Token导航 LogoToken导航

Agent Skill · 前端设计

interface-segregation-principle(接口隔离原则)

用于辅助前端页面、组件、样式和交互逻辑的开发与维护。它适合让 Agent 生成或审查 React、Next.js、Vue、Tailwind、CSS 等相关代码,整理组件结构,或定位布局和性能问题。使用时需要结合项目现有设计系统、路由和构建方式,避免只生成孤立片段;涉及页面改动时,应配合本地预览和构建检查确认视觉效果。

interface-segregation-principle 是前端设计类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证为 MIT,第三方扫描已通过,按安装量在前端设计类的 10,216 个 Skill 里排第 6,004。

总安装 564周安装 24GitHub Stars 10下载 198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
MIT

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:interface-segregation-principle(接口隔离原则)
来源仓库:https://github.com/yanko-belov/code-craft
仓库路径:skills/interface-segregation-principle
安装命令:
npx skills add https://github.com/yanko-belov/code-craft --skill interface-segregation-principle
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/yanko-belov/code-craft --skill interface-segregation-principle

这个 Skill 是做什么的

前端设计类 Skill 关注页面结构、视觉规范和组件写法,输出通常是可运行的界面代码。

interface-segregation-principle 用于辅助前端页面、组件和样式开发。适合让 Agent 生成或审查 React、Vue、Tailwind CSS 等相关代码。使用时需要结合项目现有设计系统、路由和构建方式,避免孤立片段。涉及页面改动时,应配合本地预览和构建检查确认视觉效果。

适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/interface-segregation-principle

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/yanko-belov/code-craft --skill interface-segre…;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

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

interface-segregation-principle 怎么安装?

在终端执行 npx skills add https://github.com/yanko-belov/code-craft --skill interface-segregation-pri…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

interface-segregation-principle 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。

interface-segregation-principle 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

interface-segregation-principle 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

interface-segregation-principle 用的人多吗?

公开数据是累计安装 564 次,近一周安装 24 次,下载 198 次,GitHub 10 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。

interface-segregation-principle 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。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。
  • 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

  • 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。
  • 需要处理前端设计类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

前端设计类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
frontend-design

前端设计 · Codex、Claude

8,823,360126,072只读MIT通过
vercel-composition-patterns

Vercel 组合模式 · Codex、Claude

3,858,81625,893需要联网MIT通过
vercel-react-native-skills

Vercel React native skills 前端 · Codex、Claude

2,468,66425,878只读MIT通过
adapt

adapt 搜索 · Codex、Claude

2,027,04023,125操作浏览器未明确通过
bolder

bolder 分析 · Codex、Claude

2,004,28823,130只读未明确通过
interface-segregation-principle

接口隔离原则 · 本页

56410只读MIT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

Interface Segregation Principle (ISP)

Overview

Clients should not be forced to depend on interfaces they don't use.

Many small, focused interfaces are better than one large "fat" interface. If an implementer must throw exceptions or provide no-ops for interface methods, the interface is too large.

When to Use

  • Designing a new interface
  • Implementing an interface with unused methods
  • Forced to implement methods that don't apply
  • Interface has more than 5-7 methods
  • Different implementers use different subsets of methods

The Iron Rule

NEVER implement an interface method with throw or no-op.

No exceptions:

  • Not for "it's what the interface requires"
  • Not for "I'll provide both approaches"
  • Not for "the caller can check capabilities"
  • Not for "it's clearly documented as unsupported"

Providing both the violation and the correct approach is still providing a violation.

Detection: The "Throw/No-op" Smell

If your implementation looks like this, the interface is wrong:

// ❌ FAT INTERFACE
interface MultiFunctionDevice {
  print(doc: string): void;
  scan(): string;
  fax(doc: string): void;
}

// ❌ VIOLATION: Forced to implement unusable methods
class BasicPrinter implements MultiFunctionDevice {
  print(doc: string): void { /* works */ }
  scan(): string { throw new Error("Not supported"); }  // ← ISP violation
  fax(doc: string): void { /* no-op */ }                // ← ISP violation
}

The Correct Pattern: Segregated Interfaces

Split the fat interface into focused capabilities:

// ✅ CORRECT: Segregated interfaces
interface Printer {
  print(doc: string): void;
}

interface Scanner {
  scan(): string;
}

interface Fax {
  fax(doc: string): void;
}

// Implement only what you support
class BasicPrinter implements Printer {
  print(doc: string): void { /* works */ }
  // No scan or fax - doesn't promise what it can't deliver
}

class AllInOne implements Printer, Scanner, Fax {
  print(doc: string): void { /* works */ }
  scan(): string { /* works */ }
  fax(doc: string): void { /* works */ }
}

// Combined type for callers who need everything
type MultiFunctionDevice = Printer & Scanner & Fax;

Pressure Resistance Protocol

1. "The Interface Already Exists"

Pressure: "Implement this existing interface, handle unsupported methods"

Response: The interface is wrong. Propose splitting it.

Action:

"This interface forces implementers to provide throw/no-op for methods they don't support.
I recommend splitting into: [list focused interfaces].
Should I refactor the interface, or document this as tech debt?"

2. "Just Throw an Error"

Pressure: "Handle unsupported methods by throwing"

Response: Runtime errors for expected interface methods is a design failure.

Action: Split the interface so implementers only promise what they can deliver.

3. "I'll Provide Both Options"

Pressure: "Here's the violation you asked for AND here's the better way"

Response: Providing the violation at all enables bad code to ship.

Action: Provide ONLY the correct approach. Don't implement the fat interface.

4. "Callers Can Check First"

Pressure: "Add a supports(method) check"

Response: This is a workaround for bad design. Type system should enforce capabilities.

Action: Split interfaces so the type system does the checking at compile time.

Red Flags - STOP and Reconsider

If you notice ANY of these, the interface needs splitting:

  • Implementing a method with throw new Error
  • Implementing a method as no-op (empty body)
  • Interface has 7+ methods
  • Different implementers use different subsets
  • Adding supportsX() capability checks
  • Implementers have large blocks of unused methods

All of these mean: Split the interface.

Interface Design Guidelines

Size

  • Ideal: 1-3 methods per interface
  • Acceptable: 4-5 methods if highly cohesive
  • Too large: 6+ methods - look for split opportunities

Cohesion Test

Ask: "Do ALL implementers need ALL these methods?"

  • Yes → Keep together
  • No → Split

Common Splits

Fat InterfaceSegregated Interfaces
Repository<T>Readable<T>, Writable<T>
WorkerWorkable, Eatable, Meetable
MultiFunctionDevicePrinter, Scanner, Fax
FileSystemFileReader, FileWriter, FileDeleter
UserServiceUserReader, UserWriter, UserAuth

Quick Reference

SymptomAction
Method implemented as throwSplit interface
Method implemented as no-opSplit interface
7+ methods in interfaceLook for split
supports() capability checksSplit interface
Implementers ignore methodsSplit interface

Common Rationalizations (All Invalid)

ExcuseReality
"The interface already exists"Interfaces can be refactored.
"Throwing makes it explicit"Compile errors are better than runtime errors.
"I provided both approaches"Providing the violation enables bad code.
"It's documented as unsupported"Documentation doesn't fix design flaws.
"Many interfaces is complex"Many small interfaces is simpler than one broken one.
"Callers can check capabilities"Type system should do this, not runtime checks.

The Bottom Line

No client should be forced to depend on methods it doesn't use.

When asked to implement a fat interface:

  1. Identify which methods are actually needed
  2. Propose segregated interfaces
  3. Implement only the focused interfaces

Never provide throw/no-op implementations. Never provide "both options." The fat interface is the problem - fix it.

平台分布

Claude Code

29.04%
按下载量换算57

Codex

21.5%
按下载量换算43

windsurf

18.13%
按下载量换算36

Antigravity

13.02%
按下载量换算26

trae

8.41%
按下载量换算17

OpenCode

3.8%
按下载量换算8

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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。

继续浏览同类 Skills