Token导航 LogoToken导航TokenDH.com
研究检索需要联网github未标认证来源可访问clear审计异常

dependency-inversion-principle依赖倒置原则

Agent Skill

dependency-inversion-principle 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。

总安装

588

周安装

25

GitHub Stars

10

下载量

206
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

3

许可证

MIT

最后核验

2026-05-01

来源状态

来源可访问

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:dependency-inversion-principle(依赖倒置原则)
来源仓库:https://github.com/yanko-belov/code-craft
仓库路径:skills/dependency-inversion-principle
安装命令:
npx skills add https://github.com/yanko-belov/code-craft --skill dependency-inversion-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 dependency-inversion-principle

简介

用于指导高内聚低耦合的架构设计原则。

  • 适用于 Codex、Claude、Cursor、Gemini CLI,支持 GitHub 安装。
  • 强调依赖抽象而非具体实现,禁止内部实例化。
  • 适用于数据库、API 等外部服务接入场景。
  • 输出为规则检查清单与反模式警示。dependency-inversion-principle 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Dependency Inversion Principle (DIP)

Overview

High-level modules should not depend on low-level modules. Both should depend on abstractions.

Classes should depend on interfaces, not concrete implementations. Dependencies should be injected, not instantiated internally.

When to Use

  • Creating any class that uses external services
  • Class uses database, email, file system, APIs
  • Writing new ConcreteClass() inside another class
  • Told "don't overcomplicate with DI"

The Iron Rule

NEVER instantiate dependencies inside a class. Always inject them.

No exceptions:

  • Not for "it's simpler this way"
  • Not for "don't overcomplicate"
  • Not for "it's just for this one service"
  • Not for "we can refactor later"

Dependency injection is not overcomplicating. It's correct design.

Detection: The "new" Smell

If a class instantiates its dependencies, it violates DIP:

// ❌ VIOLATION: Instantiating dependencies
class UserService {
  private emailService = new SendGridEmailService(); // ← DIP violation
  private db = new MySQLDatabase();                  // ← DIP violation

  async register(user: User): Promise<void> {
    await this.db.save(user);
    await this.emailService.send(user.email, 'Welcome!');
  }
}

Problems:

  • Can't test without real SendGrid/MySQL
  • Can't swap implementations
  • High-level policy coupled to low-level details

The Correct Pattern: Dependency Injection

Define interfaces, inject implementations:

// ✅ CORRECT: Depend on abstractions, inject dependencies

// Define abstractions
interface EmailService {
  send(to: string, subject: string, body: string): Promise<void>;
}

interface UserRepository {
  save(user: User): Promise<void>;
  findById(id: string): Promise<User | null>;
}

// High-level module depends on abstractions
class UserService {
  constructor(
    private emailService: EmailService,
    private userRepo: UserRepository
  ) {}

  async register(user: User): Promise<void> {
    await this.userRepo.save(user);
    await this.emailService.send(user.email, 'Welcome!', 'Thanks for joining!');
  }
}

// Low-level modules implement abstractions
class SendGridEmailService implements EmailService {
  async send(to: string, subject: string, body: string): Promise<void> {
    // SendGrid-specific implementation
  }
}

class MySQLUserRepository implements UserRepository {
  async save(user: User): Promise<void> { /* MySQL-specific */ }
  async findById(id: string): Promise<User | null> { /* MySQL-specific */ }
}

// Composition root - where dependencies are wired
const emailService = new SendGridEmailService();
const userRepo = new MySQLUserRepository();
const userService = new UserService(emailService, userRepo);

Pressure Resistance Protocol

1. "Don't Overcomplicate"

Pressure: "Just use SendGrid directly, DI is overkill"

Response: DI is not overcomplicating. It's the same amount of code, but testable and flexible.

Action: Create interface + inject. The "simple" way creates untestable code.

2. "It's Just One Dependency"

Pressure: "It only uses MySQL, DI is unnecessary"

Response: One tight coupling is still tight coupling. It still can't be tested or swapped.

Action: Inject even single dependencies.

3. "We Can Refactor Later"

Pressure: "Ship now, add DI when we need tests"

Response: You'll never refactor. The tight coupling will spread. DI takes 2 minutes now vs hours later.

Action: Use DI from the start.

4. "For Production You'd Want DI"

Pressure: Internal rationalization to provide bad code

Response: If production needs DI, write it with DI now.

Action: Don't provide "simple" versions that violate DIP.

Red Flags - STOP and Reconsider

If you notice ANY of these, you're violating DIP:

  • new ConcreteService() inside a class
  • Hardcoded connection strings/API keys in class
  • Class that can't be tested without real external services
  • import of concrete implementations used directly
  • No constructor parameters for external dependencies
  • Comments like "for production, inject this"

All of these mean: Define interface, inject dependency.

Testing Benefit

DIP enables testing without real services:

// Test with mock
class MockEmailService implements EmailService {
  public sentEmails: Array<{to: string; subject: string}> = [];

  async send(to: string, subject: string, body: string): Promise<void> {
    this.sentEmails.push({ to, subject });
  }
}

// Test
const mockEmail = new MockEmailService();
const mockRepo = new InMemoryUserRepository();
const userService = new UserService(mockEmail, mockRepo);

await userService.register({ id: '1', email: 'test@test.com', name: 'Test' });

expect(mockEmail.sentEmails).toHaveLength(1);
expect(mockEmail.sentEmails[0].to).toBe('test@test.com');

Without DIP, you'd need real SendGrid credentials to test.

Quick Reference

ViolationCorrect
this.db = new MySQL()constructor(db: Database)
this.email = new SendGrid()constructor(email: EmailService)
this.logger = new FileLogger()constructor(logger: Logger)
this.cache = new Redis()constructor(cache: Cache)
Hardcoded config in classConfig injected via constructor

Common Rationalizations (All Invalid)

ExcuseReality
"DI is overcomplicating"DI is the same code, just organized correctly.
"It's just one dependency"One coupling is still coupling.
"We'll refactor when we need tests"You won't. Write it right the first time.
"For production you'd want DI"Then write it with DI now.
"It's faster without interfaces"It's not. You type the same amount.
"Small project doesn't need DI"Small projects grow. Start right.

The Bottom Line

Depend on abstractions. Inject dependencies. Never instantiate internally.

When asked to create tight-coupled code:

  1. Define interface for the dependency
  2. Accept dependency via constructor
  3. Implement interface separately

Never provide "simple" versions that violate DIP. The "simple" version is untestable, inflexible code. Dependency injection IS the simple, correct approach.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

04

需要参考平台分布和安装热度时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

windsurf

28.59%
按下载量换算59

Claude Code

19.04%
按下载量换算39

Codex

18.05%
按下载量换算37

OpenCode

12.47%
按下载量换算26

Antigravity

6.43%
按下载量换算13

trae

3.49%
按下载量换算7

安全审计

Gen Agent Trust Hub

未通过

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。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。

来源信息

继续浏览同类 Skills