按影响程度列出具体修改：

[最高影响] 添加包含8个以上参数的默认值表
[高影响] 重写描述，包含10个以上触发短语
[中等影响] 为最后一步添加结构化输出模板
...

6d: Apply Changes

6d：应用修改

After user approval, edit the skill. Use

skill_manage(action='patch', ...)

for targeted changes or

skill_manage(action='edit', ...)

for full rewrites.

获得用户批准后，编辑Skill。使用

skill_manage(action='patch', ...)

进行针对性修改，或使用

skill_manage(action='edit', ...)

进行全面重写。

Step 7: Evaluate a Skill

步骤7：评估Skill

When the user asks to evaluate or score a skill:

当用户要求评估或为Skill评分时：

7a: Load and Analyze

7a：加载并分析

Read the full SKILL.md and all reference files. Count lines, steps, triggers, defaults, reference files.

阅读完整的SKILL.md和所有参考文件。统计行数、步骤数、触发词数、默认值数、参考文件数。

7b: Score Against Rubric

7b：根据评分标准打分

Use the comprehensive rubric from

references/quality-rubric.md

. Score each of the 10 dimensions on a 1-10 scale.

使用

references/quality-rubric.md

中的综合评分标准。对10个维度分别按1-10分打分。

7c: Present the Scorecard

7c：展示评分卡

undefined

undefined

Skill Quality Scorecard: [skill-name]

#	Dimension	Score	Notes
1	Trigger quality	8/10	12 triggers, includes sideways entries
2	Defaults coverage	9/10	All 11 parameters have defaults
3	Step architecture	8/10	5 clear steps with gates
4	Reference file strategy	7/10	2 files, could use troubleshooting
5	Dynamic content	10/10	Dep check + live data injection
6	Output template	9/10	5 numbered sections + verdict
7	Error handling	6/10	Missing data handling unclear
8	Code/formula quality	8/10	Working JS, copy-paste ready
9	SKILL.md conciseness	7/10	196 lines, well within target
10	Domain accuracy	9/10	BS formulas correct, edge cases covered

Overall: 81/100 -- Production quality

#	Dimension	Score	Notes
1	Trigger quality	8/10	12 triggers, includes sideways entries
2	Defaults coverage	9/10	All 11 parameters have defaults
3	Step architecture	8/10	5 clear steps with gates
4	Reference file strategy	7/10	2 files, could use troubleshooting
5	Dynamic content	10/10	Dep check + live data injection
6	Output template	9/10	5 numbered sections + verdict
7	Error handling	6/10	Missing data handling unclear
8	Code/formula quality	8/10	Working JS, copy-paste ready
9	SKILL.md conciseness	7/10	196 lines, well within target
10	Domain accuracy	9/10	BS formulas correct, edge cases covered

Overall: 81/100 -- Production quality

Top 3 Improvements

undefined

undefined

Benchmark Reference

对标参考

For context, here are scores for known high-quality skills in this repo:

Skill	Score	Why
sepa-strategy	~90/100	9 steps, 7 refs, exhaustive triggers, structured verdict
options-payoff	~85/100	Strong defaults, working code, live data, clean output
stock-correlation	~80/100	Router pattern, 4 sub-skills, good defaults

为提供上下文，以下是本仓库中已知高质量Skill的得分：

Skill	Score	原因
sepa-strategy	~90/100	9个步骤，7个参考文件，全面的触发词，结构化结论
options-payoff	~85/100	完善的默认值，可运行代码，实时数据，简洁输出
stock-correlation	~80/100	路由模式，4个子Skill，良好的默认值

Step 8: Respond to the User

步骤8：响应用户

For Create mode

Create模式

Deliver:

The complete SKILL.md content
All reference files
A README.md for the skill directory
The quality scorecard (from Step 5)
Suggested next steps (test it, iterate, publish)

交付内容：

完整的SKILL.md内容
所有参考文件
Skill目录的README.md
质量评分卡（来自步骤5）
建议后续步骤（测试、迭代、发布）

For Improve mode

Improve模式

Deliver:

Before/after quality scores
Summary of changes made
Remaining improvement opportunities

交付内容：

修改前后的质量得分
修改内容摘要
剩余改进机会

For Evaluate mode

Evaluate模式

Deliver:

The full quality scorecard
Comparison to benchmark skills
Prioritized improvement list

交付内容：

完整的质量评分卡
与标杆Skill的对比
按优先级排序的改进列表

Reference Files

参考文件

```
references/dynamic-calling.md
```
-- Core reference: Detection flows, decision trees, method fallbacks, runtime awareness, and multi-tool adaptation patterns with annotated examples from production skills
```
references/writing-guide.md
```
-- Detailed instructions for writing SKILL.md sections, environment checks, defaults tables, output templates, and reference files
```
references/architecture-patterns.md
```
-- Linear, Router, Methodology, Widget, and API Wrapper patterns with examples and anti-patterns
```
references/frontmatter-guide.md
```
-- Complete YAML frontmatter field reference (name, description, platform, env vars, config, credentials)
```
references/quality-rubric.md
```
-- 10-dimension scoring rubric with 1-10 scales, benchmark scores, and score interpretation
```
references/skill-examples.md
```
-- Annotated excerpts from top skills showing why specific patterns work

```
references/dynamic-calling.md
```
-- 核心参考：检测流程、决策树、方法备选方案、运行时感知和多工具适配模式，包含生产Skill的带注释示例
```
references/writing-guide.md
```
-- 编写SKILL.md章节、环境检查、默认值表、输出模板和参考文件的详细说明
```
references/architecture-patterns.md
```
-- Linear、Router、Methodology、Widget和API Wrapper模式，包含示例和反模式
```
references/frontmatter-guide.md
```
-- 完整的YAML前置元数据字段参考（名称、描述、平台、环境变量、配置、凭据）
```
references/quality-rubric.md
```
-- 10维度评分标准，包含1-10分刻度、标杆得分和得分解读
```
references/skill-examples.md
```
-- 顶级Skill的带注释摘录，展示特定模式有效的原因