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许可证需确认审计通过

typescript-patternsTypeScript 模式

Agent Skill

typescript-patterns 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

212

周安装

9

GitHub Stars

35

下载量

74
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

2

许可证

unknown

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:typescript-patterns(TypeScript 模式)
来源仓库:https://github.com/kazdenc/builder-skills
仓库路径:skills/typescript-patterns
安装命令:
npx skills add https://github.com/kazdenc/builder-skills --skill typescript-patterns
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/kazdenc/builder-skills --skill typescript-patterns

简介

typescript-patterns 用于查找、检索和筛选相关信息。

  • 它适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词、任务场景或来源线索快速定位候选结果。
  • 可结合来源仓库、安装命令和原始 README 继续核验具体用法。
  • 安装命令:npx skills add https://github.com/kazdenc/builder-skills --skill typescript-patterns
  • 适用于 Codex、Claude、Cursor、Gemini CLI 等宿主环境

SKILL.md

TypeScript Patterns

Apply these patterns when writing, reviewing, or refactoring TypeScript code. Prefer type safety over convenience. Catch errors at compile time, not runtime.

Type Design Principles

Default to the narrowest type that accurately represents the data. Widen only when you have a concrete reason.

PrincipleDoDon't
Narrow types`status: "active" \"inactive"`status: string
Branded types for IDstype UserId = string & {__brand: "UserId"}userId: string (mixable with any string)
Discriminated unions`{kind: "circle"; radius: number} \{kind: "rect"; w: number; h: number}`Type assertions to distinguish shapes
Readonly by defaultreadonly items: Item[]Mutable arrays unless mutation is required
Explicit return types on public APIsfunction getUser(id: UserId): Promise<User>Inferred return types on exported functions

Branded Types

Use branded types to prevent accidental mixing of structurally identical values:

type UserId = string & { readonly __brand: unique symbol };
type OrderId = string & { readonly __brand: unique symbol };

function createUserId(id: string): UserId {
  return id as UserId;
}

function getOrder(orderId: OrderId): Order { /* ... */ }

// Compile error: UserId is not assignable to OrderId
getOrder(createUserId("abc"));

Generics

Use generics when a function or type operates on a value whose type the caller determines. Don't add generics speculatively — add them when you have two or more concrete use cases.

Constraining Generics

Always constrain generics to the narrowest interface they need:

// Good — constrained to what the function actually uses
function getProperty<T extends Record<string, unknown>, K extends keyof T>(obj: T, key: K): T[K] {
  return obj[key];
}

// Bad — unconstrained, anything goes
function getProperty<T>(obj: T, key: string): unknown { /* ... */ }

Common Generic Patterns

Factory pattern — return typed instances:

function createRepository<T extends { id: string }>(collection: string) {
  return {
    findById(id: string): Promise<T | null> { /* ... */ },
    save(entity: T): Promise<void> { /* ... */ },
    delete(id: string): Promise<void> { /* ... */ },
  };
}

const users = createRepository<User>("users");

Builder pattern — chain methods with progressive type narrowing:

class QueryBuilder<T> {
  where<K extends keyof T>(field: K, value: T[K]): this { /* ... */ }
  select<K extends keyof T>(...fields: K[]): QueryBuilder<Pick<T, K>> { /* ... */ }
  build(): Query<T> { /* ... */ }
}

Inference helper — let TypeScript infer from arguments:

function createAction<TInput, TOutput>(config: {
  input: z.ZodSchema<TInput>;
  handler: (input: TInput) => Promise<TOutput>;
}) { /* ... */ }

// TInput and TOutput are inferred from usage
createAction({
  input: z.object({ name: z.string() }),
  handler: async (input) => ({ id: "1", name: input.name }),
});

Utility Types

Use built-in utility types instead of hand-rolling equivalents. Here is when to reach for each:

Utility TypeUse WhenExample
Pick<T, K>You need a subset of properties`Pick<User, "id" \"name">` for a summary view
Omit<T, K>You need everything except certain propertiesOmit<User, "password"> for a public response
Partial<T>All properties become optional (patch/update payloads)Partial<User> for an update endpoint
Required<T>All properties become requiredRequired<Config> after merging with defaults
Record<K, V>You need a typed dictionaryRecord<StatusCode, string> for a lookup table
Extract<T, U>Pull members of a union that match a conditionExtract<Event, {kind: "click"}>
Exclude<T, U>Remove members of a union that match a conditionExclude<Status, "deleted">
NonNullable<T>Strip null and undefined from a type`NonNullable<string \null>` after a null check
ReturnType<T>Get the return type of a functionReturnType<typeof fetchUser> to type a variable
Parameters<T>Get the parameter types of a function as a tupleParameters<typeof handler>[0] for the first arg

Combine them for precision: Partial<Pick<User, "name" | "email">> for an optional name-and-email update.

Error Handling

Never throw from library or shared code. Use the Result pattern to make errors part of the type signature.

Result Type

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

Apply it consistently:

type ValidationError = { field: string; message: string };
type AuthError = { code: "UNAUTHORIZED" | "FORBIDDEN"; message: string };

async function createUser(input: CreateUserInput): Promise<Result<User, ValidationError[]>> {
  const errors = validate(input);
  if (errors.length > 0) {
    return { ok: false, error: errors };
  }
  const user = await db.users.create(input);
  return { ok: true, data: user };
}

// Caller handles both paths explicitly
const result = await createUser(input);
if (!result.ok) {
  // result.error is typed as ValidationError[]
  return showErrors(result.error);
}
// result.data is typed as User
console.log(result.data.id);

Typed Error Hierarchies

Use discriminated unions for error types so callers can narrow:

type AppError =
  | { kind: "validation"; fields: { field: string; message: string }[] }
  | { kind: "not_found"; resource: string; id: string }
  | { kind: "unauthorized"; reason: string }
  | { kind: "internal"; message: string };

When to Throw

Throw only at application boundaries (request handlers, CLI entry points) where you convert Results into HTTP responses or exit codes. Never throw from utility functions, services, or domain logic.

Discriminated Unions

Use discriminated unions for any value that can be in one of several mutually exclusive states. Always use a kind, type, or status field as the discriminant.

State Machines

type RequestState<T> =
  | { status: "idle" }
  | { status: "loading" }
  | { status: "success"; data: T }
  | { status: "error"; error: AppError };

API Responses

type ApiResponse<T> =
  | { ok: true; data: T; meta: { requestId: string } }
  | { ok: false; error: { code: string; message: string; details?: unknown } };

Form States

type FormState =
  | { step: "input"; values: Partial<FormValues> }
  | { step: "review"; values: FormValues }
  | { step: "submitting"; values: FormValues }
  | { step: "complete"; result: SubmitResult }
  | { step: "error"; values: FormValues; error: string };

Exhaustiveness Checking

Always use a never check in switch statements to catch unhandled variants at compile time:

function assertNever(value: never): never {
  throw new Error(`Unhandled variant: ${JSON.stringify(value)}`);
}

function renderState(state: RequestState<User>) {
  switch (state.status) {
    case "idle": return null;
    case "loading": return <Spinner />;
    case "success": return <Profile user={state.data} />;
    case "error": return <ErrorBanner error={state.error} />;
    default: return assertNever(state);
  }
}

Module Patterns

Barrel Exports

Use barrel exports (index.ts) sparingly. They hurt tree-shaking and create circular dependency risks. Prefer direct imports.

ScenarioUse Barrel?Why
Public API of a package/libraryYesProvides a stable import surface
Internal module groupingNoImport directly from the source file
Re-exporting types onlyYesTypes are erased at compile time, no bundle cost

Re-export Types

When a module's types are needed elsewhere but its runtime code is not, re-export types separately:

// types.ts — pure types, no runtime
export type { User, CreateUserInput, UserRole };

// index.ts — runtime exports
export { createUser, getUser, deleteUser };

Isolate Side Effects

Keep side effects (database connections, env reads, logging init) in dedicated files. Don't mix side-effect code with pure business logic:

// db.ts — side effect: opens connection
export const db = createConnection(process.env.DATABASE_URL);

// user-service.ts — pure, receives db as dependency
export function createUserService(db: Database) {
  return {
    findById(id: UserId): Promise<User | null> { /* ... */ },
  };
}

Anti-Patterns

Avoid these. When you encounter them in existing code, refactor.

Anti-PatternProblemDo Instead
anyDisables all type checkingUse unknown and narrow, or define the actual type
as type assertionsLies to the compilerUse type guards or discriminated unions to narrow
Non-null assertion !Hides potential null bugsCheck for null explicitly or use optional chaining
enumNon-standard JS, poor tree-shakingUse as const objects or string literal unions
Function typeAccepts anything callableUse specific signatures: (input: T) => R
Object / {}Matches almost everythingUse Record<string, unknown> or a specific interface
namespaceLegacy pattern, poor module compatUse ES modules
Overloads for unionsVerbose, hard to maintainUse a single signature with a discriminated union parameter
@ts-ignoreSilences real errorsFix the type error or use @ts-expect-error with a comment
Index signatures everywhereToo permissiveDefine explicit interfaces for known shapes

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

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

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

Codex

37.11%
按下载量换算27

Claude

29.36%
按下载量换算22

Cursor

20.55%
按下载量换算15

Gemini CLI

9.58%
按下载量换算7

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。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills