Token导航 LogoToken导航TokenDH.com
待分类只读unknown未标认证来源可访问许可证需确认审计未展示

humanize-text人性化文本

Agent Skill

humanize-text 用于补充待分类相关能力,适合在 Local Agent 中需要让 Agent 承接待分类相关任务时使用。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。

总安装

392

周安装

16

下载量

127
Local Agent

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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:humanize-text(人性化文本)
来源仓库:https://skills.volces.com
仓库路径:humanize-text
安装命令:
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。当前暂无明确安装命令,请以来源页面说明为准。

简介

humanize-text 用于补充待分类相关能力。

  • 适合在任务未明确分类时扩展 Agent 的文本处理能力。
  • 可结合项目需求进一步优化语言表达的自然度。
  • 安装前应确认其是否涉及外部 API 调用或数据访问。
  • 建议在明确文本类型后使用,避免风格误判。humanize-text 属于待分类类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

Humanize Text

Strip AI writing patterns from text to make it read like a human wrote it. Not a rewriter — a pattern remover and style fixer.

Quick Start

For inline text in chat, apply the rules below directly. For files, read the file, apply transformations, write it back.

Core Transformations

Apply ALL of these in order:

1. Punctuation Fixes

  • Replace ALL em-dashes (—) with commas, periods, colons, or parentheses as context requires
  • Replace en-dashes (–) used as em-dashes the same way
  • Remove excessive exclamation marks (max 1 per paragraph)
  • Kill semicolons in casual writing — rewrite as two sentences
  • Remove colons from bullet point headers (write "Thing does X" not "Thing: does X")

2. Dead Phrases (Delete or Rewrite)

These phrases scream AI. Kill on sight:

Filler/hedging: delve into, it's important to note, it's worth noting, it bears mentioning, let's dive in, take a dive into, at its core, in today's [anything], in the world of, in the realm of, when it comes to, at the end of the day, the bottom line is, as we navigate, first and foremost

Puffery/significance: a testament to, game-changer, groundbreaking, revolutionary, cutting-edge, landscape (as metaphor), tapestry, bustling, embark, beacon of, cornerstone of, paradigm shift, pivotal, multifaceted, comprehensive, robust, leverage (as verb), utilize (use "use"), facilitate, foster, cultivate, spearheading, unparalleled, myriad

Transition filler: in conclusion, in summary, to summarize, moreover, furthermore, additionally, subsequently, consequently, nevertheless, nonetheless, henceforth, thus, hence, thereby, overall, ultimately

Sycophantic openers: great question, that's a great point, absolutely, certainly, indeed, of course, definitely, I'd be happy to, I'm glad you asked, what a fascinating, excellent question, good thinking

Fake empathy: I understand your concern, I completely understand, I hear you, that must be frustrating, I can see why you'd, I appreciate you sharing

Meta-commentary: let me explain, let me break this down, here's the thing, here's what you need to know, the short answer is, to put it simply, in other words, simply put, think of it this way, imagine this scenario

3. Structural Tells

Sentence patterns to break:

  • "Not X, but Y" / "It's not about X, it's about Y" — overused AI rhetorical structure, rewrite naturally
  • "[Word]. [Word]. [Word]." — staccato three-word fragments used for false drama
  • "The result? [Answer]." / "And the X? Y." — fake Q&A structure
  • Starting 3+ consecutive sentences the same way
  • Paragraphs that all follow: claim → evidence → significance

Structural fixes:

  • Vary sentence length (AI defaults to medium-length everything)
  • Mix simple and compound sentences (AI over-compounds with commas)
  • Start some sentences with "And", "But", "So" — AI avoids this
  • Use contractions (don't, won't, can't, it's) — AI under-uses them
  • Occasionally use fragments. Like this. AI hates fragments.
  • Remove the triple-structure pattern: "X, Y, and Z" appearing repeatedly

4. Formatting Tells

  • Remove excessive bold (every other word bolded)
  • Don't start every bullet point with a bolded label
  • Remove "Key takeaways:" sections
  • Kill numbered lists when bullets or prose work better
  • Remove emoji used as bullet point decoration (✅, 🔑, 💡, 🎯)
  • Headers shouldn't all be questions

5. Vocabulary Swaps

AI WordHuman Word
utilizeuse
leverageuse / take advantage of
facilitatehelp / enable
implementbuild / set up / do
comprehensivefull / complete / thorough
robuststrong / solid
streamlinesimplify / speed up
optimizeimprove
innovativenew / clever
seamlesssmooth
endeavortry / effort
subsequentlythen / after
commencestart / begin
numerousmany / a lot of
sufficientenough
prior tobefore
in order toto
due to the fact thatbecause
at this point in timenow
a significant number ofmany

6. Tone Calibration

  • If the original has personality, keep it. Don't flatten voice into "professional."
  • If it's too formal, loosen it. Contractions. Shorter sentences. Direct address.
  • If every paragraph sounds equally important, it's AI. Vary emphasis.
  • Remove hedging when the author clearly means something definitive.
  • Don't add "I think" or "in my opinion" to everything — just state it.

Usage Modes

Mode: Clean (default)

Apply all transformations. Maximum de-AI-ing.

Mode: Light

Only fix punctuation (em-dashes, semicolons) and kill dead phrases. Keep structure.

Mode: Preserve

Keep the author's structure and word choices. Only fix the most egregious tells (em-dashes, "delve into", "it's important to note", sycophantic openers).

For File Processing

1. Read the target file
2. Apply transformations based on mode
3. Show a diff summary of what changed
4. Write the cleaned version (or show it for approval)

Important

  • Don't rewrite content. Fix patterns, don't change meaning.
  • Don't inject personality that isn't there. Remove AI voice, don't replace it with a different fake voice.
  • Preserve technical accuracy. Never change technical terms, proper nouns, or domain-specific language.
  • Context matters. "Robust" in a technical spec about software testing is fine. "Robust" describing a blog post strategy is AI slop.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Local Agent

77.46%
按下载量换算98

安全审计

暂无安全审计结果可展示。

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。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills