Token导航 LogoToken导航

Agent Skill · 研究检索

clean-code(干净的代码)

clean-code 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

clean-code 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 46,976。

总安装 753周安装 32GitHub Stars 1下载 264核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:clean-code(干净的代码)
来源仓库:https://github.com/pixel-process-ug/superkit-agents
仓库路径:skills/clean-code
安装命令:
npx skills add https://github.com/pixel-process-ug/superkit-agents --skill clean-code
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/pixel-process-ug/superkit-agents --skill clean-code

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

clean-code 用于查找、检索和筛选相关信息,适合快速定位候选结果。适用于需要根据关键词或任务场景从来源线索中筛选信息的场景。通过 npx skills add 命令从指定 GitHub 仓库安装并使用该技能。安装前需确认权限范围和维护状态,注意可能触发联网或文件读写操作。

clean-code 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/clean-code

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/pixel-process-ug/superkit-agents --skill clean…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

clean-code 怎么安装?

在终端执行 npx skills add https://github.com/pixel-process-ug/superkit-agents --skill clean-code。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

clean-code 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

clean-code 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

clean-code 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

clean-code 用的人多吗?

公开数据是累计安装 753 次,近一周安装 32 次,下载 264 次,GitHub 1 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

clean-code 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
clean-code

干净的代码 · 本页

7531需要联网未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Clean Code

Overview

Apply clean code principles to produce readable, maintainable, and testable software. This skill covers SOLID principles, DRY application, code smell identification, refactoring patterns, naming conventions, error handling, and complexity management. Based on the works of Robert C. Martin, Martin Fowler, and Kent Beck.

Announce at start: "I'm using the clean-code skill to improve code quality."


Phase 1: Analyze Current Code

Goal: Read and understand the code in full context before changing anything.

Actions

  1. Read the code in its full context (not just the snippet)
  2. Identify the code's responsibility and purpose
  3. Measure cyclomatic complexity
  4. Map coupling and dependencies
  5. Note any existing tests

STOP — Do NOT proceed to Phase 2 until:

  • Code is read in full context
  • Purpose and responsibility are understood
  • Complexity hotspots are identified
  • Existing test coverage is known

Phase 2: Identify Code Smells

Goal: Catalog all code smells using the reference tables below.

Bloaters

SmellDetectionRefactoring
Long Method> 30 linesExtract Method
Large Class> 300 lines or > 5 responsibilitiesExtract Class
Long Parameter List> 3 parametersIntroduce Parameter Object
Data ClumpsSame params appear togetherExtract Class
Primitive ObsessionPrimitives instead of small objectsReplace with Value Object

Object-Orientation Abusers

SmellDetectionRefactoring
Switch StatementsSwitch on typeReplace with Polymorphism
Parallel InheritanceEvery subclass requires parallel subclassMerge hierarchies
Refused BequestSubclass ignores inherited methodsReplace Inheritance with Delegation

Change Preventers

SmellDetectionRefactoring
Divergent ChangeOne class changed for multiple reasonsExtract Class (SRP)
Shotgun SurgeryOne change touches many classesMove Method, Inline Class

Dispensables

SmellDetectionRefactoring
Dead CodeUnreachable or unusedRemove
Speculative GeneralityUnused abstractions "just in case"Collapse Hierarchy, Remove
Comments explaining bad codeComments compensating for unclear codeRename, Extract Method

STOP — Do NOT proceed to Phase 3 until:

  • All code smells are cataloged
  • Each smell has a priority (high/medium/low)
  • Refactoring approach is identified for each

Phase 3: Apply Refactoring

Goal: Apply refactoring patterns one at a time, verifying tests after each.

Actions

  1. Apply ONE refactoring at a time
  2. Run tests after each change
  3. If any test fails, revert immediately
  4. Continue until code is clean
  5. Review naming, structure, and documentation

STOP — Refactoring complete when:

  • All high-priority smells are resolved
  • All tests pass after each change
  • No behavior was changed during refactoring
  • Code is readable to a new team member

SOLID Principles

S — Single Responsibility Principle

A class/module should have one, and only one, reason to change.

Smell: A class that changes for multiple unrelated reasons. Fix: Extract responsibilities into separate classes.

O — Open/Closed Principle

Open for extension, closed for modification.

Smell: Switch statements that grow with new types. Fix: Polymorphism, strategy pattern, or plugin architecture.

L — Liskov Substitution Principle

Subtypes must be substitutable for their base types.

Smell: Subclass overrides method to throw "not supported." Fix: Restructure hierarchy; prefer composition over inheritance.

I — Interface Segregation Principle

No client should depend on methods it does not use.

Smell: Interfaces with many methods; implementors leave some as no-ops. Fix: Split into smaller, focused interfaces.

D — Dependency Inversion Principle

Depend on abstractions, not concretions.

Smell: High-level modules importing low-level modules directly. Fix: Inject dependencies via interfaces/abstract classes.


Naming Conventions

Rules

ElementConventionExample
VariablesNouns describing what they holduserCount, not n
BooleansPrefixed with is/has/can/shouldisActive, hasPermission
FunctionsVerbs describing what they docalculateTotal, fetchUsers
ConstantsUPPER_SNAKE_CASEMAX_RETRY_COUNT
ClassesPascalCase nounsUserRepository, PaymentService
InterfacesDescribe capabilitySerializable, Cacheable

Name Length Guidelines

ScopeLengthExample
Loop counters1-2 charsi, j (tiny loops only)
Lambda params1-3 chars when context clearusers.filter(u => u.active)
Local variablesShort but descriptivetotal, result
Function namesMedium, descriptivecalculateMonthlyRevenue
Class namesAs long as neededAuthenticationTokenValidator

Function Guidelines

Size and Structure

  • Functions should do one thing
  • Ideal: 5-15 lines (excluding boilerplate)
  • Maximum: 30 lines (beyond this, extract)
  • Maximum parameters: 3 (beyond this, use options object)

Guard Clauses (Early Return)

// Bad: nested conditions
function getDiscount(user) {
  if (user) {
    if (user.isPremium) {
      if (user.orderCount > 10) {
        return 0.2;
      }
    }
  }
  return 0;
}

// Good: guard clauses
function getDiscount(user) {
  if (!user) return 0;
  if (!user.isPremium) return 0;
  if (user.orderCount <= 10) return 0;
  return 0.2;
}

Error Handling Patterns

Decision Table

ApproachUse WhenExample
Result typeFunctional style, expected failuresResult<T, E> return type
Specific exceptionsOOP style, exceptional casesthrow new ValidationError(...)
Error codesC-style APIs, cross-languageReturn code + message
Option/MaybeValue may or may not existOption<User>

Result Type Pattern

type Result<T, E = Error> =
  | { success: true; data: T }
  | { success: false; error: E };

function parseConfig(raw: string): Result<Config, ParseError> {
  try {
    const config = JSON.parse(raw);
    if (!isValidConfig(config)) {
      return { success: false, error: new ParseError('Invalid config structure') };
    }
    return { success: true, data: config };
  } catch {
    return { success: false, error: new ParseError('Invalid JSON') };
  }
}

Error Handling Never List

  • Never catch and swallow errors silently
  • Never use exceptions for control flow
  • Never return null to indicate an error
  • Never log and rethrow without adding context

Complexity Metrics

RangeRisk LevelAction
1-5LowNo action needed
6-10ModerateConsider refactoring
11-20HighShould refactor
21+CriticalMust refactor

Reducing Complexity

  1. Extract complex conditions into named booleans
  2. Replace nested conditionals with guard clauses
  3. Use polymorphism instead of type checking
  4. Decompose into smaller functions
  5. Use lookup tables instead of switch/if chains

DRY Application Decision Table

SituationApply DRY?Rationale
Exact duplication of logicYesSame logic should live in one place
Three or more occurrencesYesRule of Three confirms the pattern
Two occurrences onlyWaitMay be coincidental similarity
Similar structure, different purposeNoDifferent reasons to change
Abstracting adds more complexityNoClarity over DRY

Comment Philosophy

Good Comments

TypeExample
Why (reasoning)// Use binary search because list is pre-sorted and >10K items
LegalCopyright, license headers
TODO with ticket// TODO(PROJ-123): Add rate limiting
Warning// WARNING: This is not thread-safe
Public API docsJSDoc/TSDoc for public interfaces

Bad Comments (remove and fix code instead)

TypeExample
Restating code// increment counter before counter++
Commented-out codeUse version control instead
Journal commentsUse git log instead
Closing brace comments} // end if

Anti-Patterns / Common Mistakes

Anti-PatternWhy It Is WrongCorrect Approach
Premature abstractionDRYing code that differs in intentWait for Rule of Three
God classesKnow everything, do everythingSplit by responsibility (SRP)
Feature envyMethod uses another class's data more than its ownMove method to the data owner
Stringly typed dataStrings where enums/types belongDefine proper types
Magic numbersUnclear meaning, error-proneNamed constants
Boolean trapFunction with boolean params that change behaviorUse named options or separate functions
Over-engineeringAbstractions for problems that do not existYAGNI — You Ain't Gonna Need It

Integration Points

SkillRelationship
code-reviewReview identifies code smells for clean-code to resolve
test-driven-developmentTDD ensures behavior preservation during refactoring
senior-frontendFrontend components follow clean code principles
senior-backendBackend services follow SOLID and clean architecture
performance-optimizationClean code enables easier performance optimization
systematic-debuggingClean code is easier to debug

Immutability Preferences

  • Default to const (JavaScript/TypeScript)
  • Use readonly properties and ReadonlyArray
  • Prefer spread/destructuring over mutation
  • Use immutable update patterns for state
  • Only mutate when performance profiling demands it

Skill Type

FLEXIBLE — Apply principles based on context. Not every function needs to be 5 lines; not every pattern needs to be SOLID. Use judgment and optimize for team readability over theoretical purity.

平台分布

Codex

35.34%
按下载量换算93

Claude

31.56%
按下载量换算83

Cursor

17.85%
按下载量换算47

Gemini CLI

9.14%
按下载量换算24

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills