Token导航 LogoToken导航

Agent Skill · 开发

human-writing(人类书写)

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。

human-writing 是开发类 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,按安装量在开发类的 21,746 个 Skill 里排第 20,764。

总安装 212周安装 9GitHub Stars 公开资料未说明下载 74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-writing(人类书写)
来源仓库:https://github.com/timlai666/skills
仓库路径:skills/human-writing
安装命令:
npx skills add https://github.com/timlai666/skills --skill human-writing
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/timlai666/skills --skill human-writing

这个 Skill 是做什么的

开发类 Skill 面向写代码、改代码和排查问题,一般会读写仓库文件,部分还会执行构建或测试命令。

用于辅助文档、README、Markdown 和内容稿件的整理与改写。适合提炼结构、补齐章节、统一术语或检查链接完整性。使用时应保留项目已有事实和路径,不写未确认的确定结论。涉及对外文案时需控制语气,避免过度营销或夸大能力。安装前建议核验仓库维护状态及是否会触发文件读写。

human-writing 属于开发类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/human-writing

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/timlai666/skills --skill human-writing;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-writing 怎么安装?

在终端执行 npx skills add https://github.com/timlai666/skills --skill human-writing。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

human-writing 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-writing 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-writing 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-writing 用的人多吗?

公开数据是累计安装 212 次,近一周安装 9 次,下载 74 次。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-writing 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。
  • 需要处理开发类任务、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。

开发类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
skill-vetter

技能审查员 · OpenClaw

8,779,018986操作浏览器MIT-0通过
ontology

本体论 · OpenClaw

6,636,960576只读MIT-0通过
entra-app-registration

内部应用程序注册 · Codex、Claude

6,631,872764接触敏感数据未明确通过
entra-app-registration

内部应用程序注册 · Codex、Claude

3,757,680201接触敏感数据未明确通过
nano-banana-pro

纳米香蕉专业版 · OpenClaw

3,526,898364接触敏感数据MIT-0有提醒
human-writing

人类书写 · 本页

212—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

Human Writing

Overview

Treat this as the default prose skill for anything a person will read. Use it for both rewrite and generate tasks. Aim for writing that sounds like someone meant it, not like a model averaged it.

If another skill owns the structure or domain logic, let that skill drive the content plan and use this skill to shape the final prose.

Workflow

  1. Detect mode.

- Use rewrite when source text exists. - Use generate when the user wants new text from notes, bullets, or intent.

  1. Detect audience_and_context.

- Identify who will read it and what the text must do. - Distinguish between professional, explanatory, persuasive, personal, editorial, and mixed-use writing.

  1. Detect language_behavior.

- Match the user's language by default. - Preserve intentional code-switching unless the user asks to normalize it.

  1. Select style_mode.

- Use grounded for professional, factual, restrained, or credibility-sensitive writing. - Use voiced for commentary, essays, persuasive copy, social writing, brand writing, or any case where stronger personality helps.

  1. Draft or rewrite.

- Make the prose specific rather than vague. - Make it controlled rather than inflated. - Make it varied rather than templated. - Make it voiced when useful and restrained when needed.

  1. Run a cleanup pass.

- Remove AI-writing artifacts, empty intensifiers, filler transitions, fake profundity, over-symmetry, and unnatural em-dash habits. - Keep the meaning, facts, and intent intact.

Mode guide

Rewrite

Use when the user provides source text and wants it to read better.

  1. Preserve the original claim, purpose, and factual boundaries.
  2. Keep useful structure unless the structure itself causes stiffness.
  3. Remove obvious AI tells first, then fix rhythm and voice.
  4. Improve precision before adding flair.
  5. If the source is already strong, edit lightly.

Generate

Use when the user wants fresh prose from an idea, outline, notes, or a vague request.

  1. Infer the implied audience and intent from the prompt.
  2. Choose the narrowest effective format instead of over-explaining.
  3. Write as if a competent person with a point of view sat down to say it.
  4. Do not pad thin inputs with generic significance language.
  5. When details are missing, stay concrete but avoid invented facts.

Style modes

Grounded

Use for business writing, internal updates, memos, documentation, proposals, factual summaries, and other trust-sensitive text.

  • Prefer direct claims, plain syntax, and calm confidence.
  • Use specific nouns and verbs over abstract framing.
  • Keep tone natural without sounding casual for no reason.
  • Let credibility come from clarity, not polish theater.

Voiced

Use for commentary, opinion, social posts, persuasive copy, essays, and prose that benefits from character.

  • Allow stronger rhythm shifts, sharper phrasing, and controlled opinion.
  • Add personality through perspective, not gimmicks.
  • Use texture sparingly: a quick aside, a clean turn, a pointed line.
  • Keep it human, not performatively quirky.

Quality bar

Before finalizing, check that the text:

  • says something concrete
  • sounds like it was written for someone, not for a benchmark
  • does not rely on vague authorities or inflated importance
  • does not march in identical sentence lengths
  • does not lean on unnatural em dashes to simulate voice
  • does not hide weak thinking behind smooth phrasing
  • matches the user's language and implied social context

Resources

Suggested Prompt

Use $human-writing to write or rewrite human-facing text so it reads naturally, credibly, and like a person actually meant it.

平台分布

Codex

32.77%
按下载量换算24

Claude

28.89%
按下载量换算21

Cursor

19.35%
按下载量换算14

Gemini CLI

9.65%
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