Token导航 LogoToken导航TokenDH.com
研究检索敏感数据github未标认证来源可访问许可证需确认审计通过

documentation文档

Agent Skill

用于辅助文档、README、Markdown、说明文和内容稿件的整理与改写。它适合让 Agent 提炼结构、补齐章节、统一术语、检查链接或把零散材料整理成可读文档。使用时应保留项目已有事实、命令和路径,不要把未确认的信息写成确定结论;涉及对外文案时,还需要控制语气,避免过度营销或夸大能力。

总安装

570

周安装

24

GitHub Stars

8

下载量

216
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:documentation(文档)
来源仓库:https://github.com/phrazzld/claude-config
仓库路径:skills/documentation
安装命令:
npx skills add https://github.com/phrazzld/claude-config --skill documentation
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/phrazzld/claude-config --skill documentation

简介

documentation 用于辅助文档、README 和内容稿件的整理与改写,适合提炼结构或统一术语。

  • 适用于文档编写和内容优化场景,需保留项目已有事实,避免过度营销。
  • 通过 npx skills add 命令从指定 GitHub 仓库安装使用。
  • 使用时应控制语气,不要把未确认的信息写成确定结论。
  • 当前分类为研究检索,支持主流 Agent 宿主平台。

SKILL.md

/documentation

Ensure this project has complete, current documentation. Audit, fix, verify—every time.

What This Does

Examines project documentation, identifies gaps and staleness, generates/updates what's needed, and verifies links and examples work. No partial modes.

Branching

Assumes you start on master/main. Before making documentation changes:

git checkout -b docs/update-$(date +%Y%m%d)

Process

1. Audit

Quick scan for what exists:

# Core docs
[ -f "README.md" ] && echo "✓ README" || echo "✗ README"
[ -f "ARCHITECTURE.md" ] || [ -f "docs/ARCHITECTURE.md" ] || [ -f "docs/CODEBASE_MAP.md" ] && echo "✓ Architecture" || echo "✗ Architecture"
[ -f "CONTRIBUTING.md" ] && echo "✓ CONTRIBUTING" || echo "✗ CONTRIBUTING"
[ -f ".env.example" ] && echo "✓ .env.example" || echo "✗ .env.example"
[ -d "docs/adr" ] || [ -d "docs/adrs" ] && echo "✓ ADR directory" || echo "✗ ADR directory"

# API docs (for libraries)
[ -f "docs/api.md" ] || [ -d "docs/api" ] && echo "✓ API docs" || echo "✗ API docs"

# Subdirectory READMEs (for complex projects)
find . -type d -name "src" -o -name "packages" -o -name "apps" 2>/dev/null | head -5 | while read d; do
  [ -f "$d/README.md" ] && echo "✓ $d/README.md" || echo "⚠ $d/README.md missing"
done

Check staleness:

# Find docs not updated in 90+ days
find . -name "*.md" -path "./docs/*" -o -name "README.md" -o -name "CONTRIBUTING.md" | while read f; do
  if [ -f "$f" ]; then
    age=$(( ($(date +%s) - $(stat -f %m "$f" 2>/dev/null || stat -c %Y "$f" 2>/dev/null)) / 86400 ))
    [ $age -gt 90 ] && echo "STALE ($age days): $f"
  fi
done

Check link validity (if lychee installed):

command -v lychee >/dev/null && lychee --offline *.md docs/**/*.md 2>/dev/null || echo "Install lychee for link checking: brew install lychee"

Spawn agent for deep review: Spawn documentation-quality-reviewer agent for comprehensive analysis if docs exist but quality is uncertain.

2. Plan

Prioritize gaps based on project type:

Every project needs:

  • README.md (what, why, quick start, setup)
  • .env.example (if any env vars used)

Applications need:

  • ARCHITECTURE.md or CODEBASE_MAP.md (system design)
  • ADRs for significant decisions

Libraries need:

  • API documentation (JSDoc → generated docs)
  • CONTRIBUTING.md
  • Examples in README

Monorepos need:

  • Root README with overview
  • Subdirectory READMEs for each package/app
  • Workspace setup instructions

Open source projects need:

  • CONTRIBUTING.md (must have)
  • CODE_OF_CONDUCT.md
  • Issue/PR templates

3. Execute

Fix everything. Don't stop at a report.

Generate README.md (if missing): Delegate to Codex with context from package.json, code structure:

codex exec --full-auto "Generate README.md for this project. \
Include: project name, description, features, prerequisites, installation, \
quick start, configuration, development commands. \
Reference package.json for scripts and dependencies. \
Follow documentation-standards skill patterns." \
--output-last-message /tmp/codex-readme.md 2>/dev/null

Generate architecture docs (if missing): Run Cartographer to create CODEBASE_MAP.md:

Invoke /cartographer to map this codebase

This creates docs/CODEBASE_MAP.md with architecture diagrams, module guide, navigation.

Generate.env.example (if missing): Scan codebase for env var usage:

# Find env var references
grep -r "process\.env\." --include="*.ts" --include="*.tsx" --include="*.js" . | \
  grep -oE "process\.env\.[A-Z_]+" | sort -u | \
  sed 's/process.env.//' | \
  awk '{print $1"="}'

Create .env.example with discovered vars and placeholder values.

Generate CONTRIBUTING.md (if missing and needed): Delegate to Codex:

codex exec --full-auto "Generate CONTRIBUTING.md for this project. \
Include: how to set up dev environment, coding standards, \
commit message format, PR process, testing requirements. \
Reference existing configs (ESLint, Prettier, Lefthook) for standards." \
--output-last-message /tmp/codex-contributing.md 2>/dev/null

Create ADR directory and template (if missing):

mkdir -p docs/adr

Create docs/adr/template.md and docs/adr/README.md per documentation-standards.

Update stale docs: For each stale doc, analyze what's changed and update:

  • Check git log for related changes since last doc update
  • Identify sections that may be outdated
  • Update or flag for manual review if significant

Subdirectory READMEs: For complex projects with src/, packages/, apps/ directories, generate module-level READMEs explaining purpose and key exports.

4. Verify

Link checking:

# Install if needed
command -v lychee >/dev/null || brew install lychee

# Check all markdown files
lychee --offline *.md docs/**/*.md README.md

Example validation: For code examples in docs, verify they're syntactically valid:

# Extract code blocks and check syntax (TypeScript)
grep -A 20 '```typescript' README.md | grep -v '```' > /tmp/example.ts
pnpm tsc --noEmit /tmp/example.ts 2>/dev/null && echo "✓ Examples valid" || echo "⚠ Example syntax issues"

Completeness check:

# Minimum viable docs
[ -f "README.md" ] && echo "✓ README exists" || echo "✗ README MISSING"
grep -q "## Installation" README.md && echo "✓ Installation section" || echo "⚠ No installation section"
grep -q "## Quick Start" README.md || grep -q "## Getting Started" README.md && echo "✓ Quick start section" || echo "⚠ No quick start section"
[ -f ".env.example" ] || ! grep -r "process.env" --include="*.ts" . -q && echo "✓ Env documented" || echo "⚠ Env vars used but no .env.example"

If any verification fails, go back and fix it.

Documentation Standards

Reference documentation-standards skill throughout. Key principles:

Comments: Why, not what. Code should be self-documenting.

README structure:

  1. What (one sentence)
  2. Why (problem it solves)
  3. Quick Start (fastest path to running)
  4. Setup (prerequisites, installation, configuration)
  5. Usage examples
  6. Contributing (if applicable)
  7. License

Architecture docs: Use Cartographer for comprehensive mapping. Generate diagrams with /beautiful-mermaid for visual architecture representation.

ADR format: Context → Decision → Consequences → Alternatives Considered

Tool Choices

Link checking: lychee (Rust, fast, offline-capable) Doc linting: Vale (if style enforcement needed) Architecture mapping: Cartographer (parallel subagent approach)

What You Get

When complete:

  • README.md with what, why, quick start, setup
  • .env.example documenting all env vars
  • Architecture docs (CODEBASE_MAP.md or ARCHITECTURE.md)
  • ADR directory with template (for significant decisions)
  • CONTRIBUTING.md (if open source or team project)
  • All links verified working
  • No stale documentation (or flagged for review)

User can:

  • Clone repo and get running via README quick start
  • Understand system architecture from docs
  • Know what env vars to configure
  • Find contributor guidelines easily
  • Trust that documentation is current

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

35.01%
按下载量换算76

Claude

30.7%
按下载量换算66

Cursor

17.75%
按下载量换算38

Gemini CLI

9.56%
按下载量换算21

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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