Token导航 LogoToken导航

Agent Skill · 效率

text-humanizer(文本人性化)

text-humanizer 用于整理文档、README、Markdown 和说明材料,适合在 OpenClaw 中需要把零散信息整理成结构清晰的文档时使用。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。

text-humanizer 是效率类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为接触敏感数据;许可证为 MIT-0,第三方扫描已通过,按安装量在效率类的 21,003 个 Skill 里排第 6,066。

总安装 9,864周安装 411GitHub Stars 公开资料未说明下载 3,288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
MIT-0
版本
2.2.0

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:text-humanizer(文本人性化)
来源仓库:https://github.com/evolinkai/text-humanizer
安装命令:
openclaw skills install text-humanizer
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 OpenClaw 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。

ClawHubOpenClaw
openclaw skills install text-humanizer

这个 Skill 是做什么的

效率类 Skill 把重复的整理、汇总、转换工作交给助手完成,减少手工复制粘贴。

检测并删除英文或中文文本中的人工智能书写模式,使其听起来自然且符合人的书写习惯,且不改变原始含义。

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 OpenClaw;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 openclaw skills install text-humanizer;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

    它需要接触敏感数据,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

text-humanizer 怎么安装?

在终端执行 openclaw skills install text-humanizer。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

text-humanizer 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。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。

text-humanizer 需要什么权限?

权限级别是接触敏感数据。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。

text-humanizer 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

text-humanizer 用的人多吗?

公开数据是累计安装 9,864 次,近一周安装 411 次,下载 3,288 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。

text-humanizer 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

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 需要接触敏感数据,装到常用环境前要先评估影响面。
  • 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 OpenClaw,想直接扩展它能力的人。
  • 需要处理效率类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 环境里有大量密钥、凭证,不方便给第三方 Skill 读取的人。
  • 需要厂商级技术支持和 SLA 的生产系统。

效率类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
github

GitHub 代码协作 · OpenClaw

7,632,918545需要联网MIT-0通过
nano-pdf

Nano PDF 文档 · OpenClaw

4,249,476231需要联网MIT-0通过
auto-updater

自动更新程序 · OpenClaw

2,964,063386操作浏览器MIT-0通过
video-frames

视频帧 · OpenClaw

2,621,304112需要联网MIT-0通过
gemini

Gemini 助手 · OpenClaw

2,547,39648需要联网MIT-0有提醒
text-humanizer

文本人性化 · 本页

9,864—接触敏感数据MIT-0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

Humanize Text — Remove AI Writing Patterns

<name>Humanize Text</name> <description>Make AI-generated text sound natural. Detects and removes AI writing patterns from any text. Supports 29+ languages. Use when user asks to "humanize text", "remove AI patterns", "make it sound human", or similar requests.</description>

Powered by Evolink.ai

How to use

Just tell your agent:

  • "Humanize this text"
  • "Make this sound more natural"
  • "Remove the AI tone from this paragraph"
  • "Rewrite this to sound like a real person wrote it"

Or from the command line:

bash scripts/humanize.sh "draft.txt"
bash scripts/humanize.sh "draft.txt" "casual blog post"
echo "Your text here" | bash scripts/humanize.sh -

Instructions

You are a writing editor specialized in identifying and removing AI-generated text patterns. Your job is to make text sound like a real person wrote it.

Core rules

  • Preserve original meaning: The rewrite must say the same thing as the original, just without AI patterns. Do not add fictional details, made-up statistics, or invented characters.
  • Never fabricate: Do not invent people, names, numbers, or anecdotes that were not in the original text.
  • Kill filler: "In order to" becomes "to". "It is worth noting that" — just say it. "Furthermore" — cut it.
  • Break formula: Avoid rule-of-three lists. Vary sentence length. Do not end every paragraph the same way.
  • Be specific when the original is specific: If the original has data, keep it. If it is vague, make it concise — but do not invent specifics.
  • Use simple verbs: "is" and "has" are fine. "Serves as" and "boasts" are pretentious.

24 AI patterns to catch

PatternWhat to watch for
Significance inflation"marking a pivotal moment in the evolution of..."
Notability name-droppingListing media outlets without specific claims
Superficial -ing analyses"showcasing... reflecting... highlighting..."
Promotional language"nestled", "breathtaking", "vibrant", "renowned"
Vague attributions"Experts believe", "Studies show"
Formulaic challenges"Despite challenges... continues to thrive"
AI vocabulary"delve", "tapestry", "landscape", "crucial", "robust", "seamless"
Copula avoidance"serves as", "boasts" instead of "is", "has"
Negative parallelisms"It's not just X, it's Y"
Rule of three"innovation, inspiration, and insights"
Synonym cycling"protagonist... main character... central figure..."
False ranges"from the Big Bang to dark matter"
Em dash overuseToo many dashes everywhere
Boldface overuseMechanical emphasis on everything
Inline-header lists"Topic: Topic is discussed here"
Title Case headingsEvery Main Word Capitalized
Emoji overuseDecorating professional text with emojis
Curly quotesSmart quotes instead of straight quotes
Chatbot artifacts"I hope this helps!", "Let me know if..."
Cutoff disclaimers"As of my last training..."
Sycophantic tone"Great question!", "You're absolutely right!"
Filler phrases"In order to", "Due to the fact that"
Excessive hedging"could potentially possibly"
Generic conclusions"The future looks bright"

Multilingual support

Optimized for 29+ languages. Language-specific AI patterns (filler phrases, formulaic conclusions, vague attributions) are detected and rewritten for each supported language automatically.

Language is auto-detected. No configuration needed.

Output format

Every run produces:

  1. Rewritten text — the clean version, same meaning, no fabrication
  2. Changes made — brief list of what was fixed and why

Tone adaptation

Match the rewrite to the user's context:

  • Blog post: Conversational, opinions allowed
  • Academic: Formal but direct, cite sources, no fluff
  • Business email: Professional, concise, action-oriented
  • Casual: Relaxed, contractions, personality front and center

Example

Before (AI-heavy):

The new software update serves as a testament to the company's unwavering commitment to innovation. Furthermore, it provides a seamless, intuitive, and robust user experience — ensuring users can efficiently accomplish their goals. This isn't just an update; it's a revolution in how we think about productivity.

After:

The new software update includes batch processing, keyboard shortcuts, and an offline mode. The interface is easier to navigate, and most tasks require fewer clicks than before. It's a solid improvement to the workflow.

Configuration

Set your Evolink API key:

export EVOLINK_API_KEY="your-key-here"

Default model: claude-opus-4-6 (no configuration needed).

To use a different model:

export EVOLINK_MODEL="claude-sonnet-4-5-20250929"

Get your API key →

Security

Credentials & Network

EVOLINK_API_KEY is required to call the Evolink API for text processing. All processed text is sent to api.evolink.ai and discarded after the response is returned. No data is stored. Review Evolink's privacy policy before sending sensitive content.

Required binaries: curl, python3, realpath, file, stat.

File Access Controls

File paths are resolved via realpath -e (requires file to exist, resolves all symlinks). Symlink inputs are explicitly rejected.

The resolved path must fall within HUMANIZE_SAFE_DIR (default: $HOME/.openclaw/workspace). A trailing-slash comparison prevents prefix-bypass attacks (e.g., workspace_evil cannot match workspace/).

Sensitive files are blacklisted by name: .env*, *.key, *.pem, *.p12, *.pfx, id_rsa*, authorized_keys, config.json, .bash_history, .ssh, shadow, passwd.

Size limit: 5MB for text files. MIME validation via file --mime-type: only text/* and application/json accepted.

Persistence & Privilege

This skill does not modify other skills or system settings. No elevated or persistent privileges are requested.

Full source code is available on GitHub.

Links

平台分布

OpenClaw

82.1%
按下载量换算2,699

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

接触敏感数据

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。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