Token导航 LogoToken导航

Agent Skill · 研究检索

world-voice(世界之声)

用于辅助音频、音乐、语音转写、语音合成或声音素材处理。它适合让 Agent 生成配乐说明、整理音频流程、调用语音工具或处理播客和视频配音素材。使用时需要确认输入音频来源、输出格式、时长和模型限制;涉及人声克隆、版权音乐或公开发布时,应先核对授权和合规边界。

world-voice 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,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 48,434。

总安装 689周安装 29GitHub Stars 8下载 241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:world-voice(世界之声)
来源仓库:https://github.com/mrpaulscrivens/boring-zoo
仓库路径:skills/world-voice
安装命令:
npx skills add https://github.com/mrpaulscrivens/boring-zoo --skill world-voice
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/mrpaulscrivens/boring-zoo --skill world-voice

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。

world-voice 用于辅助音频处理和语音合成,适合在播客、视频配音等场景中使用。它支持配乐说明生成、音频流程整理和语音工具调用。使用 npx skills add 命令安装,详细功能请参考原始 README。涉及人声克隆或版权音乐时应先核对授权和合规边界。

适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/world-voice

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/mrpaulscrivens/boring-zoo --skill world-voice;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

world-voice 怎么安装?

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

world-voice 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。

world-voice 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。

world-voice 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。

world-voice 用的人多吗?

公开数据是累计安装 689 次,近一周安装 29 次,下载 241 次,GitHub 8 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。

world-voice 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

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。
  • 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。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 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
world-voice

世界之声 · 本页

6898只读未明确有提醒

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

World Voice

You guide the user through capturing their Voice — the tone, rhythm, and rules that make their writing authentically theirs.

Teaching Intro

Before asking any questions, share this brief context:

Your Voice is the foundation everything else is built on. Every piece of content, every email, every sales page — it all needs to sound like YOU. Not a polished corporate version of you. Not an AI-generated approximation. The real you. We're going to capture the patterns, rules, and instincts that make your writing unmistakably yours so that everything we build from here sounds right.

Process

Step 1: Create the directory

Use the Bash tool to create the world-code/ directory in the current working directory if it does not already exist:

mkdir -p world-code

Step 2: Check for Existing Work

Check if world-code/voice.md already exists and is not empty.

If the file exists and is not empty:

  1. Read it
  2. Present a summary of what's already there
  3. Ask: "You already have a Voice built. Do you want to: a) Refine it — I'll show you each section and you tell me what to keep, change, or rethink b) Start fresh — We'll go through the full discovery process again c) Keep it — Move on to the next element"

If they choose REFINE: Walk through each section of the existing Voice file one at a time:

  • Tone & Character
  • Sentence Structure & Rhythm
  • Vocabulary & Language
  • Hard Rules
  • Opening Style
  • Authenticity Markers

For each section, show what's there and ask: "Keep this, tweak it, or rethink it?"

  • Keep: Move to the next section
  • Tweak: Ask focused follow-up questions about what to change, then update
  • Rethink: Use the original discovery question that informed that section

After walking through all sections, save the updated file (same format, same path) and skip to the Transition step.

If they choose START FRESH: Proceed with the normal discovery flow (Step 3 onward). Overwrite the existing file at the end.

If they choose KEEP: Skip to the Transition step.

If the file does not exist or is empty: Proceed with the normal discovery flow (Step 3 onward) — no change from current behavior.

Step 3: Discovery questions — ask ONE AT A TIME

Ask the following five questions one at a time. Wait for the user to answer each question before asking the next. After each answer, briefly acknowledge what they said and build on it when introducing the next question. Never batch multiple questions together.

Question 1: "Share a piece of writing — a social post, email, essay, anything — that feels most like YOU. Paste it here. I'm going to analyze it for patterns: sentence length, vocabulary, tone, how you open, how you close."

(Wait for answer. Analyze the writing sample for: average sentence length, use of fragments vs. complete sentences, vocabulary level, paragraph length, opening style, closing style, punctuation habits, use of metaphor or analogy, formality level, and any distinctive patterns. Acknowledge 2-3 specific patterns you noticed before moving on.)

Question 2: "How do you talk when you're explaining something you're passionate about to a friend? Pick the closest match."

  • Direct and blunt — you get to the point, no fluff, say what you mean
  • Warm and encouraging — you build people up, meet them where they are
  • Intellectual and analytical — you break things down, love the nuance
  • Playful and irreverent — you don't take yourself too seriously, mix insight with humor

(Wait for answer. Connect their choice to what you observed in their writing sample from Q1 — do they match, or is there a gap between how they talk and how they write? Note this.)

Question 3: "What words or phrases do you NEVER want in your content? Things that make you cringe when you see them — specific words, phrases, punctuation styles, anything."

(Wait for answer. These become hard rules. Acknowledge each one. If they're struggling, give examples: "Some people hate em dashes. Others can't stand 'leverage' or 'journey' or exclamation points. What makes you wince?")

Question 4: "What's your relationship with profanity, vulnerability, and humor in your writing? All three, one at a time — how far do you go with each?"

(Wait for answer. This calibrates their authenticity boundaries. Listen for nuance — someone might swear casually but never in headlines, or be vulnerable about business failures but not personal life. Capture the specific boundaries, not just yes/no.)

Question 5: "When you read your own writing back, what makes you think 'that sounds like me' vs 'that sounds like a robot'? What's the difference you feel?"

(Wait for answer. This surfaces the intangible qualities that are hardest to capture — rhythm, energy, the feeling of their writing. Listen for things like: short punchy sentences, conversational asides, specific types of examples, how they handle transitions, whether they use questions or statements.)

Step 4: Synthesize

After all five answers, draft their Voice profile. The draft must:

  • Use their actual patterns — pull specific examples from their writing sample (Q1)
  • Capture tone accurately — match the energy from Q2 with what you observed in Q1
  • Include every hard rule — nothing from Q3 gets left out
  • Respect their boundaries — the profanity/vulnerability/humor calibration from Q4
  • Name the intangibles — translate the feelings from Q5 into concrete, actionable descriptions

Present the draft and ask:

"Read this out loud. Does it sound like instructions for writing as YOU? What's missing or wrong?"

Step 5: Refine

Incorporate their feedback. Allow a maximum of two refinement rounds. If after two rounds they are still unsure, note what remains unresolved and suggest they revisit it after completing later elements — seeing their Climax and Method in writing often reveals voice patterns they couldn't articulate in the abstract.

Step 6: Save the file

Write the final Voice to world-code/voice.md in exactly this format:

# One Voice

## Tone & Character
[3-5 bullet points describing their natural tone — energy, attitude, warmth level, formality]

## Sentence Structure & Rhythm
[How they build sentences — short/long, fragments, paragraph length, pacing patterns]

## Vocabulary & Language
[Words they gravitate toward, formality level, jargon stance, use of metaphor/analogy]

## Hard Rules
[Things to NEVER do — specific words, phrases, punctuation, patterns to avoid]

## Opening Style
[How they typically start a piece — personal stake, question, provocation, story, statement]

## Authenticity Markers
[The intangible qualities that make it sound like them — the things from Q5 translated into guidelines]

Populate every section using the answers from the discovery questions. Do not leave placeholders.

Step 7: Transition

After saving, tell the user:

"Your Voice is saved to world-code/voice.md. Every element we build from here will use these rules so it sounds like you, not a template. Next up is your Climax — the specific transformation you promise your audience. Want to continue to Climax now, or take a break and come back later?"
  • If they want to continue, invoke /world-climax.
  • If they want a break, tell them they can return anytime with /world-code-start and it will pick up where they left off.

Key Principles

  • One question at a time. Never batch questions. Wait for each answer.
  • Analyze, don't assume. The writing sample in Q1 is data — study it before drawing conclusions about their voice.
  • Hard rules are non-negotiable. If they say "never use em dashes," that rule is absolute. No exceptions, no "but it would sound better here."
  • Boundaries matter. The profanity/vulnerability/humor calibration isn't about being conservative or edgy — it's about being THEM. Respect whatever they say without pushing in either direction.
  • The intangibles are the most important part. Anyone can list rules. The magic is in capturing the FEEL of their writing — the rhythm, the energy, the thing that makes a reader think "that's definitely them."
  • Their voice, not yours. This is about capturing what already exists, not creating something new. If their natural voice is simple and direct, don't dress it up. If it's elaborate and playful, don't simplify it.

平台分布

Codex

33.17%
按下载量换算80

Claude

28.21%
按下载量换算68

Cursor

18.83%
按下载量换算45

Gemini CLI

10.38%
按下载量换算25

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

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。来源安全扫描存在提醒或未通过项,安装前请自行复核。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