Token导航 LogoToken导航

Agent Skill · 研究检索

article-writer(文章作者)

article-writer 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。

article-writer 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描有未通过项,按安装量在研究检索类的 73,675 个 Skill 里排第 59,221。

总安装 392周安装 16GitHub Stars 29下载 127核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

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:article-writer(文章作者)
来源仓库:https://github.com/mwguerra/claude-code-plugins
仓库路径:skills/article-writer
安装命令:
npx skills add https://github.com/mwguerra/claude-code-plugins --skill article-writer
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/mwguerra/claude-code-plugins --skill article-writer

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于查找、检索和筛选相关信息。article-writer 属于研究检索类 Skill,可作为该场景下的辅助能力补充。适合根据关键词、任务场景或来源线索快速定位候选结果。可结合来源仓库和原始 README 继续核验具体用法。安装前建议确认权限范围和维护状态,避免误操作。注意是否会触发联网、命令执行或文件读写。

来源仓库内路径:skills/article-writer

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/mwguerra/claude-code-plugins --skill article-w…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

article-writer 怎么安装?

在终端执行 npx skills add https://github.com/mwguerra/claude-code-plugins --skill article-writer。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

article-writer 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

article-writer 需要什么权限?

权限级别是需要联网。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。

article-writer 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

article-writer 用的人多吗?

公开数据是累计安装 392 次,近一周安装 16 次,下载 127 次,GitHub 29 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

article-writer 安全吗?

第三方扫描有未通过项,风险较高。除非你确认过源码,否则不建议直接安装到常用环境。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
article-writer

文章作者 · 本页

39229需要联网未明确未通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Article Writer

Create technical articles with companion projects and multi-language support.

Quick Start

  1. Determine author (from task or default author in database)
  2. Create folder structure (including code/ folder)
  3. Load author profile from database
  4. Load settings (including article_limits.max_words)
  5. Follow phases: Initialize → Plan → Research → Draft → Companion Project → Integrate → Review → Condense → Translate → Finalize

Workflow Overview

Plan → Research → Draft (initial) → Create Companion Project → Update Draft → Review → Condense → Translate → Finalize
                         ↑                              ↓               ↓
                         └──────── Iterate ─────────────┘               │
                                                                        ↓
                                                            (if over max_words)

Folder Structure

content/articles/YYYY_MM_DD_slug/
├── 00_context/              # author_profile.json
├── 01_planning/             # classification.md, outline.md
├── 02_research/
│   ├── sources.json         # All researched sources
│   └── research_notes.md
├── 03_drafts/
│   ├── draft_v1.{lang}.md   # Initial draft
│   └── draft_v2.{lang}.md   # After companion project integration
├── 04_review/               # checklists
├── 05_assets/images/
├── code/                    # COMPANION PROJECT
│   ├── README.md            # How to run the companion project
│   ├── src/                 # Companion project source code/files
│   └── tests/               # Tests (if applicable)
├── {slug}.{primary_lang}.md # Primary article
└── {slug}.{other_lang}.md   # Translations

Phases

Phase 0: Initialize

  • Get author, generate slug, create folder
  • Create code/ directory for companion project
  • Copy author profile to 00_context/
  • Load article_limits.max_words from settings

Phase 1: Plan

  • Classify article type
  • Create outline
  • Plan companion project type and scope
  • CHECKPOINT: Get approval

Phase 2: Research (Web Search)

  • Search official documentation
  • Find recent news (< 1 year for technical)
  • Record all sources

Phase 3: Draft (Initial)

  • Write initial draft in primary language
  • Mark places where example code will go: <!-- EXAMPLE: description -->
  • Save as 03_drafts/draft_v1.{lang}.md
Applying Author Voice

Use ALL author profile data when writing:

  1. Manual Profile Data

- tone.formality: 1=very casual, 10=very formal - tone.opinionated: 1=always hedge, 10=strong opinions - phrases.signature: Use naturally (don't overdo) - phrases.avoid: Never use these - vocabulary.use_freely: Assume reader knows these - vocabulary.always_explain: Explain on first use

  1. Voice Analysis Data (if present in voice_analysis)

- sentence_structure.avg_length: Target this sentence length - sentence_structure.variety: Match style (short/moderate/long) - communication_style: Reflect top traits in tone - characteristic_expressions: Sprinkle these naturally - sentence_starters: Use these patterns - signature_vocabulary: Prefer these words

Example:

For author with:

{
  "tone": { "formality": 4, "opinionated": 7 },
  "voice_analysis": {
    "sentence_structure": { "avg_length": 14, "variety": "moderate" },
    "communication_style": [{ "trait": "enthusiasm", "percentage": 32 }],
    "characteristic_expressions": ["na prática", "o ponto é"],
    "sentence_starters": ["Então", "O interessante é"]
  }
}

Write with:

  • Conversational but confident tone
  • Medium sentences (~14 words)
  • Enthusiastic energy
  • Occasional "na prática" and "o ponto é"
  • Some sentences starting with "Então" or "O interessante é"

Phase 4: Create Companion Project ⭐

Use Skill(companion-project-creator) for this phase

CRITICAL: Companion projects must be COMPLETE, RUNNABLE, and VERIFIED.

A Laravel companion project is a FULL Laravel installation. A Node companion project is a FULL Node project.

The companion project is NOT complete until you have actually run and tested it.

Step 1: Load Companion Project Defaults from Settings

Load settings from database first:

# View defaults for the companion project type
bun run "${CLAUDE_PLUGIN_ROOT}"/scripts/show.ts settings code

Or read JSON and extract:

// Settings are loaded from the database via show.ts or article-stats.ts
const codeDefaults = settings.companion_project_defaults.code;
// codeDefaults.scaffold_command
// codeDefaults.verification.install_command
// codeDefaults.verification.run_command
// codeDefaults.verification.test_command
Step 2: Merge with Article Overrides

If article task has companion_project field, those values override settings defaults.

Step 3: Execute Scaffold Command
# From settings.companion_project_defaults.code.scaffold_command
composer create-project laravel/laravel code --prefer-dist
Step 4: Add Article-Specific Code

Add your custom code on top of the scaffolded project:

  • Models, Controllers, Routes
  • Migrations, Seeders
  • Tests

Never create partial projects with just a few files.

Step 5: VERIFY (Mandatory) ⚠️

You MUST actually run these commands and confirm they succeed:

cd code

# 1. Install dependencies - MUST SUCCEED
composer install
# ✓ Check: No errors, vendor/ directory exists

# 2. Setup - MUST SUCCEED
cp .env.example .env
php artisan key:generate
touch database/database.sqlite
php artisan migrate
# ✓ Check: No errors

# 3. Run application - MUST START
php artisan serve &
# ✓ Check: "Server running on http://127.0.0.1:8000"
# Stop the server after confirming

# 4. Run tests - ALL MUST PASS
php artisan test
# ✓ Check: "Tests: X passed" with 0 failures

If ANY step fails:

  1. Read the error message
  2. Fix the code
  3. Re-run verification from step 1
  4. Repeat until ALL steps pass

DO NOT proceed to Phase 5 until verification passes.

Companion Project Types
Article TopicProject TypeWhat to Create
Laravel/PHP codecodeFull Laravel project via composer create-project
JavaScript/NodenodeFull Node project via npm init
PythonpythonFull Python project with venv
DevOps/DockerconfigComplete docker-compose setup
ArchitecturediagramComplete Mermaid diagrams
Project managementdocumentComplete templates + filled examples
Verification Checklist

Before proceeding to Phase 5:

  • Scaffold command executed successfully
  • All article-specific code added
  • install_command succeeded (vendor/node_modules exists)
  • run_command starts application without errors
  • test_command runs with 0 failures
  • README.md explains setup and usage
For Code Companion Projects (Laravel)
code/
├── README.md              # Setup and run instructions
├── app/
│   └── ...                # Minimal app code
├── database/
│   ├── migrations/
│   └── seeders/
├── tests/
│   └── Feature/           # Pest tests for main features
├── composer.json
└── .env.example           # SQLite by default

Standards for Laravel companion projects:

  • Use SQLite (no external DB needed)
  • Use Pest for tests
  • Include at least 2-3 tests for main features
  • Add comments referencing article: // See article section: "Rate Limiting Basics"
  • Keep dependencies minimal
  • Include setup script if complex
For Document Companion Projects
code/
├── README.md              # What the documents demonstrate
├── templates/
│   └── ...                # Reusable templates
└── examples/
    └── ...                # Filled-in examples

Phase 5: Integrate Companion Project into Draft

  • Replace <!-- EXAMPLE: --> markers with actual code snippets
  • Add file references: "See code/app/Models/Post.php"
  • Add run instructions in appropriate sections
  • Save as 03_drafts/draft_v2.{lang}.md

Phase 6: Review (Comprehensive)

Review the article as a whole:

  1. Explanation Flow

- Does the narrative flow logically? - Are concepts introduced before being used? - Does the companion project appear at the right time?

  1. Companion Project Integration

- Do code snippets match the full companion project? - Are file paths correct? - Can readers follow along?

  1. Voice Compliance

- Matches author's formality level? - Uses signature phrases appropriately? - Avoids forbidden phrases? - Opinions expressed match author's positions? - If voice_analysis present: - Sentence length matches avg_length? - Communication style traits reflected? - Characteristic expressions used (not overused)?

  1. Technical Accuracy

- Code snippets are correct? - Companion project actually runs? - Tests pass?

  1. Completeness

- All outline points covered? - Sources properly cited? - Companion project fully demonstrates topic?

CHECKPOINT: Confirm article + companion project are ready

Phase 6b: Condense (Word Limit Enforcement) ⚠️

This phase is MANDATORY if article exceeds max_words from settings.

Step 1: Check Word Count
# Count words in article (excluding frontmatter and code blocks)
# Frontmatter: lines between first --- and second ---
# Code blocks: lines between ``` markers

# Simple word count of prose content only:
sed '/^---$/,/^---$/d; /^```/,/^```$/d' draft_v2.{lang}.md | wc -w
Step 2: Load Word Limit from Settings
# Read max_words from settings
bun run "${CLAUDE_PLUGIN_ROOT}"/scripts/show.ts settings
# Or read JSON directly:
# bun run "${CLAUDE_PLUGIN_ROOT}"/scripts/show.ts settings
Step 3: Condense if Over Limit

If word count > max_words:

  1. Identify Condensation Targets (in order of priority):

- Redundant explanations of the same concept - Overly verbose transitions - Repeated caveats or disclaimers - Extended tangents not central to the topic - Excessive examples when fewer would suffice

  1. Condensation Techniques (preserve quality):

- Combine related paragraphs - Replace lengthy explanations with concise summaries - Convert verbose lists to compact bullet points - Remove filler words and phrases - Tighten sentence structure

  1. CRITICAL: Preserve Author Voice

- Keep signature phrases and expressions - Maintain the same tone (formality level) - Preserve the author's opinion style - Keep characteristic sentence structures - Retain enthusiasm/energy level from voice profile

  1. DO NOT Remove:

- Code examples or snippets (these don't count toward word limit) - Critical technical explanations - Prerequisites or setup instructions - Safety warnings or important notes - References to the companion project

Step 4: Verify Condensed Version

After condensing:

  • Word count is now ≤ max_words
  • Article still reads naturally (not choppy)
  • All key points are preserved
  • Technical accuracy maintained
  • Author voice is consistent throughout
  • Flow and narrative structure intact
Step 5: Save Condensed Draft
# Save as draft_v3 (condensed version)
# 03_drafts/draft_v3.{lang}.md

If unable to condense below max_words without quality loss:

  • Document the reason in the task
  • Note the final word count achieved
  • Flag for human review

CHECKPOINT: Article is within word limit while maintaining quality

Phase 7: Translate

  • Create versions for other languages
  • Keep code snippets unchanged
  • Translate comments in code if needed

Phase 8: Finalize

  • Write final article with frontmatter
  • Update database with:

- output_files - sources_used - companion_project info

  • Verify companion project README is complete

When to Skip Companion Projects

Only skip if a companion project makes absolutely no sense:

  • Pure opinion pieces with no actionable content
  • News/announcement summaries
  • Historical retrospectives
  • Philosophical discussions

If skipping, document in task:

{
  "companion_project": {
    "skipped": true,
    "skip_reason": "Opinion piece with no actionable code or templates"
  }
}

Companion Project README Template

# Companion Project: [Topic]

Demonstrates [what this companion project shows] from the article "[Article Title]".

## Requirements

- PHP 8.2+
- Composer
- (any other requirements)

## Setup

\`\`\`bash
composer install
cp .env.example .env
php artisan key:generate
php artisan migrate --seed
\`\`\`

## Run Tests

\`\`\`bash
php artisan test
\`\`\`

## Key Files

| File | Description |
|------|-------------|
| `app/Models/Post.php` | Demonstrates eager loading |
| `tests/Feature/QueryTest.php` | Tests N+1 detection |

## Article Reference

This companion project accompanies the article:
- **Title**: [Article Title]
- **Section**: See "Implementing Eager Loading" section

Companion Project Comments Style

<?php
// ===========================================
// ARTICLE: Rate Limiting in Laravel 11
// SECTION: Creating Custom Rate Limiters
// ===========================================

namespace App\Providers;

use Illuminate\Support\Facades\RateLimiter;

class AppServiceProvider extends ServiceProvider
{
    public function boot(): void
    {
        // Custom rate limiter for API endpoints
        // See article section: "Dynamic Rate Limits"
        RateLimiter::for('api', function (Request $request) {
            return Limit::perMinute(60)->by($request->user()?->id ?: $request->ip());
        });
    }
}

Recording Companion Project in Task

{
  "companion_project": {
    "type": "code",
    "path": "code/",
    "description": "Minimal Laravel app demonstrating rate limiting",
    "technologies": ["Laravel 11", "SQLite", "Pest 3"],
    "has_tests": true,
    "files": [
      "app/Providers/AppServiceProvider.php",
      "routes/api.php",
      "tests/Feature/RateLimitTest.php"
    ],
    "run_instructions": "composer install && php artisan test"
  }
}

References

平台分布

Codex

34.97%
按下载量换算44

Claude

27.14%
按下载量换算34

Cursor

18.87%
按下载量换算24

Gemini CLI

9.1%
按下载量换算12

安全审计

Gen Agent Trust Hub

未通过

Socket

通过

Snyk

可疑

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills