Token导航 LogoToken导航TokenDH.com
前端设计需要联网github未标认证来源可访问许可证需确认审计通过

bootstrapBootstrap 初始化

Agent Skill

bootstrap 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。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

4,630

周安装

191

GitHub Stars

45

下载量

1,513
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:bootstrap(Bootstrap 初始化)
来源仓库:https://github.com/buiducnhat/agent-skills
仓库路径:skills/bootstrap
安装命令:
npx skills add https://github.com/buiducnhat/agent-skills --skill bootstrap
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/buiducnhat/agent-skills --skill bootstrap

简介

bootstrap 用于从零开始搭建项目基础结构,包括目录、配置、文档和 README,也可在已有基础上增强组织。

  • 它通过多维度需求收集(如技术栈、产品定义)生成完整脚手架,并移交至下游工作流继续开发。
  • 使用时需提供项目类型、目标平台和可选参数(如 --scratch 或 --existing),系统会智能补全必要信息。
  • 安装前请确认仓库权限、维护状态,以及是否会覆盖现有文件或执行初始化命令。
  • bootstrap 属于前端设计类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Bootstrap

Collect project requirements through structured dialogue, generate foundational documentation, and scaffold the project structure — then hand off to downstream workflow skills.

Parameters

  • --scratch: Start from nothing. Full scaffold: directories, configs, docs, README.
  • --existing: Project already initialized (e.g., npm init done, framework scaffolded). Enhance with docs and structure without overwriting existing files.

Requirement Categories

Collect information across these 8 dimensions. All categories except Techstack and Product definition are optional and may fall back to smart defaults.

#CategoryWhat to collect
1TechstackLanguage, framework, runtime, package manager
2Product definitionPurpose, target users, core features, success criteria
3ArchitectureComponent structure, data flow, API design, deployment target
4RoadmapPhases, milestones, MVP scope
5ToolingLinter, formatter, test framework, CI provider
6Code standardsNaming conventions, file structure patterns, commit conventions
7Design systemUI library, styling approach — *frontend projects only*
8Auth & dataAuth method, database, ORM — *if applicable*

Workflow

Step 1: Detect Project State

  1. Check if a package.json, pyproject.toml, Cargo.toml, or similar manifest exists.
  2. Check if a docs/ directory and documentation files exist.
  3. Determine mode:

- If the user passed --scratch or nothing exists → proceed as --scratch. - If the user passed --existing or a project manifest exists → proceed as --existing.

  1. Announce the detected mode and ask for confirmation before continuing.

Step 2: Gather Requirements

Follow the Question Tool mandate for all questions. Ask one question at a time using multiple-choice options when possible.

Sequence:

  1. Techstack (required) — Ask about language/framework first. Offer common options based on context clues.
  2. Product definition (required) — Ask: what does this project do, who uses it, what are the 2-3 core features?
  3. Architecture — Ask: what are the main components, is there an API, where will it deploy?
  4. Roadmap — Ask: what is the MVP scope, are there phases?
  5. Tooling — Offer defaults from the Smart Defaults table; ask to confirm or override.
  6. Code standards — Ask about naming conventions, file layout preferences, commit style (Conventional Commits?).
  7. Design system — *Skip for non-frontend projects.* Ask about UI library and styling approach.
  8. Auth & data — *Skip if not applicable.* Ask about authentication and database/ORM if the project has these concerns.

Rules for gathering:

  • If user says "I don't know", "default", or is vague → apply Smart Defaults for the detected stack and confirm.
  • Skip categories that are clearly irrelevant (e.g., Design system for a CLI tool, Auth & data for a static site).
  • Minimum required: Techstack + Product definition. All others can use defaults.

Step 3: Confirm Requirements Summary

Present a concise summary of all collected requirements in a structured list. Ask for approval before proceeding:

  • Looks good / Confirmed → proceed to Step 4.
  • User provides corrections → update the relevant category and re-present the summary.

Step 4: Generate Documentation

Create the docs/ directory if it doesn't exist. Generate all four foundational docs populated with concrete project-specific content — no placeholders.

FileSource categories
docs/project-pdr.mdProduct definition, Roadmap
docs/architecture.mdArchitecture, Techstack, Auth & data
docs/codebase.mdGenerated from actual structure after Step 5
docs/code-standard.mdTechstack, Code standards, Tooling

Follow the same content requirements as the docs --init skill for each file.

For --existing mode: read existing docs first. Only add missing sections; do not overwrite content that is already accurate.

Step 5: Scaffold Project Structure

Create standard directories and essential config files based on the chosen techstack.

General rules:

  • Create src/, tests/ (or framework equivalent), public/ for web projects.
  • Create config files: tsconfig.json, .eslintrc.json, .prettierrc, .gitignore, etc.
  • Initialize git (git init) if not already a repository.
  • For --existing mode: only add missing files/directories. Never overwrite files that already exist.

Common scaffolds:

  • Next.js: src/app/, src/components/, src/lib/, public/, tests/
  • React (Vite): src/components/, src/hooks/, src/lib/, public/, tests/
  • Express/Node: src/routes/, src/middleware/, src/lib/, tests/
  • Python: src/<package>/, tests/, scripts/
  • CLI (Node): src/commands/, src/lib/, tests/

After scaffolding, regenerate docs/codebase.md to reflect the actual directory structure.

Step 6: Initialize Tooling

Install and configure selected tools:

  1. Run package manager install: npm install, bun install, pnpm install, pip install, etc.
  2. Install linter/formatter dev dependencies.
  3. Write or update linter config (biome.json, .eslintrc.json, ruff.toml, etc.).
  4. Write or update formatter config (biome.json, .prettierrc, etc.).
  5. Add lint/format scripts to package.json (or equivalent).
  6. If a test framework was selected, install it and create one example test file under tests/.

Skip any sub-step where the file already exists (--existing mode).

Step 7: Generate README

Create README.md with:

  • Project name and one-sentence description
  • Tech stack badges or a concise stack list
  • Prerequisites
  • Quick start (install + run commands)
  • Documentation links section pointing to all 4 docs files
  • License (if specified)

For --existing mode: update README only if it is missing or significantly incomplete.

Step 8: Handoff

Summarize everything created:

  • List all new files/directories created
  • List all docs generated
  • List all tools configured

Then recommend the next skill based on project readiness:

  • If requirements are exploratory or architecture is uncertain → recommend brainstorm
  • If the plan is clear → recommend write-plan to create the first implementation plan
  • If there's an immediate small task → recommend quick-implement

Smart Defaults

StackPackage ManagerLinterFormatterTest FrameworkStyling
Next.jsbunBiomeBiomeVitestTailwind CSS
React (Vite)bunBiomeBiomeVitestTailwind CSS
Express/NodebunBiomeBiomeVitestN/A
CLI tool (Node)bunBiomeBiomeVitestN/A
PythonuvRuffRuffpytestN/A
Rustcargoclippyrustfmtcargo testN/A

Rules

  • Never overwrite existing files in --existing mode unless the user explicitly approves.
  • Do not invent product requirements — always collect them from the user.
  • Skip irrelevant categories; don't ask questions that don't apply to the project type.
  • Generate documentation with concrete, project-specific content — no generic placeholders.
  • Keep docs/codebase.md in sync with the actual scaffolded structure.
  • If unsure about a decision, apply the smart default and confirm with the user.
  • Do not run destructive commands (e.g., rm -rf) without explicit user approval.

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

32.9%
按下载量换算498

Claude

31.83%
按下载量换算482

Cursor

18.13%
按下载量换算274

Gemini CLI

9.13%
按下载量换算138

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