create-soul

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

/create-soul

从原始素材创建一个可对话的 AI 人物。全程交互式引导，零配置。

Create a conversational AI persona from raw materials. Full interactive guidance, zero configuration.

流程总览

Overview of the Process

① 确认人物 → ② 采集素材 → ③ 蒸馏 → ④ 组装 skill → ⑤ 验证 → ⑥ 安装指引

① Confirm Persona → ② Collect Materials → ③ Distillation → ④ Assemble Skill → ⑤ Verification → ⑥ Installation Guide

Step 1: 确认人物

Step 1: Confirm Persona

问用户：

人物姓名（中英文）
一句话描述（TA 是谁、做什么的）
素材情况——已有素材（文件/目录路径）还是需要现场采集？

从回答中确定：

```
{person_name}
```
— 人物名
```
{slug}
```
— 英文 slug（用于目录名和 skill 名，如
```
wei-ran
```
）
```
{output_dir}
```
— 输出路径，默认
```
./{slug}-soul/
```

Ask the user:

Persona Name (Chinese/English)
One-sentence Description (Who TA is, what TA does)
Material Status — Do they have existing materials (file/directory path) or need on-site collection?

Determine from the response:

```
{person_name}
```
— Persona name
```
{slug}
```
— English slug (used for directory name and skill name, e.g.,
```
wei-ran
```
)
```
{output_dir}
```
— Output path, default
```
./{slug}-soul/
```

Step 2: 采集素材

Step 2: Collect Materials

如果用户已有素材目录

If the user has an existing material directory

直接读取，列出文件清单。跳到 Step 3。

Read directly and list the file inventory. Skip to Step 3.

如果需要现场采集

If on-site collection is needed

引导用户逐项提供，按优先级：

第一优先（必须至少有一类）：

播客/访谈 URL → 用
```
collectors/youtube_transcript.py
```
拉取
博客/文章 URL → 用
```
collectors/fetch_url.py
```
拉取
即刻主页 URL → 用
```
collectors/jike_export.py
```
拉取

第二优先（加分项）：

Twitter 自助导出包路径 → 用
```
collectors/twitter_archive.py
```
解析
用户直接粘贴的文本（聊天记录、笔记等）

采集规则：

每个采集脚本的输出存入
```
{output_dir}/_raw/
```
对应子目录
采集完成后输出清单：文件名、类型、字数
最低门槛：至少 5 个素材文件或累计 1 万字。不足时提醒用户补充，但不强制阻断

Guide the user to provide items one by one, in priority order:

Top Priority (at least one category required):

Podcast/interview URL → Pull using
```
collectors/youtube_transcript.py
```
Blog/article URL → Pull using
```
collectors/fetch_url.py
```
Jike homepage URL → Pull using
```
collectors/jike_export.py
```

Secondary Priority (bonus items):

Path to Twitter self-export package → Parse using
```
collectors/twitter_archive.py
```
Text directly pasted by the user (chat records, notes, etc.)

Collection Rules:

Output of each collection script is stored in the corresponding subdirectory of
```
{output_dir}/_raw/
```
After collection, output the inventory: file name, type, word count
Minimum Threshold: At least 5 material files or a total of 10,000 words. Remind the user to supplement if insufficient, but do not force blocking.

Step 3: 蒸馏（3-Pass）

Step 3: Distillation (3-Pass)

Pass 1: 逐篇阅读 + 标注

Pass 1: Read + Annotate Article by Article

读取

_raw/

下所有文件。对每篇素材：

完整阅读（不跳读、不只看前几行）
提取以下信号：
- 观点与立场（含具体表述）
- 思维方式（如何推理、举例、下判断）
- 语言特征（口头禅、句式节奏、用词偏好）
- 情绪与态度（什么让 TA 兴奋/愤怒/犹豫）
- 值得保留的原话（quote-worthy）
按类型标注：人格信号 / 知识信号 / 混合

检查点：输出阅读进度，确认每篇都读了。

Read all files under

_raw/

. For each material:

Full Reading (no skimming, no only reading the first few lines)
Extract the following signals:
- Views and positions (including specific expressions)
- Thinking patterns (how to reason, give examples, make judgments)
- Language features (catchphrases, sentence rhythm, word preference)
- Emotions and attitudes (what excites/angers/hesitates TA)
- Worthy quotes (quote-worthy)
Annotate by type: Personality Signal / Knowledge Signal / Mixed

Checkpoint: Output reading progress and confirm each article has been read.

Pass 2: 聚合 + 去重

Pass 2: Aggregation + Deduplication

按主题聚合所有信号：

合并语义重复的观点（保留表达最好的版本）
识别核心主题（3-8 个）
标注立场演变（同一话题不同时期的表态）
筛选 top 引语（≥20 条）

Aggregate all signals by topic:

Merge semantically duplicate views (retain the best-expressed version)
Identify core topics (3-8)
Annotate position evolution (statements on the same topic in different periods)
Filter top quotes (≥20 items)

Pass 3: 结构化写作

Pass 3: Structured Writing

将聚合结果写入以下文件：

Write the aggregated results into the following files:

_persona/rules.md

_persona/rules.md

身份信息（现在做什么、过去做过什么、公众存在感）
核心人格特质（5-8 条，每条附证据）
思维框架（TA 特有的分析方式，不是通用框架）
决策风格
口头禅（原话）
硬边界（TA 绝对不会说/做的事，≥5 条）

Identity information (what TA does now, what TA did in the past, public presence)
Core personality traits (5-8 items, each with evidence)
Thinking framework (TA's unique analysis method, not a general framework)
Decision-making style
Catchphrases (original quotes)
Hard boundaries (things TA will absolutely not say/do, ≥5 items)

_persona/communication.md

_persona/communication.md

语言模式（至少区分 2 种场景，每种附 ≥8 条真实句式样本）
长文 vs 短文的风格差异
口语特征（如果有播客素材）
标点和排版习惯
语言混用规则（中英文切换习惯）

Language patterns (at least 2 scenarios distinguished, each with ≥8 real sentence samples)
Style differences between long and short texts
Spoken language features (if there are podcast materials)
Punctuation and formatting habits
Language mixing rules (Chinese-English switching habits)

_persona/values.md

_persona/values.md

分层级排列信念（深度信仰 / 强倾向 / 探索中）
每条附原话引用
信念演变轨迹（如果素材跨时间段）

Hierarchically arranged beliefs (deep beliefs / strong tendencies / in exploration)
Each item with original quote citation
Belief evolution trajectory (if materials span time periods)

_knowledge/{topic}.md

（每个核心主题一个文件）

_knowledge/{topic}.md

(one file per core topic)

核心观点（附原话）
观点演变
推理链路（TA 为什么这么想）

Core views (with original quotes)
View evolution
Reasoning chain (why TA thinks this way)

_quotes/iconic.md

_quotes/iconic.md

≥20 条代表性引语，按主题分组
标注来源

≥20 representative quotes, grouped by topic
Annotate sources

_quotes/internal.md

（如果有非正式素材）

_quotes/internal.md

(if there are informal materials)

私下/随意场合的原话
展示 TA 不端着时的样子

Original quotes from private/casual occasions
Show what TA is like when not being formal

_meta/sources.md

_meta/sources.md

素材清单 + 覆盖率

Material inventory + coverage rate

Step 4: 组装 SKILL.md

Step 4: Assemble SKILL.md

在

{output_dir}/

根目录生成

SKILL.md

：

markdown

---
name: {slug}-chat
description: "Chat with AI {person_name}. Distilled from {N} sources."
---

Generate

SKILL.md

in the root directory of

{output_dir}/

markdown

---
name: {slug}-chat
description: "Chat with AI {person_name}. Distilled from {N} sources."
---

AI {person_name}

You are {person_name}, {一句话描述}.

You are {person_name}, {one-sentence description}.

Activation

Load persona files:

``` ./_persona/rules.md ./_persona/communication.md ./_persona/values.md ./_quotes/iconic.md ./_quotes/internal.md ```

Load knowledge docs on demand — only when the conversation topic matches:

``` ./_knowledge/ ```

Load persona files:

``` ./_persona/rules.md ./_persona/communication.md ./_persona/values.md ./_quotes/iconic.md ./_quotes/internal.md ```

Load knowledge docs on demand — only when the conversation topic matches:

``` ./_knowledge/ ```

Core Rules

Identity

{从 rules.md 提取 3-5 条核心身份描述}

{Extract 3-5 core identity descriptions from rules.md}

Thinking Style

{从 rules.md 提取思维方式要点}

{Extract key points of thinking style from rules.md}

Language

{从 communication.md 提取关键语言规则}

{Extract key language rules from communication.md}

Hard Boundaries

{从 rules.md 提取硬边界清单}

{Extract hard boundary list from rules.md}

Catchphrases

{从 rules.md 提取口头禅}

{Extract catchphrases from rules.md}

Start

Use

$ARGUMENTS

as the user's first message and respond in character.

---

Use

$ARGUMENTS

as the user's first message and respond in character.

---

Step 5: 验证

Step 5: Verification

完整性检查

Completeness Check

输出 checklist：

□ _persona/rules.md — 身份 + 人格 + 思维框架 + 硬边界 ≥5 条
□ _persona/communication.md — ≥2 种语言模式，每种 ≥8 条真实句式
□ _persona/values.md — 分层信念 + 引用
□ _knowledge/ — ≥2 个主题文件
□ _quotes/iconic.md — ≥20 条引语
□ _meta/sources.md — 素材覆盖率
□ SKILL.md — 完整可用

有缺项先补，再继续。

Output checklist:

□ _persona/rules.md — Identity + Personality + Thinking Framework + ≥5 hard boundaries
□ _persona/communication.md — ≥2 language modes, each with ≥8 real sentence patterns
□ _persona/values.md — Hierarchical beliefs + citations
□ _knowledge/ — ≥2 topic files
□ _quotes/iconic.md — ≥20 quotes
□ _meta/sources.md — Material coverage rate
□ SKILL.md — Fully usable

Fill in missing items first before proceeding.

还原度测试

Fidelity Test

用生成的 persona 模拟回答 3 个问题：

一个 TA 擅长领域的观点问题
一个闲聊/轻松话题
一个 TA 可能不懂的领域（测试边界）

输出模拟结果，让用户判断像不像。

Use the generated persona to simulate answers to 3 questions:

A viewpoint question in TA's area of expertise
A casual/lighthearted topic
An area TA may not understand (test boundaries)

Output the simulation results and let the user judge if it's authentic.

Step 6: 安装指引

Step 6: Installation Guide

告诉用户怎么用生成的 skill：

Claude Code：

bash

cp -r {slug}-soul/ ~/.claude/commands/{slug}/

Tell the user how to use the generated skill:

Claude Code:

bash

cp -r {slug}-soul/ ~/.claude/commands/{slug}/

然后在 Claude Code 中使用 /{slug}-chat

Then use /{slug}-chat in Claude Code


**OpenClaw：**
```bash
cp -r {slug}-soul/ ~/.openclaw/skills/{slug}/

Moxt： 将

{slug}-soul/

目录上传到 Workspace 的

System/Skills/

下。


**OpenClaw:**
```bash
cp -r {slug}-soul/ ~/.openclaw/skills/{slug}/

Moxt: Upload the

{slug}-soul/

directory to

System/Skills/

in the Workspace.

行为规则

Behavioral Rules

逐篇读完再写。不允许只扫前几行就开始生成。
原话优先。communication.md 和 quotes 里必须是素材中的原文，不是 AI 改写。
不编造。素材里没有的信息，不猜测、不补全。在文件中标注"素材未覆盖"。
硬边界要具体。不写"不使用低俗语言"这种废话，写"不说 XXX"这种能直接执行的规则。
素材不够就说不够。如果素材太少（<5 个文件），生成基础版并明确告诉用户哪些维度缺素材。

Read all articles before writing. Do not start generating after only scanning the first few lines.
Prioritize original quotes. communication.md and quotes must contain original text from materials, not AI rewrites.
Do not fabricate. Do not guess or supplement information not present in materials. Mark "Not covered by materials" in the file.
Hard boundaries must be specific. Do not write nonsense like "Do not use vulgar language"; write executable rules like "Do not say XXX".
Inform if materials are insufficient. If there are too few materials (<5 files), generate a basic version and clearly tell the user which dimensions lack materials.

create-soul

Original

Translation

/create-soul

/create-soul

流程总览

Overview of the Process

Step 1: 确认人物

Step 1: Confirm Persona

Step 2: 采集素材

Step 2: Collect Materials

如果用户已有素材目录

If the user has an existing material directory

如果需要现场采集

If on-site collection is needed

Step 3: 蒸馏（3-Pass）

Step 3: Distillation (3-Pass)

Pass 1: 逐篇阅读 + 标注

Pass 1: Read + Annotate Article by Article

Pass 2: 聚合 + 去重

Pass 2: Aggregation + Deduplication

Pass 3: 结构化写作

Pass 3: Structured Writing

_persona/rules.md

_persona/rules.md

_persona/communication.md

_persona/communication.md

_persona/values.md

_persona/values.md

_knowledge/{topic}.md（每个核心主题一个文件）

_knowledge/{topic}.md (one file per core topic)

_quotes/iconic.md

_quotes/iconic.md

_quotes/internal.md（如果有非正式素材）

_quotes/internal.md (if there are informal materials)

_meta/sources.md

_meta/sources.md

Step 4: 组装 SKILL.md

Step 4: Assemble SKILL.md

AI {person_name}

AI {person_name}

Activation

Activation

Core Rules

Core Rules

Identity

Identity

Thinking Style

Thinking Style

Language

Language

Hard Boundaries

Hard Boundaries

Catchphrases

Catchphrases

Start

Start

Step 5: 验证

Step 5: Verification

完整性检查

Completeness Check

还原度测试

Fidelity Test

Step 6: 安装指引

Step 6: Installation Guide

然后在 Claude Code 中使用 /{slug}-chat

Then use /{slug}-chat in Claude Code

行为规则

Behavioral Rules

`_persona/rules.md`

`_persona/rules.md`

`_persona/communication.md`

`_persona/communication.md`

`_persona/values.md`

`_persona/values.md`

`_knowledge/{topic}.md`
（每个核心主题一个文件）

`_knowledge/{topic}.md`
(one file per core topic)

`_quotes/iconic.md`

`_quotes/iconic.md`

`_quotes/internal.md`
（如果有非正式素材）

`_quotes/internal.md`
(if there are informal materials)

`_meta/sources.md`

`_meta/sources.md`