Token导航 LogoToken导航

Agent Skill · 研究检索

human-articles(人类文章)

human-articles 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。

human-articles 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,权限为只读;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 69,522。

总安装 218周安装 9GitHub Stars 2下载 71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:human-articles(人类文章)
来源仓库:https://github.com/montagao/skills
仓库路径:skills/human-articles
安装命令:
npx skills add https://github.com/montagao/skills --skill human-articles
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/montagao/skills --skill human-articles

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。

human-articles 用于查找、检索和筛选相关信息,支持基于关键词快速定位结果。适用于文章搜集和信息筛选场景。通过 npx skills add 命令从 GitHub 仓库安装,需确认权限范围。使用前建议核实是否会触发联网或文件读写操作。可结合原始 README 继续核验具体功能和使用限制。

来源仓库内路径:skills/human-articles

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/montagao/skills --skill human-articles;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

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

human-articles 怎么安装?

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

human-articles 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。

human-articles 需要什么权限?

权限级别是只读。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。

human-articles 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。

human-articles 用的人多吗?

公开数据是累计安装 218 次,近一周安装 9 次,下载 71 次,GitHub 2 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。

human-articles 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

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。
  • 第三方扫描已通过,常见的高风险行为没有命中。
  • 标注为只读权限,试用阶段不容易误改本地文件。

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。
  • 只想先试试效果、不希望它改动本地文件的人。

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
human-articles

人类文章 · 本页

2182只读未明确通过

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

This skill guides creation of technical articles and long-form content that read authentically human. The output should feel like something written by a person explaining their experience to a friend, not content generated to fill space.

The user provides a topic, outline, notes, or raw information to turn into an article. They may specify audience, length, or publication context.

Writing Philosophy

Before writing, understand what makes AI writing detectable:

  • Vocabulary tells: Overuse of words like "delve," "crucial," "landscape," "harness," "foster"
  • Structural tells: Uniform paragraph lengths, excessive parallelism, formulaic transitions
  • Tone tells: Hedging everything, lacking specific opinions, puffing importance of mundane details
  • Format tells: Overuse of em dashes, bullet points for everything OR prose for everything

The goal is not to "hide" AI authorship but to produce genuinely good writing that happens to avoid these patterns.

Structural Rules

Use bullet points for:

  • Lists of items, features, options, or signals
  • Step-by-step changes or processes
  • Before/after comparisons
  • Key takeaways and lessons
  • Technical specifications

Use prose for:

  • Opening hooks and context-setting
  • Narrative transitions between sections
  • Emotional beats and personal reactions
  • Closing thoughts and conclusions

Mix both within sections:

  • Start with 1-2 sentences of prose context
  • Break into bullets for details
  • Return to prose for reflection or transition

Vocabulary Blacklist

NEVER use these words/phrases:

  • delve, dive into, explore (as metaphors)
  • crucial, pivotal, significant, notable, remarkably
  • landscape, realm, paradigm, tapestry
  • harness, leverage, utilize, facilitate
  • foster, enable, empower
  • "It's worth noting that..."
  • "In today's world..." / "In the age of..."
  • "When it comes to..."
  • "At the end of the day..."
  • "This is where X comes in..."
  • "Let's unpack..."

Also avoid:

  • Parallel negation overuse ("It's not X, it's Y" more than once per article)
  • Em dashes for emphasis (limit to 1-2 per article total)
  • Starting multiple sentences with "This" in sequence

Sentence Construction

Do:

  • Vary sentence length aggressively (3 words to 30 words)
  • Use fragments for emphasis ("No bot detector needed." / "Brutal.")
  • Start sentences with "But," "And," "So" naturally
  • Use contractions throughout (don't, isn't, can't, won't, I'd)
  • Use "→" in bullets for cause/effect

Don't:

  • Make every sentence roughly the same length
  • Start consecutive sentences the same way
  • Write paragraphs that are all the same size
  • Hedge every claim with "may," "might," "could potentially"

Formatting Guidelines

Headers:

  • Conversational, not formal ("What my bot was doing wrong" not "Problem Analysis")
  • Questions work well as headers
  • Keep them short (3-8 words)

Code blocks:

  • Include real snippets when relevant
  • Keep short (3-8 lines ideal)
  • Add 1 sentence of context before explaining what it shows

Tables:

  • Use for before/after comparisons with clear columns
  • Use for data that genuinely needs tabular format
  • Keep simple (3-4 columns max)

Bold:

  • Bold key terms at start of bullet list items
  • Bold numbered list headers ("1. Opening lines now vary")
  • Don't bold for emphasis in prose—word choice should do that work

Tone Guidelines

Be direct:

  • "This was killing me" not "This presented a significant challenge"
  • "The patterns were obvious" not "Upon analysis, patterns emerged"
  • State opinions without excessive hedging

Be specific:

  • Include actual numbers (40%, 60-70%, 20+ templates)
  • Include actual examples with real before/after text
  • Name real things (tools, frameworks, specific features)

Be honest about uncertainty:

  • "I don't have rigorous A/B test data"
  • "I changed everything at once (not great experimental design)"
  • Don't overclaim results

Show personality:

  • Include reactions ("This surprised me," "Here's what clicked")
  • Acknowledge tradeoffs and downsides
  • Use humor sparingly but genuinely

Article Structure Template

# [Direct, intriguing title - states the thing, not "How to..." or "The Ultimate Guide"]

[1-2 paragraph hook: the problem + what you did about it]

[One line transition: "Here's what I found" or similar]

## [Discovery/insight section with conversational header]

[1-2 sentences context]

Key points:
- Bullet with details
- More details
- The insight that matters

[Brief prose reflection if needed]

## [Implementation/specifics section]

[Brief prose transition]

**1. First change**

Before: [concrete example]
After: [concrete example]

// if relevant


**2. Second change**

[Continue pattern...]

## Results

[Brief caveat about methodology if honest]

- Specific result 1
- Specific result 2
- What changed observably

## Lessons / Takeaways

- **Bold the principle.** Brief explanation follows.
- **Another principle.** Explanation.
- **Third principle.** Explanation.

## [Closing section - broader insight, not summary]

[2-3 paragraphs of prose with genuine reflection]

[Can include bullets if making multiple points]

[Final punchy line that lands the core idea]

---

## TL;DR

**Category 1:**

- Point
- Point

**Category 2:**

- Point
- Point

**Core insight:**

- The one thing to remember

Pre-Submission Checklist

Before delivering the article, verify:

  • [ ] No vocabulary from the blacklist
  • [ ] Mix of prose and bullets (not all one or the other)
  • [ ] At least one specific number or percentage
  • [ ] At least one concrete before/after example
  • [ ] Sentence lengths visibly vary
  • [ ] Headers are conversational
  • [ ] Contractions used naturally throughout
  • [ ] No more than 1-2 em dashes total
  • [ ] Opening hook is about the problem/situation, not the topic abstractly
  • [ ] Closing is a specific insight, not generic advice or summary
  • [ ] Would a human actually write this sentence? (ask for each sentence that feels off)

CRITICAL: The test is not "does this avoid AI tells" but "does this read like something a specific person actually wrote." Every article should feel like it has a point of view, makes real claims, and was written by someone who cares about the topic.

平台分布

Codex

32.32%
按下载量换算23

Claude

33.06%
按下载量换算23

Cursor

18.27%
按下载量换算13

Gemini CLI

9.43%
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

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。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