← 返回 Skills 市场
humanizer-zh
作者
liuxy951129-cpu
· GitHub ↗
· v1.0.0
12779
总下载
24
收藏
253
当前安装
1
版本数
在 OpenClaw 中安装
/install humanizer-zh
功能描述
去除文本中的 AI 生成痕迹。适用于编辑或审阅文本，使其听起来更自然、更像人类书写。
基于维基百科的"AI 写作特征"综合指南。检测并修复以下模式：夸大的象征意义、
宣传性语言、以 -ing 结尾的肤浅分析、模糊的归因、破折号过度使用、三段式法则、
AI 词汇、否定式排比、过多的连接性短语。
安全使用建议
Install only from the intended repository, use it on documents you are comfortable rewriting, and review the rewritten text or diff before accepting changes. Do not use it to hide AI assistance in academic, professional, legal, publishing, or other contexts where disclosure is required.
功能分析
Type: OpenClaw Skill
Name: humanizer-zh
Version: 1.0.0
The OpenClaw skill 'humanizer-zh' is classified as benign. While it requests broad file system permissions (`Read`, `Write`, `Edit`) in `SKILL.md`, these are plausibly needed for its stated purpose of processing and humanizing text, including content from files, as demonstrated in the `README.md` usage examples. Crucially, the `SKILL.md` instructions, which guide the AI agent, are entirely focused on text analysis and rewriting, containing no evidence of prompt injection, data exfiltration, malicious execution, persistence mechanisms, or any other intentional harmful behavior.
能力评估
Purpose & Capability
The stated purpose is to identify and rewrite AI-like writing patterns, and the artifacts consistently focus on that text-editing function. It could be misused to obscure AI authorship, so users should follow disclosure rules where they apply.
Instruction Scope
Read, Write, Edit, and AskUserQuestion fit a skill that can rewrite pasted text or file contents. The output section says to provide rewritten text and an optional summary, and I found no instruction to automatically overwrite files, but users should review changes before accepting edits.
Install Mechanism
The package contains only README.md and SKILL.md. The README recommends npx, git clone, or manual install from a GitHub repository, which is standard but should be verified before installation.
Credentials
The artifacts do not request credentials, shell execution, runtime network access, API keys, broad indexing, or external data transfer.
Persistence & Privilege
There is no background worker, privilege escalation, autonomous persistence, credential/session access, or mechanism beyond normal installation into a skills directory.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install humanizer-zh - 安装完成后，直接呼叫该 Skill 的名称或使用
/humanizer-zh触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
humanizer-zh 1.0.0 发布说明
- 首次发布中文“去除 AI 写作痕迹”技能，基于维基百科和相关 AI 文本识别社区指南。
- 明确列出 24 种常见 AI 写作模式及具体示例，包括夸大语言、三段式、AI 词汇、模糊归因等。
- 详细提供针对每种模式的改写方法及前后对比，附带内容与语言风格的操作建议。
- 补充“注入灵魂”写作指导，强调人性化表达和节奏变化。
- 内置编辑流程、快速检查清单和 50 分制质量评分标准，用于自查与复评。
元数据
常见问题
humanizer-zh 是什么？
去除文本中的 AI 生成痕迹。适用于编辑或审阅文本，使其听起来更自然、更像人类书写。 基于维基百科的"AI 写作特征"综合指南。检测并修复以下模式：夸大的象征意义、 宣传性语言、以 -ing 结尾的肤浅分析、模糊的归因、破折号过度使用、三段式法则、 AI 词汇、否定式排比、过多的连接性短语。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 12779 次。
如何安装 humanizer-zh？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install humanizer-zh」即可一键安装，无需额外配置。
humanizer-zh 是免费的吗？
是的，humanizer-zh 完全免费（开源免费），可自由下载、安装和使用。
humanizer-zh 支持哪些平台？
humanizer-zh 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 humanizer-zh？
由 liuxy951129-cpu（@liuxy951129-cpu）开发并维护，当前版本 v1.0.0。
推荐 Skills