rate-skill

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Rate Skill

Skill 评分

Overview

概述

Audit SKILL.md files against quality standards from generate-skill best practices. Provides letter grade (A-F) and actionable recommendations.

Core principle: Measure skill quality objectively to improve activation reliability and context efficiency.

依据generate-skill最佳实践中的质量标准审核SKILL.md文件，提供A-F的字母评级以及可执行的改进建议。

核心原则： 客观衡量Skill质量，以提升激活可靠性和上下文效率。

When to Use

使用场景

Always use when:

Reviewing skills before publishing
Validating skill structure and formatting
Checking if skill meets quality standards
User asks to "rate", "grade", or "review" a skill

Useful for:

Skill authors validating their work
Maintainers reviewing PRs with new skills
Quality audits of skill repositories
Before submitting skills to marketplaces

Avoid when:

Evaluating non-skill documentation
Reviewing code (not skill definitions)
General code quality auditing

必须使用的场景：

发布前审核Skill
验证Skill的结构与格式
检查Skill是否符合质量标准
用户要求"评分""评级"或"审核"Skill时

适用场景：

Skill作者验证自己的作品
维护者审核包含新Skill的PR
Skill仓库的质量审计
提交Skill至市场前

避免使用的场景：

评估非Skill类文档
审核代码（而非Skill定义）
通用代码质量审计

How It Works

工作原理

Read specified SKILL.md file
Evaluate against quality criteria
Calculate scores per category
Generate letter grade (A-F)
Output findings with priorities
Provide actionable recommendations

读取指定的SKILL.md文件
依据质量标准进行评估
按类别计算得分
生成A-F的字母评级
按优先级输出评估结果
提供可执行的改进建议

Quality Criteria

质量评估标准

Category	Weight	Criteria
Length	20%	Under 500 lines (or progressive disclosure)
Conciseness	20%	Clear, scannable, no fluff
Repetitiveness	15%	No redundant content
Structure	15%	Required sections present and ordered
Triggers	15%	3-5+ specific activation phrases
Examples	10%	Good/Bad code comparisons
Troubleshooting	5%	Common issues addressed

类别	权重	评估标准
篇幅	20%	少于500行（或采用渐进式披露）
简洁性	20%	表述清晰、易于扫描、无冗余内容
重复性	15%	无冗余内容
结构	15%	包含必填章节且顺序正确
触发词	15%	包含3-5个以上具体的激活短语
示例	10%	包含好坏代码对比
故障排查	5%	涵盖常见问题解决方案

Length (20%)

篇幅（20%）

Scores: A: <500 or progressive disclosure | B: 500-600 | C: 600-800 | D: 800-1000 | F: >1000

Checks: Line count, reference/ directory, progressive disclosure links

评分标准： A: 少于500行或采用渐进式披露 | B: 500-600行 | C: 600-800行 | D: 800-1000行 | F: 超过1000行

检查项： 行数统计、reference/目录、渐进式披露链接

Conciseness (20%)

简洁性（20%）

Scores: A: High info density, scannable | B: Mostly concise | C: Some wordiness | D: Verbose | F: Excessive

Red flags: Long paragraphs (>5 sentences), redundant explanations, flowery language

评分标准： A: 信息密度高、易于扫描 | B: 基本简洁 | C: 存在部分冗余表述 | D: 表述冗长 | F: 过于啰嗦

警示信号： 长段落（超过5句）、重复解释、修饰性语言过多

Repetitiveness (15%)

重复性（15%）

Scores: A: Zero redundancy | B: 1-2 overlaps | C: 3-4 overlaps | D: 5+ overlaps | F: Heavy redundancy

Common: Format in section AND example, repeated "use when", duplicate trigger phrases

评分标准： A: 完全无冗余 | B: 1-2处重复 | C: 3-4处重复 | D: 5处以上重复 | F: 大量冗余

常见问题： 章节与示例中格式重复、"使用场景"重复、触发短语重复

Structure (15%)

结构（15%）

Scores: A: All required sections | B: Missing 1 optional | C: Missing 2-3 | D: Missing required | F: Severely lacking

Required: Frontmatter, Overview, When to Use, Main content, Examples (Good/Bad), Troubleshooting, Integration

评分标准： A: 包含所有必填章节 | B: 缺失1个可选章节 | C: 缺失2-3个章节 | D: 缺失必填章节 | F: 结构严重缺失

必填章节： 前置元数据、概述、使用场景、主要内容、示例（好坏对比）、故障排查、集成说明

Triggers (15%)

触发词（15%）

Scores: A: 5+ specific | B: 3-4 good | C: 2 phrases | D: 1 vague | F: None

Quality: User language ("when asked to X"), specific situations, multiple contexts, concrete not abstract

评分标准： A: 5个以上具体触发词 | B: 3-4个优质触发词 | C: 2个触发词 | D: 1个模糊触发词 | F: 无触发词

优质触发词标准： 符合用户语言习惯（如"当被要求X时"）、针对具体场景、覆盖多上下文、表述具体而非抽象

Examples (10%)

示例（10%）

Scores: A: 3+ with Good/Bad | B: 2 with comparisons | C: 1 comparison | D: No comparisons | F: None

Quality: Uses tags, includes explanations, real scenarios, syntax highlighting

评分标准： A: 3个以上带好坏对比的示例 | B: 2个带对比的示例 | C: 1个对比示例 | D: 无对比示例 | F: 无示例

优质示例标准： 使用标签、包含解释、基于真实场景、带语法高亮

Troubleshooting (5%)

故障排查（5%）

Scores: A: 5+ pairs | B: 3-4 pairs | C: 1-2 basic | D: Vague | F: None

Quality: Clear problem, cause identified, solution with code, explanation

评分标准： A: 5组以上问题-解决方案 | B: 3-4组 | C: 1-2组基础方案 | D: 表述模糊 | F: 无故障排查内容

优质故障排查标准： 问题清晰、明确原因、带代码的解决方案、有解释说明

Output Format

输出格式

markdown

undefined

markdown

undefined

Skill Rating: [Letter Grade]

Skill 评级: [字母等级]

Summary

摘要

File: path/to/SKILL.md
Lines: XXX lines
Overall Grade: [A/B/C/D/F] ([Score]/100)
Status: [Production Ready / Needs Work / Not Ready]

文件: path/to/SKILL.md
行数: XXX 行
总体评级: [A/B/C/D/F] ([得分]/100)
状态: [可投入生产 / 需要优化 / 未达标]

Category Scores

类别得分

Category	Score	Grade	Status
Length	XX/20	[A-F]	[✅/⚠️/❌]
Conciseness	XX/20	[A-F]	[✅/⚠️/❌]
Repetitiveness	XX/15	[A-F]	[✅/⚠️/❌]
Structure	XX/15	[A-F]	[✅/⚠️/❌]
Triggers	XX/15	[A-F]	[✅/⚠️/❌]
Examples	XX/10	[A-F]	[✅/⚠️/❌]
Troubleshooting	XX/5	[A-F]	[✅/⚠️/❌]

类别	得分	评级	状态
篇幅	XX/20	[A-F]	[✅/⚠️/❌]
简洁性	XX/20	[A-F]	[✅/⚠️/❌]
重复性	XX/15	[A-F]	[✅/⚠️/❌]
结构	XX/15	[A-F]	[✅/⚠️/❌]
触发词	XX/15	[A-F]	[✅/⚠️/❌]
示例	XX/10	[A-F]	[✅/⚠️/❌]
故障排查	XX/5	[A-F]	[✅/⚠️/❌]

Findings by Priority

优先级问题清单

❌ Critical Issues (Fix Before Publishing)

❌ 关键问题（发布前必须修复）

[Issue description]
- Impact: [Why this matters]
- Fix: [Specific action to take]

[问题描述]
- 影响: [问题的严重性]
- 修复方案: [具体操作步骤]

⚠️ Important Issues (Should Fix)

⚠️ 重要问题（建议修复）

[Issue description]
- Impact: [Why this matters]
- Fix: [Specific action to take]

[问题描述]
- 影响: [问题的严重性]
- 修复方案: [具体操作步骤]

📋 Nice to Have

📋 优化建议

[Suggestion]
- Benefit: [Why this helps]

[建议内容]
- 收益: [优化后的好处]

Strengths

优势

[What this skill does well]
[Another strength]

[Skill的优点]
[另一个优点]

Priority Action Items

优先级行动项

[Priority 1 action]
[Priority 2 action]
[Priority 3 action]

[优先级1行动]
[优先级2行动]
[优先级3行动]

Estimated Improvements

预期提升效果

Fix critical issues: +[X] points
Address important issues: +[X] points
Potential grade: [Current] → [Target]

undefined

修复关键问题: +[X] 分
解决重要问题: +[X] 分
预期评级: [当前评级] → [目标评级]

undefined

Usage

使用方法

Basic rating:

bash

/rate-skill skills/example-skill/SKILL.md

Rate after changes:

bash

undefined

基础评分：

bash

/rate-skill skills/example-skill/SKILL.md

优化后重新评分：

bash

undefined

Make improvements

进行优化

[edit SKILL.md]

[编辑 SKILL.md]

Re-rate

重新评分

/rate-skill skills/example-skill/SKILL.md


**Compare before/after:**
```bash

/rate-skill skills/example-skill/SKILL.md


**对比优化前后评分：**
```bash

Rate original

对原始版本评分

/rate-skill skills/track-session/SKILL.md

Make improvements

进行优化

[condense, remove redundancy]

[精简内容、移除冗余]

Rate again to see improvement

重新评分查看提升效果

/rate-skill skills/track-session/SKILL.md

undefined

/rate-skill skills/track-session/SKILL.md

undefined

Grading Scale

评级量表

Grade	Score	Meaning
A	90-100	Excellent - Production ready
B	80-89	Good - Minor improvements recommended
C	70-79	Acceptable - Needs work before publishing
D	60-69	Poor - Significant issues to address
F	0-59	Failing - Major overhaul needed

Status mapping:

A-B: Production Ready ✅
C: Needs Work ⚠️
D-F: Not Ready ❌

等级	得分	含义
A	90-100	优秀 - 可投入生产
B	80-89	良好 - 建议小幅优化
C	70-79	合格 - 发布前需要优化
D	60-69	较差 - 存在严重问题需解决
F	0-59	不合格 - 需要全面重构

状态映射：

A-B: 可投入生产 ✅
C: 需要优化 ⚠️
D-F: 未达标 ❌

Examples

示例

Example 1: Rating a High-Quality Skill

示例1：高质量Skill评级

Input:

/rate-skill skills/track-session/SKILL.md

Output:

markdown

undefined

输入：

/rate-skill skills/track-session/SKILL.md

输出：

markdown

undefined

Skill Rating: A

Skill 评级: A

Summary

摘要

File: skills/track-session/SKILL.md
Lines: 489 lines
Overall Grade: A (92/100)
Status: Production Ready

文件: skills/track-session/SKILL.md
行数: 489 行
总体评级: A (92/100)
状态: 可投入生产

Category Scores

类别得分

Category	Score	Grade	Status
Length	20/20	A	Pass
Conciseness	18/20	A	Pass
Repetitiveness	14/15	A	Pass
Structure	15/15	A	Pass
Triggers	15/15	A	Pass
Examples	9/10	A	Pass
Troubleshooting	5/5	A	Pass

类别	得分	评级	状态
篇幅	20/20	A	通过
简洁性	18/20	A	通过
重复性	14/15	A	通过
结构	15/15	A	通过
触发词	15/15	A	通过
示例	9/10	A	通过
故障排查	5/5	A	通过

Strengths

优势

Excellent progressive disclosure with reference/VERIFICATION.md
10+ diverse trigger phrases
Strong examples with Good/Bad comparisons

采用了优秀的渐进式披露，搭配reference/VERIFICATION.md
包含10个以上多样化的触发短语
示例质量高，包含好坏对比

Priority Action Items

优先级行动项

Consider adding one more example for edge cases (optional)


**Note:** High-scoring skills get a short report focused on strengths and optional improvements.

考虑新增1个边缘场景示例（可选）


**说明：** 高分Skill的报告重点突出优势和可选优化建议。

Example 2: Rating a Skill That Needs Work

示例2：需要优化的Skill评级

Input:

/rate-skill skills/problematic-skill/SKILL.md

Output:

markdown

undefined

输入：

/rate-skill skills/problematic-skill/SKILL.md

输出：

markdown

undefined

Skill Rating: C

Skill 评级: C

Summary

摘要

File: skills/problematic-skill/SKILL.md
Lines: 742 lines
Overall Grade: C (74/100)
Status: Needs Work

文件: skills/problematic-skill/SKILL.md
行数: 742 行
总体评级: C (74/100)
状态: 需要优化

Findings by Priority

优先级问题清单

Critical Issues

关键问题

Length: 742 lines without progressive disclosure
- Impact: High context usage, harder to scan
- Fix: Move detailed content to reference/ directory
Only 2 trigger phrases in description
- Impact: Poor activation reliability
- Fix: Add 3-5 specific user phrases and situations

篇幅：742行且未采用渐进式披露
- 影响: 上下文占用高，难以扫描
- 修复方案: 将详细内容移至reference/目录
描述中仅包含2个触发短语
- 影响: 激活可靠性差
- 修复方案: 添加3-5个具体的用户场景短语

Important Issues

重要问题

Verbose mode descriptions (30+ lines each)
- Fix: Condense to 2-3 lines per mode

模式描述过于冗长（每个模式超过30行）
- 修复方案: 精简为每个模式2-3行

Priority Action Items

优先级行动项

Implement progressive disclosure (move 200+ lines to reference/)
Add 3+ trigger phrases to description
Condense verbose sections

实现渐进式披露（将200多行内容移至reference/）
在描述中添加3个以上触发短语
精简冗长章节

Estimated Improvements

预期提升效果

Fix critical issues: +12 points -> 86 (B)
Potential grade: C -> A


**Note:** Lower-scoring skills get detailed findings with specific fixes and an improvement roadmap.

修复关键问题: +12分 → 86分（B级）
预期评级: C → A


**说明：** 低分Skill的报告包含详细问题清单、具体修复方案和优化路线图。

Troubleshooting

故障排查

Problem: Can't find SKILL.md file

问题：找不到SKILL.md文件

Cause: Path incorrect or file doesn't exist.

Solution:

bash

undefined

原因： 路径错误或文件不存在。

解决方案：

bash

undefined

Verify file exists

验证文件是否存在

ls skills/skill-name/SKILL.md

Use correct path

使用正确路径

/rate-skill skills/skill-name/SKILL.md

undefined

/rate-skill skills/skill-name/SKILL.md

undefined

Problem: Rating seems too harsh

问题：评级似乎过于严格

Cause: Standards are strict for good reason - quality matters for activation.

Solution:

Review specific findings
Compare to high-quality skills
Focus on critical issues first
Remember: B grade is still "good"

原因： 标准严格是有原因的——质量直接影响激活效果。

解决方案：

查看具体评估结果
与高质量Skill对比
优先解决关键问题
记住：B级已经属于"良好"

Problem: Grade improved but still low

问题：评级有所提升但仍偏低

Cause: Multiple categories need attention.

Solution:

Focus on highest-weight categories first (Length, Conciseness)
Fix critical issues before nice-to-haves
Re-rate after each major change
Use "Estimated Improvements" as roadmap

原因： 多个类别需要优化。

解决方案：

优先优化权重最高的类别（篇幅、简洁性）
先修复关键问题再处理优化建议
每次重大优化后重新评分
以"预期提升效果"作为优化路线图

Problem: Don't know how to fix an issue

问题：不知道如何修复某个问题

Cause: Fix recommendation unclear.

Solution:

Check generate-skill examples for patterns
Review high-rated skills for reference
Ask for specific help on that issue
Consult CLAUDE.md for SkillBox guidelines

原因： 修复建议不够明确。

解决方案：

查看generate-skill示例中的模式
参考高评级Skill的写法
针对该问题寻求具体帮助
查阅CLAUDE.md中的SkillBox指南

Integration

集成

This skill works with:

generate-skill - Use after generating to validate quality
Skill development workflow - Rate before committing/publishing
Quality control - Gate for accepting skills into repositories
Continuous improvement - Track quality metrics over time

Workflow:

bash

undefined

本Skill可与以下工具配合使用：

generate-skill - 生成Skill后用于验证质量
Skill开发流程 - 提交/发布前进行评级
质量控制 - 作为Skill仓库的准入门槛
持续改进 - 跟踪质量指标变化

工作流：

bash

undefined

Create skill

创建Skill

/generate-skill new-feature

Rate it

进行评级

/rate-skill skills/new-feature/SKILL.md

Fix issues

修复问题

[make improvements]

[进行优化]

Re-rate

重新评级

/rate-skill skills/new-feature/SKILL.md

When A or B grade, publish

达到A或B级后发布

git add skills/new-feature/ git commit -m "Add new-feature skill"


**Quality gates:**
- A-B: Merge to main ✅
- C: Request changes ⚠️
- D-F: Reject until improved ❌

git add skills/new-feature/ git commit -m "Add new-feature skill"


**质量门禁：**
- A-B级：合并至主分支 ✅
- C级：要求修改 ⚠️
- D-F级：拒绝合并，需优化后重新提交 ❌

References

参考资料

Based on:

generate-skill best practices
SkillBox CLAUDE.md guidelines
obra/superpowers patterns
Vercel agent-skills standards

Related:

generate-skill
SkillBox CLAUDE.md

基于以下标准：

generate-skill最佳实践
SkillBox CLAUDE.md指南
obra/superpowers模式
Vercel agent-skills标准

相关链接：

generate-skill
SkillBox CLAUDE.md