Skip to content

ZJU-REAL/Easel / skills/builtin/skill-voice-builder

skill-voice-builder

Verified

通过结构化访谈和写作样本分析,构建创作者个人声音画像(语气/用词/节奏/风格),确保后续内容风格一致。 当用户说"声音画像""我的写作风格""风格一致""建立人设语气""voice/tone""我的表达习惯""统一文风"时使用。 和 skill-audience-profiler 的区别:本 SKILL 建创作者自己的声音画像;audience-profiler 建目标受众/粉丝画像。

FileKindSize
EASEL-META.mdmarkdown2.6 KB
SKILL.mdskill md4.7 KB
references/interview-questions.mdmarkdown2.8 KB
references/sample-content.mdmarkdown2.9 KB
references/voice-analysis-dimensions.mdmarkdown2.6 KB

EASEL-META.md

View on GitHub
# Easel SKILL 元数据

| 字段 | 值 |
|------|-----|
| **SKILL 名称** | skill-voice-builder |
| **所属层** | plan |
| **来源仓库** | charlie947/social-media-skills |
| **GitHub 地址** | https://github.com/charlie947/social-media-skills |
| **Star 数** | ~1.7k |
| **功能描述** | 结构化访谈 + 写作样本分析,生成个人声音画像(about-me.md + voice.md) |
| **SKILL.md 行数** | 119 |
| **文件总数** | 4 |
| **自包含** | 是 |
| **备注** | 原始为英文 + B2B/LinkedIn 语境(作者 Charlie Hills),已本地化为中文社媒创作者语境 |

> 收录时间: 2026-07-16
> 用途: Easel SKILL 验证测试

## 本地化说明

原 SKILL 面向 LinkedIn/B2B 场景(访谈选项为 Founder/Marketing lead/Job seekers,默认样本为英文 LinkedIn 帖,产物模板用英文 header)。本次本地化为中文社媒创作者(个人 IP)语境:

| 项 | 改动 |
|------|-----|
| interview-questions.md | 身份选项改为 知识博主/好物测评/生活方式/职场副业/垂类专家;受众/主题/渠道全部换成中文社媒语境;追问要求具体化 |
| sample-content.md | 换成中文样本:小红书笔记 + 抖音口播 + 公众号段落各一篇,体现跨平台语气差异(原为英文 LinkedIn 帖) |
| voice-analysis-dimensions.md | voice.md 模板 header 中文化;缺失信号示例改为中文社媒场景(emoji/话题标签/"家人们"等) |
| SKILL.md | about-me.md / voice.md 产物模板 header 中文化;批次与输出字段术语对齐;frontmatter 未改动 |

## 参考的 GitHub 库(致谢)

本地化过程中调研了以下声音/风格画像相关实现,吸收其访谈维度与缺失信号设计:

| 仓库 | 地址 | 借鉴点 |
|------|------|--------|
| charlie947/social-media-skills | https://github.com/charlie947/social-media-skills | 原始来源:voice-builder 的 about-me.md + voice.md 双文件结构、访谈 + 样本分析流程 |
| omeyazic/ai-voice-capture-framework | https://github.com/omeyazic/ai-voice-capture-framework | Ruben Hassid "I am just a text file" 方法:多维度访谈、对模糊回答追问逼具体 |
| angelarose210/ghostwriter | https://github.com/angelarose210/ghostwriter | 反 AI 感模式库、句子节奏/用词/口头禅等可复用画像维度 |
| danielrosehill/My-Tone-Of-Voice | https://github.com/danielrosehill/My-Tone-Of-Voice | 跨格式收集真实样本、保留原始排版做参数化分析(段落长度/句子节奏) |
| jaimeschwarz/brandvoice | https://github.com/jaimeschwarz/brandvoice | 品牌声音治理理念(阶段化构建、以"拒绝什么"定义声音) |
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall