skill-creator

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Skill Creator

Create, improve, and audit AI agent skills. Every skill follows 14 proven structural patterns.

Scope: Skills only. NOT for creating agents (

wagents new agent

), building MCP servers (

/mcp-creator

), or running existing skills. This repo uses raw

SKILL.md

format committed directly to

skills/

创建、改进和审计AI Agent技能。每个技能都遵循14种经过验证的结构模式。

**适用范围：**仅针对技能。不适用于创建Agent（

wagents new agent

）、搭建MCP服务器（

/mcp-creator

）或运行现有技能。本仓库使用原始

SKILL.md

格式，直接提交至

skills/

目录。

Dispatch

调度（Dispatch）

$ARGUMENTS	Action	Example
`create <name>` / `new <name>`	Develop (new)	`/skill-creator create my-analyzer`
`create <name> --from <source>`	Develop (new, from exemplar)	`/skill-creator create my-analyzer --from wargame`
`improve <name>` / `improve <path>`	Develop (existing)	`/skill-creator improve add-badges`
`audit <name>`	Audit	`/skill-creator audit honest-review`
`audit --all`	Audit All	`/skill-creator audit --all`
`dashboard`	Dashboard	`/skill-creator dashboard`
`package <name>` / `package --all`	Package	`/skill-creator package wargame`
Natural language skill idea	Auto: Develop (new)	`"tool that audits Python type safety"`
Skill name + modification verb	Auto: Develop (existing)	`"refactor the wargame skill"`
Path to SKILL.md	Auto: Develop (existing)	`skills/wargame/SKILL.md`
"MCP server" / "agent" / "run"	Refuse + redirect	—
Empty	Gallery	`/skill-creator`

参数	操作	示例
`create <name>` / `new <name>`	开发（新技能）	`/skill-creator create my-analyzer`
`create <name> --from <source>`	开发（基于示例创建新技能）	`/skill-creator create my-analyzer --from wargame`
`improve <name>` / `improve <path>`	开发（现有技能优化）	`/skill-creator improve add-badges`
`audit <name>`	审计	`/skill-creator audit honest-review`
`audit --all`	全量审计	`/skill-creator audit --all`
`dashboard`	仪表盘	`/skill-creator dashboard`
`package <name>` / `package --all`	打包	`/skill-creator package wargame`
自然语言描述的技能需求	自动：开发（新技能）	`"用于审计Python类型安全性的工具"`
技能名称+修改动词	自动：开发（现有技能优化）	`"重构wargame技能"`
SKILL.md文件路径	自动：开发（现有技能优化）	`skills/wargame/SKILL.md`
"MCP server" / "agent" / "run"	拒绝操作并引导至正确工具	—
无参数	技能库	`/skill-creator`

Auto-Detection Heuristic

自动检测规则

If no explicit mode keyword is provided:

Path ending in
```
SKILL.md
```
or directory under
```
skills/
```
→ Develop (existing)
Existing skill name + modification verb (improve, refactor, enhance, update, fix, rewrite, optimize, polish, revise, change) → Develop (existing)
```
--from <source>
```
in arguments → Develop (new, from exemplar)
New capability description ("I want to build...", "tool that...", "skill for...") → Develop (new) — derive name, confirm before scaffolding
"MCP server", "agent", "run" → refuse gracefully and redirect
Ambiguous → ask the user which mode they want

如果未提供明确的模式关键字：

以
```
SKILL.md
```
结尾的路径或
```
skills/
```
下的目录 → 开发（现有技能优化）
现有技能名称+修改动词（improve, refactor, enhance, update, fix, rewrite, optimize, polish, revise, change） → 开发（现有技能优化）
参数中包含
```
--from <source>
```
→ 开发（基于示例创建新技能）
新能力描述（"我想构建...", "用于...的工具", "面向...的技能"） → 开发（新技能） — 自动生成名称，在搭建脚手架前需确认
包含"MCP server"、"agent"、"run" → 友好拒绝并引导至正确工具
模糊需求 → 询问用户选择对应模式

Quick Start

快速开始

bash

wagents new skill <name>           # Scaffold from template
wagents validate                   # Check all skills
uv run python skills/skill-creator/scripts/audit.py skills/<name>/  # Score quality
wagents package <name>             # Package as portable ZIP

bash

wagents new skill <name>           # 从模板搭建技能脚手架
wagents validate                   # 检查所有技能
uv run python skills/skill-creator/scripts/audit.py skills/<name>/  # 对技能进行质量评分
wagents package <name>             # 打包为可移植ZIP文件

Skill Development

技能开发

Unified process for creating new skills and improving existing ones. Load

references/workflow.md

for the full procedure.

Step	New Skill	Existing Skill
1. Understand	Define use cases, scope, patterns	Audit + understand user's intent
2. Plan	Structure, description, frontmatter	Gap analysis + improvement plan (approval gate)
3. Scaffold	`wagents new skill <name>`	Skip
4. Build	Write/edit body, references, scripts, templates, evals	Same
5. Validate	`wagents validate` + `audit.py`	Same
6. Iterate	Test, identify issues, loop to Step 4	Same

适用于新技能创建和现有技能优化的统一流程。查看

references/workflow.md

获取完整步骤。

步骤	新技能	现有技能
1. 需求理解	定义使用场景、范围和模式	审计+理解用户需求
2. 规划	结构设计、描述撰写、前置元数据	差距分析+优化方案（需用户确认）
3. 脚手架搭建	`wagents new skill <name>`	跳过
4. 开发实现	编写/编辑正文、参考文档、脚本、模板、评估用例	相同
5. 验证	`wagents validate` + `audit.py`	相同
6. 迭代优化	测试、定位问题、回到步骤4	相同

Audit

审计

Score a skill using deterministic analysis + AI review. Load

references/audit-guide.md

通过确定性分析+AI评审对技能进行质量评分。查看

references/audit-guide.md

获取详细说明。

Audit All

全量审计

Comparative ranking of all repository skills. Load

references/audit-guide.md

§ Audit All.

对仓库中所有技能进行对比排名。查看

references/audit-guide.md

中「全量审计」章节。

Dashboard

仪表盘

Render visual creation process monitor or audit quality dashboard. Load

references/audit-guide.md

§ Dashboard.

Auto-detects mode from data:

phases

field → process monitor;

skills

array → audit overview.

渲染可视化的技能创建过程监控或审计质量仪表盘。查看

references/audit-guide.md

中「仪表盘」章节。

会自动根据数据检测模式：存在

phases

字段→过程监控；存在

skills

数组→审计概览。

Gallery (Empty Arguments)

技能库（无参数调用）

Present skill inventory with scores and available actions. Run

uv run python skills/skill-creator/scripts/audit.py --all --format table

, display results, offer mode menu.

展示技能清单及评分和可用操作。运行

uv run python skills/skill-creator/scripts/audit.py --all --format table

，显示结果并提供模式选择菜单。

Package

打包

Package skills into portable ZIP files for Claude Code Desktop import. Load

references/packaging-guide.md

for ZIP structure, manifest schema, portability checks, and cross-agent compatibility.

bash

wagents package <name>            # Single skill → <name>-v<version>.skill.zip
wagents package --all             # All skills → dist/ with manifest.json
wagents package --all --dry-run   # Check portability without creating ZIPs

将技能打包为可移植ZIP文件，用于Claude Code Desktop导入。查看

references/packaging-guide.md

了解ZIP结构、清单schema、可移植性检查及跨Agent兼容性。

bash

wagents package <name>            # 单个技能→<name>-v<version>.skill.zip
wagents package --all             # 所有技能→打包至dist/目录并生成manifest.json
wagents package --all --dry-run   # 检查可移植性但不创建ZIP文件

Hooks

钩子（Hooks）

PreToolUse hooks intercept tool calls during skill execution. The

hooks:

frontmatter field scopes hooks to this skill only — they activate when the skill is loaded and deactivate when it completes.

PreToolUse钩子会在技能执行期间拦截工具调用。

hooks:

前置元数据字段仅将钩子作用于当前技能——钩子在技能加载时激活，完成时停用。

State Management

状态管理

Creation progress persists at

~/.claude/skill-progress/<name>.json

. Read/write via

scripts/progress.py

. Survives session restarts. Use

--state-dir

to override the default location.

技能创建进度会持久化存储在

~/.claude/skill-progress/<name>.json

。可通过

scripts/progress.py

进行读写。进度会在会话重启后保留。使用

--state-dir

参数可覆盖默认存储位置。

Reference File Index

参考文件索引

File	Content	Read When
`references/workflow.md`	Unified 6-step skill development process for new and existing skills	Develop (new), Develop (existing)
`references/audit-guide.md`	Audit procedure, Audit All, Dashboard rendering, Gallery, grade thresholds	Audit, Audit All, Dashboard, Gallery
`references/proven-patterns.md`	14 structural patterns with examples from repo skills	Step 4 (Build), gap analysis
`references/best-practices.md`	Anthropic guide + superpowers methodology + cross-agent awareness	Step 2 (Plan), Step 4 (Build), description writing
`references/frontmatter-spec.md`	Full field catalog, invocation matrix, decision tree	Step 3 (Scaffold), frontmatter configuration
`references/packaging-guide.md`	ZIP structure, manifest schema, portability checks, import instructions	Package
`references/evaluation-rubric.md`	11 scoring dimensions, grade thresholds, pressure testing	Audit (pressure testing), scoring targets

Read reference files as indicated by the "Read When" column above. Do not rely on memory or prior knowledge of their contents.

文件	内容	阅读时机
`references/workflow.md`	适用于新技能和现有技能的6步统一开发流程	开发（新技能）、开发（现有技能优化）
`references/audit-guide.md`	审计流程、全量审计、仪表盘渲染、技能库、评分阈值	审计、全量审计、仪表盘、技能库
`references/proven-patterns.md`	14种结构模式及仓库内技能示例	步骤4（开发实现）、差距分析
`references/best-practices.md`	Anthropic官方指南+超能力方法论+跨Agent适配	步骤2（规划）、步骤4（开发实现）、描述撰写
`references/frontmatter-spec.md`	完整字段目录、调用矩阵、决策树	步骤3（脚手架搭建）、前置元数据配置
`references/packaging-guide.md`	ZIP结构、清单schema、可移植性检查、导入说明	打包
`references/evaluation-rubric.md`	11个评分维度、等级阈值、压力测试规则	审计（压力测试）、评分目标

请根据「阅读时机」列的提示查阅对应参考文件。请勿依赖记忆或过往知识。

Core Principles

核心原则

Conciseness is respect — The context window is shared. Every line competes with the agent's working memory. Earn every line or delete it.

Progressive disclosure — Frontmatter for discovery (~100 tokens), body for dispatch (<5K tokens), references for deep knowledge (on demand), scripts/templates for execution (never loaded).

Self-exemplar — This skill follows every pattern it teaches. When in doubt, look at how skill-creator applies it.

简洁即是尊重——上下文窗口是共享资源。每一行内容都在与Agent的工作内存竞争。确保每一行都有价值，否则删除。

渐进式披露——前置元数据用于技能发现（约100 tokens），正文用于调度（<5K tokens），参考文件用于深度知识（按需查阅），脚本/模板用于执行（不会加载到上下文）。

自我示例——本技能遵循它所教授的所有模式。如有疑问，可查看skill-creator自身的实现方式。

Critical Rules

关键规则

Run
```
uv run wagents validate
```
before declaring any skill complete

Run

uv run python skills/skill-creator/scripts/audit.py

after every significant SKILL.md change

Never create a skill without a dispatch table — it is the routing contract
Never create a dispatch table without an empty-args handler — unrouted input is a bug
Every reference file must appear in the Reference File Index — orphan refs are invisible
Every indexed reference must exist on disk — phantom refs cause agent errors
Body must stay under 500 lines (below frontmatter) — move detail to references
Description must include "Use when" trigger phrases AND "NOT for" exclusions
Names must be kebab-case, 2-64 chars, no consecutive hyphens, no reserved words
Scripts use argparse + JSON to stdout — no custom output formats
Templates are self-contained HTML with no external dependencies
Do NOT call
```
wagents docs generate
```
— delegate to docs-steward
Do NOT create agents or MCP servers — refuse gracefully and redirect
Improving existing skills requires presenting an improvement plan and getting user approval before implementing changes
Audit mode is read-only — never modify the skill being audited
Update evals when dispatch behavior or modes change — stale evals are invisible bugs

Canonical terms (use these exactly throughout):

Modes: "Develop (new)", "Develop (existing)", "Audit", "Audit All", "Dashboard", "Package", "Gallery"
Steps (Development): "Understand", "Plan", "Scaffold", "Build", "Validate", "Iterate"
Grade scale: "A" (90+), "B" (75-89), "C" (60-74), "D" (40-59), "F" (<40)
Patterns: "dispatch-table", "reference-file-index", "critical-rules", "canonical-vocabulary", "scope-boundaries", "classification-gating", "scaling-strategy", "state-management", "scripts", "templates", "hooks", "progressive-disclosure", "body-substitutions", "stop-hooks"
Audit dimensions: "frontmatter", "description", "dispatch-table", "body-structure", "pattern-coverage", "reference-quality", "critical-rules", "script-quality", "portability", "conciseness", "canonical-vocabulary"

完成技能开发前必须运行
```
uv run wagents validate
```
每次对SKILL.md进行重大修改后，必须运行
```
uv run python skills/skill-creator/scripts/audit.py
```
绝不能创建没有调度表的技能——调度表是路由契约
绝不能创建没有无参数处理器的调度表——未路由的输入属于Bug
所有参考文件必须出现在参考文件索引中——未索引的参考文件无法被发现
所有已索引的参考文件必须存在于磁盘上——不存在的参考文件会导致Agent报错
正文（前置元数据下方）必须控制在500行以内——详细内容请移至参考文件
描述必须包含「适用场景」触发短语及「不适用场景」排除项
技能名称必须使用短横线命名法（kebab-case），长度2-64字符，无连续短横线，无保留字
脚本使用argparse+JSON输出到标准输出——禁止使用自定义输出格式
模板为独立HTML文件，无外部依赖
请勿调用
```
wagents docs generate
```
——该操作由docs-steward负责
请勿创建Agent或MCP服务器——友好拒绝并引导至正确工具
优化现有技能前必须提交优化方案并获得用户批准
审计模式为只读——绝不能修改被审计的技能
当调度行为或模式发生变化时，必须更新评估用例——过时的评估用例相当于隐形Bug

标准术语（请在所有场景中严格使用以下术语）：

模式："Develop (new)", "Develop (existing)", "Audit", "Audit All", "Dashboard", "Package", "Gallery"
开发步骤："Understand", "Plan", "Scaffold", "Build", "Validate", "Iterate"
评分等级："A"（90分以上）、"B"（75-89分）、"C"（60-74分）、"D"（40-59分）、"F"（40分以下）
结构模式："dispatch-table", "reference-file-index", "critical-rules", "canonical-vocabulary", "scope-boundaries", "classification-gating", "scaling-strategy", "state-management", "scripts", "templates", "hooks", "progressive-disclosure", "body-substitutions", "stop-hooks"
审计维度："frontmatter", "description", "dispatch-table", "body-structure", "pattern-coverage", "reference-quality", "critical-rules", "script-quality", "portability", "conciseness", "canonical-vocabulary"