Token导航 LogoToken导航TokenDH.com
研究检索只读github未标认证来源可访问clear审计异常

validate-docs-implementation验证文档实施

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

404

周安装

17

GitHub Stars

2

下载量

141
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:validate-docs-implementation(验证文档实施)
来源仓库:https://github.com/abdullahbeam/nexus-design-abdullah
仓库路径:skills/validate-docs-implementation
安装命令:
npx skills add https://github.com/abdullahbeam/nexus-design-abdullah --skill validate-docs-implementation
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/abdullahbeam/nexus-design-abdullah --skill validate-docs-implementation

简介

validate-docs-implementation 检查文档内容与实际代码实现是否一致。

  • 适用于开源项目维护、技术文档同步或内部知识库管理。
  • 可提取代码注释、README 描述并比对关键信息点。
  • 自动化检查无法覆盖所有细节,复杂逻辑仍需人工确认。
  • 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。

SKILL.md

Validate Documentation vs Implementation

Purpose: Systematically identify and fix mismatches between actual implementation code and documentation.

Load When:

  • User says: "validate docs", "check documentation", "docs vs implementation"
  • User mentions: "documentation mismatch", "docs don't match code"
  • After major code changes that may affect docs
  • During pre-release validation

Core Value: Ensures documentation accurately reflects actual implementation, preventing user confusion and wasted debugging time.


Quick Reference

What This Skill Does:

  1. ✅ Analyzes implementation code (scripts, core logic)
  2. ✅ Identifies what the code actually creates/does
  3. ✅ Searches all documentation for references to those entities
  4. ✅ Compares documented behavior vs actual behavior
  5. ✅ Fixes all mismatches systematically
  6. ✅ Provides summary of changes made

Typical Mismatches Found:

  • File names (e.g., docs say "design.md" but code creates "plan.md")
  • Folder structure (e.g., docs missing "02-resources/" directory)
  • Counts (e.g., docs say "3 files" but script creates "4 directories + 3 files")
  • Naming conventions (e.g., old vs new terminology)
  • References to deprecated features

Workflow: 5-Step Validation Process

Step 1: Analyze Implementation

Goal: Understand what the code actually does

Actions:

  1. Read the implementation file(s) identified by user
  2. Document exactly what it creates/generates/does:

- Files created (names, locations, count) - Folders created (names, structure, count) - Data structures used - Behavior patterns

  1. Create a reference list of "ground truth" facts

Example Output:

## Implementation Analysis: init_project.py

**Creates**:
- 4 directories: 01-planning/, 02-resources/, 03-working/, 04-outputs/
- 3 files in 01-planning/: overview.md, plan.md, steps.md

**Does NOT create**:
- design.md (old name)
- tasks.md (old name)
- requirements.md (never existed)

Step 2: Search Documentation

Goal: Find all references to the entities/behavior

Actions:

  1. Use Grep to search for old/mismatched terms across documentation: # Search for old file names grep -r "design\.md\|tasks\.md\|requirements\.md" 00-system/ # Search for old folder structure grep -r "02-working\|03-outputs" 00-system/
  2. Identify all files that contain references
  3. For each file, note line numbers and context

Output: List of files needing updates with specific locations


Step 3: Compare & Identify Mismatches

Goal: Categorize what needs fixing

Actions:

  1. For each documentation file found:

- Read the relevant sections - Compare against implementation truth - Categorize mismatch type: - File name mismatch: Wrong file names referenced - Structure mismatch: Wrong folder structure shown - Count mismatch: Wrong counts ("3 files" vs "4 directories + 3 files") - Missing info: Implementation creates X but docs don't mention it

  1. Prioritize fixes:

- Critical: User-facing docs, core system maps - Important: Skill workflows, reference materials - Nice-to-have: Examples, comments


Step 4: Fix All Mismatches

Goal: Update documentation to match implementation

Actions:

  1. For each file with mismatches:

- Read the file - Use Edit tool to fix each mismatch - Preserve context and formatting - Update related references (e.g., if fixing folder name, update all mentions)

  1. Common fix patterns: # File name fixes OLD: "design.md" → NEW: "plan.md" OLD: "tasks.md" → NEW: "steps.md" # Structure fixes OLD: "02-working/" → NEW: "03-working/" OLD: "03-outputs/" → NEW: "04-outputs/" ADD: "02-resources/" (was missing) # Count fixes OLD: "3 core files" NEW: "4 directories (01-planning/, 02-resources/, 03-working/, 04-outputs/) + 3 planning files"
  2. Track progress with TodoWrite:

- Mark each file as completed after fixing - Maintains visibility for user


Step 5: Verify & Generate Report

Goal: Confirm all fixes are consistent and create permanent validation record

Actions:

5A. Verification:

  1. Re-search for old terms to verify they're gone: grep -r "design\.md" 00-system/ # Should return no results
  2. Spot-check fixed files to ensure consistency

5B. Determine Report Location:

First time or if location not set:

AI: "Where should I save validation reports? Options:
    1. 04-workspace/validation-reports/ (recommended - organized, searchable)
    2. 01-memory/validation-reports/ (persistent record)
    3. Custom location (you specify)

    Or say 'skip' to not save reports."

User: [Chooses option]

AI: "Got it! I'll save reports to [location]. This will be remembered for future validations."

If location already configured (from previous validation):

AI: "Saving report to configured location: [location]"

5C. Generate & Save Report:

  1. Create validation report using template (see references/report-template.md)
  2. Generate filename: validation-{implementation}-YYYY-MM-DD.md

- Example: validation-init-project-2025-11-24.md

  1. Save to configured location
  2. Display summary to user with link to full report

Report includes:

  • Implementation analyzed
  • Files updated (with line-by-line changes)
  • Mismatch breakdown by type
  • Verification results
  • Status and next steps

Example output:

✅ Validation complete!

**Summary**:
- 6 files updated
- 23 total fixes
- 0 grep results for old terms

📄 Full report saved: [04-workspace/validation-reports/validation-init-project-2025-11-24.md](04-workspace/validation-reports/validation-init-project-2025-11-24.md)

Common Validation Scenarios

Scenario 1: Script Creates Different Files Than Documented

Symptoms:

  • User reports "docs mention design.md but I don't see it"
  • Script creates files not mentioned in docs

Process:

  1. Read script to see what it actually creates
  2. Search docs for old file names
  3. Replace all references systematically

Example: The create-project validation we just did


Scenario 2: Folder Structure Changed

Symptoms:

  • Documentation shows old folder structure
  • New folders not mentioned in docs

Process:

  1. Check actual project structure (ls)
  2. Compare against documented structure
  3. Update all structure diagrams and references

Scenario 3: Counts/Descriptions Wrong

Symptoms:

  • "Creates 3 files" but actually creates more
  • Missing important details about what's created

Process:

  1. Count actual artifacts created
  2. Find all count references in docs
  3. Update with complete, accurate descriptions

Best Practices

Be Systematic

  • Don't skip steps or assume you've found everything
  • Use grep to search exhaustively
  • Check related files (if you fix system-map.md, check framework-overview.md too)

Preserve Context

  • Don't just fix the immediate error
  • Update related references nearby
  • Ensure examples/diagrams stay consistent

Use Parallel Searches

  • When searching for multiple patterns, use multiple Grep calls in parallel
  • Saves time and provides complete picture faster

Track Progress

  • Use TodoWrite to show user what's being fixed
  • Mark items complete as you go
  • Provides transparency and confidence

Verify Fixes

  • Re-search for old terms after fixing
  • Spot-check that new terminology is consistent
  • Run validation commands if available

Integration Points

  • After code changes: Run this validation before committing
  • Pre-release: Validate all docs match current implementation
  • User reports mismatch: Use this skill to fix comprehensively
  • Documentation updates: Verify changes didn't introduce new mismatches

Success Criteria

Zero grep results for old/deprecated terms ✅ All structure diagrams match actual folder structure ✅ All counts match actual artifact counts ✅ All file references use current names ✅ User can follow docs without confusion


Remember: Documentation debt compounds quickly. Fix mismatches systematically and completely the first time!

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

Claude Code

30.93%
按下载量换算44

Antigravity

24.95%
按下载量换算35

windsurf

15.84%
按下载量换算22

Codex

11.85%
按下载量换算17

OpenCode

7.83%
按下载量换算11

Gemini CLI

3.74%
按下载量换算5

安全审计

Gen Agent Trust Hub

未通过

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。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