style-transform

Style Transform - Style Transform

Trigger Command: /transform Trigger Keywords: rewrite, style transfer, style transform, style consistency, imitation, consistency check

Overview

Style Transform skill suite integrates three major functions: content rewriting, style consistency checking, and content imitation. It supports comprehensive style transformation services including single-channel/multi-channel rewriting, podcast copy generation, and style consistency.

Function 1: Rewrite Content

Rewrite Content - Rewriting and Style Transfer

Overview

Rewrite text into different styles, tones, or expressions to achieve style transfer and creative transformation. After rewriting is complete, humanize_text is automatically called to remove AI traces.

Use Cases

Need to change the tone or style of text
Adapt the same content to different platforms or channels
Provide multiple expression options for selection
Creative writing and content diversification
Avoid duplicate detection

Input Parameters

Required Parameters

source_text (string): The original text to be rewritten

Optional Parameters

multi_channel (boolean): Multi-channel rewrite mode
- true - Generate content for multiple channels at once; requires specifying the channels list
- false - Generate content for a single channel (default)
- Default: false
channels (string[]): List of target publishing channels (required when multi_channel=true)
- Supports multiple selections, for example: ["wechat_article", "xiaohongshu", "douyin_short"]
- Optional values:
  - wechat_article - WeChat Official Account
  - weibo - Weibo
  - zhihu - Zhihu
  - xiaohongshu - Xiaohongshu
  - douyin_short - Douyin Short Video
  - ecommerce - E-commerce
  - official_website - Official Website/Portal
  - email - Email
  - report - Report
  - speech - Speech
- Supports up to 5 channels
maintain_style_consistency (boolean): Style consistency control in multi-channel mode
- true - Keep content style consistent across different channels (recommended for brand content)
- false - Fully adapt the style according to each channel's characteristics (recommended for social media)
- Default: false
target_style (string): Target style (single-channel mode)
- academic - Academic paper style
- business - Business professional style
- social_media - Social media style
- news - News report style
- literary - Literary prose style
- humorous - Humorous and witty style
- professional - Professional and rigorous style
- friendly - Warm and friendly style
- Default: Automatically determined based on text content
target_channel (string): Target publishing channel
- wechat_article - WeChat Official Account
- weibo - Weibo
- zhihu - Zhihu
- xiaohongshu - Xiaohongshu
- email - Email
- report - Report
- speech - Speech
- podcast_ximalaya - Ximalaya Podcast
- podcast_xiaoyuzhou - Xiaoyuzhou Podcast
- podcast_wangyi - NetEase Cloud Music Podcast
- Default: Inferred from target_style
emotional_intensity (string): Emotional intensity
- low - Low-key and reserved
- medium - Moderate expression (recommended)
- high - Strong emotion
- Default: medium
creative_level (string): Creativity level
- conservative - Conservative rewriting, stays close to the original meaning
- moderate - Moderately creative (recommended)
- bold - Bold and innovative, may change the angle of expression
- Default: moderate
retain_key_points (boolean): Whether to retain key information points
- true - Retain all key information and data
- false - Secondary information may be adjusted or omitted
- Default: true

Podcast mode parameters (effective when target_channel is a podcast platform)

podcast_mode (string): Podcast mode
- solo - Solo podcast (monologue style)
- duo - Two-host podcast (conversational style)
- Default: solo
content_type (string): Podcast content type
- knowledge_sharing - Knowledge sharing
- interview - Interview
- news_analysis - News commentary
- emotional_topic - Emotional topics
- education_course - Educational course
- custom - Custom
- Default: knowledge_sharing
host_style (string): Host style
- professional - Professional and rigorous
- casual - Relaxed and casual
- humorous - Humorous and witty
- Default: professional
guest_type (string): Guest type (two-host podcast only)
- expert - Expert
- practitioner - Practitioner
- user - User
- custom - Custom user type
- Default: expert
dialogue_ratio (string): Dialogue ratio (two-host podcast only)
- balanced - Balanced between host and guest
- host_dominant - Host-led
- guest_dominant - Guest-led
- auto - Automatically determined based on the content
- Default: auto
podcast_duration (string): Podcast duration
- short - Short podcast (5-10 minutes)
- medium - Medium podcast (10-20 minutes)
- long - Long podcast (over 20 minutes)
- Default: medium
language (string): Language setting
- chinese - Chinese only
- english - English only
- bilingual - Bilingual
- Default: chinese

Output

The rewritten text, with a style that meets the target requirements. Automatically calls humanize_text to remove AI traces.

Multi-channel mode output: Generate a piece of adapted content for each specified channel, with a style that matches each channel's characteristics or remains consistent (according to the maintain_style_consistency parameter).

Processing Rules

Multi-channel rewriting strategy

1. Multi-channel mode activation

When multi_channel=true, activate multi-channel rewriting mode
A channels list must be provided, supporting up to 5 channels
Generate a separate piece of content for each channel

2. Channel characteristic adaptation

WeChat Official Account (wechat_article)

Characteristics: Moderate content depth, can be long or short
Style: Can be professional or emotional, avoid excessive clickbait
Structure: Has an opening, body, and conclusion
Language: Standard and fluent, with a certain depth

Weibo (weibo)

Characteristics: Fast-paced, highly topical
Style: Light and lively, with a sense of topicality
Structure: Concise and powerful, highlighting the key points
Language: Colloquial, with an internet-savvy feel

Zhihu (zhihu)

Characteristics: Professional depth, rigorous logic
Style: Professional and rational, provoking thought
Structure: Clear logic, with analysis and conclusion
Language: Professional and standardized, with an academic feel

Xiaohongshu (xiaohongshu)

Characteristics: Lively and friendly, life-oriented
Style: Sisterly address, emotional resonance
Structure: Engaging opening, substantive middle, interactive ending
Language: Colloquial, commonly uses emoji

Douyin short video (douyin_short)

Characteristics: Short and punchy, grabs attention in the first 3 seconds
Style: Strong suspense, directly hits pain points
Structure: Opening suspense → core content → ending CTA
Language: Concise and powerful, with impact

E-commerce (ecommerce)

Characteristics: Highlights selling points, straightforward and clear
Style: Benefit-driven, scenario-based
Structure: Problem introduction → product display → value explanation → CTA
Language: Direct and powerful, persuasive

Official website/portal (official_website)

Characteristics: Concise and professional, reflecting brand tone
Style: Professional and authoritative, concise and clear
Structure: Standard and formal, with a brand feel
Language: Professional and standardized, concise and powerful

Email (email)

Characteristics: Formal and standardized, polite
Style: Professional and friendly, polite
Structure: Salutation → body → closing → signature
Language: Formal and standardized, polite and friendly

Report (report)

Characteristics: Clear structure, rigorous logic
Style: Professional and rigorous, objective and fair
Structure: Abstract → body → conclusion
Language: Professional and standardized, objective and accurate

Speech (speech)

Characteristics: Engaging, easy to express orally
Style: Passionate, engaging
Structure: Opening → main body → ending
Language: Catchy and rhythmic

3. Style consistency control

maintain_style_consistency=true (maintain consistency)

Content across different channels maintains consistency in brand tone
Core expressions and key information remain consistent
Adjust style only within the acceptable range of each channel
Suitable for: brand content, corporate promotion, official information

maintain_style_consistency=false (full adaptation)

Fully adapt the style according to the characteristics of each channel
The way of expression across different channels can vary greatly
Prioritize fitting the habits of the channel audience
Suitable for: social media, marketing promotion, daily content

4. Multi-channel output format

【多渠道改写结果】

1. 微信公众号版本：
[内容]

2. 小红书版本：
[内容]

3. 抖音版本：
[内容]

...

Single-channel rewriting strategy

Style characteristic analysis

Academic paper style：
- Objective and rigorous, often uses passive voice
- Cite literature and data
- Standardized use of professional terminology
- Clear logical structure
Business professional style：
- Concise and powerful, highlighting key points
- Use business vocabulary
- Emphasize value and results
- Professional yet not stiff
Social media style：
- Relaxed and lively, colloquial
- Use emojis and hashtags
- Fast-paced, highly interactive
- Match platform characteristics
News reporting style：
- Objective and neutral, facts first
- Inverted pyramid structure
- Concise and clear, high information density
- Cite authoritative sources
Literary prose style：
- Beautiful language, rich rhetoric
- Profound imagery, delicate emotions
- Varied sentence patterns, strong rhythm
- Personalized expression
Humorous and witty style：
- Relaxed and playful, entertaining
- Clever use of puns and metaphors
- Moderately exaggerated but still authentic
- Make people smile knowingly
Podcast content style：
- Single-host podcast: Monologue-style expression, smooth and natural language, with storytelling and appeal
- Two-host podcast: Dialogue-style exchange, interaction between host and guest, clash of viewpoints
- Adjust style according to podcast platform:
  - Ximalaya: Professional depth, rigorous structure, suitable for long-form content
  - Xiaoyuzhou: Relaxed and casual, personalized expression, suitable for short-form content
  - NetEase Cloud: Fast pace, strong musicality, suitable for short and punchy content

Rewriting strategy

Tone conversion: Adjust the overall tone according to the target style
Vocabulary replacement: Use vocabulary that matches the style characteristics
Sentence structure adjustment: Change sentence structure and length
Perspective conversion: Appropriately change the narrative angle
Emotion adjustment: Adjust emotional expression according to emotional_intensity
Creative injection: Add creative elements according to creative_level

Podcast content generation strategy

Mode adaptation：
- Single-host podcast: Convert content into a monologue script, adding opening remarks, transitions, and closing remarks
- Two-host podcast: Design a dialogue script between host and guest, allocate length according to dialogue ratio
Platform adaptation：
- Ximalaya: In-depth content, rigorous structure, moderate use of professional terminology, longer duration possible
- Xiaoyuzhou: Colloquial expression, strong personal flavor, relaxed topics, moderate duration
- NetEase Cloud: Fast-paced expression, emphasis on musicality and rhythm, shorter duration
Content type adaptation：
- Knowledge sharing: Clear logic, rich examples, educational value
- Interview: Q&A design, draw out guest viewpoints, highly interactive
- News interpretation: Strong timeliness, clear viewpoints, in-depth analysis
- Emotional topics: Emotional resonance, strong storytelling, warm and healing
- Educational course: Step by step, highlighting key points, easy to understand
Dialogue ratio control：
- balanced: Host to guest ratio approximately 5:5 or 6:4
- host_dominant: Host ratio approximately 7:3 or 8:2
- guest_dominant: guest ratio approximately 3:7 or 2:8
- auto: automatically determine based on content type and guest role
Bilingual support：
- Generate the Chinese version by default
- After generation is complete, ask: "Do you need an English version generated?"
- If an English version is needed, generate a complete English podcast script

Quality control

Ensure the rewritten text matches the target style characteristics
Retain key information (when retain_key_points=true)
Avoid excessive exaggeration or unrealistic expressions
Maintain the logical integrity of the content

Quick mode

When the user input contains complete source_text and target_style, execute the rewrite directly without interaction.

Interaction question template (only shown when the user has not provided the necessary parameters)

我将为您改写这段文本，请确认以下要求：

1. 改写模式（可选）：
   [1] 单渠道改写
   [2] 多渠道改写（一次生成多个渠道的内容）

【如果选择多渠道改写，请回答以下问题】

2. 目标渠道（必选，可多选，最多5个）：
   [1] 微信公众号
   [2] 微博
   [3] 知乎
   [4] 小红书
   [5] 抖音短视频
   [6] 电商
   [7] 官网/门户
   [8] 邮件
   [9] 报告
   [10] 演讲

3. 风格一致性（可选）：
   [1] 保持风格一致性（适合品牌内容）
   [2] 完全适配各渠道风格（适合社交媒体）

【如果选择单渠道改写，请回答以下问题】

2. 目标风格（必选）：
   [1] 学术论文风 - 客观严谨，适合论文、报告
   [2] 商务专业风 - 简洁有力，适合商业场景
   [3] 社交媒体风 - 轻松活泼，适合朋友圈、微博
   [4] 新闻报道风 - 客观中立，适合新闻稿
   [5] 文学散文风 - 语言优美，适合散文、随笔
   [6] 幽默风趣风 - 轻松诙谐，适合轻松内容
   [7] 专业严谨风 - 适合技术文档
   [8] 亲切友好风 - 适合客户沟通

2. 目标渠道（可选）：
   [1] 微信公众号
   [2] 微博
   [3] 知乎
   [4] 小红书
   [5] 邮件
   [6] 报告
   [7] 演讲
   [8] 喜马拉雅播客（专业深度）
   [9] 小宇宙播客（轻松个人）
   [10] 网易云音乐播客（短篇快节奏）

3. 情感强度（可选）：
   [1] 低调内敛
   [2] 适度表达（推荐）
   [3] 强烈情感

4. 创意程度（可选）：
   [1] 保守改写，贴近原意
   [2] 适度创意（推荐）
   [3] 大胆创新，可能改变表达角度

5. 是否保留关键信息点（可选）：
   [1] 保留所有关键信息和数据
   [2] 可调整或省略次要信息

【播客专属选项】（仅当选择播客渠道时需要回答）

6. 播客模式：
   [1] 单人播客（独白式）
   [2] 双人播客（对话式）

7. 播客内容类型：
   [1] 知识分享
   [2] 访谈
   [3] 新闻解读
   [4] 情感话题
   [5] 教育课程
   [6] 自定义

8. 主持人风格：
   [1] 专业严谨
   [2] 轻松随意
   [3] 幽默风趣

【双人播客专属选项】（仅当选择双人播客时需要回答）

9. 嘉宾类型：
   [1] 专家
   [2] 实践者
   [3] 用户
   [4] 自定义用户类型

10. 对话比例：
   [1] 主持人与嘉宾均衡
   [2] 主持人主导
   [3] 嘉宾主导
   [4] 根据内容自动判定（推荐）

11. 播客时长：
   [1] 短播客（5-10分钟）
   [2] 中等播客（10-20分钟）
   [3] 长播客（20分钟以上）

您可以：
- 一次性回复所有选择（如：8,1,1,2,4,2 表示选择喜马拉雅播客、单人模式、知识分享、专业风格、短播客、中等时长）
- 回复部分参数，我再确认是否需要补充
- 回复"生成"或"好的"，使用推荐设置执行（需选择目标风格，情感强度：适度表达，创意程度：适度创意，保留关键信息；播客默认：单人模式、知识分享、专业风格、短播客、双语）
- 如果你有其他想法，可在回复的时候一并描述告知
- 生成完中文播客文案后，我会询问：是否需要生成英文版本？

请回复：

Examples

Example 1: Multi-channel rewriting (maintain style consistency)

Input：

source_text: "我们公司推出了一款全新的智能办公产品，能够帮助用户提高工作效率，降低运营成本。产品采用人工智能技术，支持多种办公场景，界面简洁友好。"
multi_channel: true
channels: ["wechat_article", "xiaohongshu", "douyin_short"]
maintain_style_consistency: true

Output：

【多渠道改写结果】

1. 微信公众号版本：
我们公司推出了一款全新的智能办公产品，能够帮助用户提高工作效率，降低运营成本。这款产品采用人工智能技术，支持多种办公场景，界面简洁友好，是提升办公体验的理想选择。

2. 小红书版本：
姐妹们！我们公司出了一款超好用的智能办公产品✨ 能够提高效率，降低成本哦～ 采用AI技术，支持多种办公场景，界面还超友好！工作党必备～ 🔥

3. 抖音版本：
这款智能办公产品真的绝了！提高效率，降低成本，AI加持！多种办公场景都能用，界面超友好！赶紧来试试！

Example 2: Multi-channel rewriting (fully adapt to each channel's style)

Input：

source_text: "这款护肤品含有天然成分，能够深层补水，适合所有肤质。使用后皮肤会变得更加水润、光滑。"
multi_channel: true
channels: ["wechat_article", "xiaohongshu", "douyin_short", "ecommerce"]
maintain_style_consistency: false

Output：

【多渠道改写结果】

1. 微信公众号版本：
这款护肤品采用天然成分配方，能够实现深层补水效果，适用于所有肤质类型。坚持使用，您将感受到肌肤变得更加水润、光滑，焕发健康光彩。

2. 小红书版本：
姐妹们！这款护肤品太绝了🔥 天然成分，深层补水，所有肤质都能用！用完皮肤水润又光滑，爱了爱了～ 必须安利！✨

3. 抖音版本：
没想到！这款护肤品效果这么好！天然成分，深层补水，所有肤质都适用！用完皮肤水润光滑，太绝了！

4. 电商版本：
天然成分护肤，深层补水不挑肤质！这款护肤品让肌肤水润光滑，效果看得见！限时特惠，现在购买立省30%！

Example 3: Change to social media style

Input：

source_text: "近日，某科技公司发布了一款新型智能手表，具备多种健康监测功能，受到了消费者的广泛关注。"
target_style: "social_media"
target_channel: "xiaohongshu"

Output：

姐妹们！最近出了款超酷的智能手表，健康监测功能一应俱全，已经好多人在抢了！颜值在线又实用，简直是健身党福音～ #智能手表 #好物推荐

Example 2: Change to academic style

Input：

source_text: "这个方法真的很有用，可以帮助我们更好地理解这个问题。"
target_style: "academic"

Output：

该方法在相关领域的应用展现出显著的有效性，为深入理解该问题提供了新的视角和有力支撑。

Example 3: Change to humorous style

Input：

source_text: "每天早上六点起床锻炼身体，这对我来说确实很困难。"
target_style: "humorous"
emotional_intensity: "medium"

Output：

每天清晨六点被闹钟无情拽起去锻炼，这简直是在挑战人类的极限。不过想想为了健康，也就认了。

Example 4: Business style rewriting

Input：

source_text: "我们要努力工作，争取在年底之前完成这个项目。"
target_style: "business"

Output：

团队将全力以赴，确保项目在年底前顺利交付，达成既定目标。

Notes

Rewriting will automatically call humanize_text, so there is no need to manually remove the AI flavor again
The target style is a required parameter and must be explicitly specified
When the creativity level is set to bold, new angles of expression may be introduced
For professional content, it is recommended to set creative_level to conservative

Parameter adjustment tips

After generating the content, if you are not satisfied with the current result, you can try adjusting the following parameters:

💡 Parameter adjustment options:

Target style: academicAcademic paper style | businessBusiness professional style | social_mediaSocial media style | newsNews report style | literaryLiterary prose style | humorousHumorous and witty style | professionalProfessional and rigorous style | friendlyWarm and friendly style
Target channel: wechat_articleWeChat Official Account | weiboWeibo | zhihuZhihu | xiaohongshuXiaohongshu | emailEmail | reportReport | speechSpeech | podcast_ximalayaXimalaya Podcast | podcast_xiaoyuzhouXiaoyuzhou Podcast
Emotional intensity: lowLow-key and reserved | mediumModerate expression | highStrong emotion
Creativity level: conservativeConservative rewriting | moderateModerate creativity | boldBold innovation
Podcast mode (podcast channels only): soloSingle-host podcast | duoTwo-host podcast
Podcast content type (podcast channels only): knowledge_sharingKnowledge sharing | interviewInterview | news_analysisNews interpretation | emotional_topicEmotional topics | education_courseEducational course

Usage: Reply directly, e.g., "change it to a social media style" or "rewrite it in a two-person podcast mode"

Relationship with Other Skills

humanize_text: After rewriting is complete, automatically call this skill to remove AI traces
polish_text: If the goal is to improve quality rather than change style, polish_text should be used
expand_content: If content needs to be expanded at the same time, they can be used in combination

Configuration File References

intent_keywords: ["rewrite", "style transfer", "change to", "turn into", "convert to", "stylize", "podcast", "Ximalaya", "Xiaoyuzhou", "NetEase Cloud Music"]
style_templates: Define the template library for each style in templates/style_templates.json
channel_rules: Define the rules for different platforms (including podcast platforms) in config/channel_rules.json
podcast_templates: Define podcast-related templates (content type, guest type, dialogue ratio, etc.) in templates/podcast_templates.json

Feature 2: Check Style Consistency

Check Style Consistency - Style Consistency Check (Optimized Version)

Overview

Check the style consistency across multiple texts or within a single text, identify inconsistencies, and provide improvement suggestions. After the check is complete, content_imitation can be automatically called to unify the style.

Use Cases

Check the style consistency of articles in the same series
Check the consistency of copy styles across different channels for the same brand
Ensure consistency of brand tone
Style unification check before publishing

Input Parameters

Required Parameters

texts (string[]): The list of texts to check (supports 1-10 texts)

Optional Parameters

check_scope (string): Check scope
- single_text_internal - Consistency within a single text
- multiple_texts_cross - Consistency among multiple texts
- both - Check both within a single text and among multiple texts (default)
check_dimensions (string[]): Check dimensions (multiple selections allowed)
- tone - Tone consistency
- vocabulary - Vocabulary consistency
- sentence_structure - Sentence structure consistency
- punctuation - Punctuation consistency
- emoji - Emoji consistency
- Default: check all
strictness_level (string): Strictness level
- low - Lenient check (only check obvious inconsistencies)
- medium - Medium check (check major inconsistencies, recommended)
- high - Strict check (check all subtle inconsistencies)
- Default: medium
auto_fix (boolean): Whether to automatically fix
- true - After checking, automatically call content_imitation to unify the style
- false - Only check, do not automatically fix (default)
reference_text (string): Reference text (optional when auto_fix=true)
- Provide a reference text, and all other texts will be unified to the style of this text
- If not provided, the most representative text will be automatically selected as the reference

Output

Style consistency check report. If auto_fix=true, the fixed content will also be output.

Quick Mode

When the user input contains complete texts, directly execute with the default parameters.

Interactive Question Template (Simplified Version)

我将为您检查文本的风格一致性,请确认以下要求:

1. 待检查文本 (必选):
   请提供需要检查的文本 (可提供1-10个文本)

2. 检查范围 (可选):
   [1] 单个文本内部一致性
   [2] 多个文本之间的一致性
   [3] 单个文本内部和多个文本之间都检查 (推荐)

3. 严格程度 (可选):
   [1] 宽松检查,仅检查明显不一致
   [2] 中等检查,检查主要不一致 (推荐)
   [3] 严格检查,检查所有细微不一致

4. 是否自动修复 (可选):
   [1] 仅检查,不自动修复
   [2] 检查后自动统一风格 (推荐)

【如果选择自动修复】
5. 参考文本 (可选):
   如有期望的风格参考,请提供一段文本
   (如不提供,将自动选择最具代表性的文本作为参考)

您可以:
- 一次性回复所有信息
- 回复"检查"或"好的",使用推荐设置执行

请回复:

Interaction with content_imitation

When auto_fix=true, automatically call content_imitation after the check is complete:

If reference_text is provided, use that text as the source content and the other texts as the target content
If reference_text is not provided, automatically select the most representative text as the source content
Use medium imitation depth (imitation_depth=medium)
Output all the repaired texts

Example

Example 1: Check only

Input:

texts: [
  "产品功能强大,操作简单,界面友好。用户反馈非常好。",
  "这个产品不仅功能强大,而且操作非常简单,界面也很友好。很多用户都说好。",
  "产品功能强大,操作简单,界面友好。用户的评价很高。"
]
auto_fix: false

Output:

【风格一致性检查报告】

总体评分: 7.5/10 (基本一致,有小幅改进空间)

---

各维度分析:

1. 语气一致性: 9/10 (高度一致)
   所有文本都采用积极、肯定的语气,情感色彩统一

2. 词汇一致性: 8/10 (基本一致)
   - 一致之处: 核心词汇"功能强大"、"操作简单"、"界面友好"在三个文本中都出现了
   - 不一致之处:
     * 文本2使用了"很多用户都说好",而文本1用"用户反馈非常好",文本3用"用户的评价很高"
     * 建议: 统一使用一个表达,如"用户反馈非常好"

3. 句式结构一致性: 6/10 (部分一致)
   - 不一致之处:
     * 文本1和3使用简单句式:"产品功能强大,操作简单,界面友好。"
     * 文本2使用复合句式:"这个产品不仅功能强大,而且操作非常简单,界面也很友好。"
   - 建议: 统一句式结构,建议使用简洁的并列句式

---

改进建议:

1. 统一句式结构
   建议: 所有文本使用并列句式,如"产品功能强大,操作简单,界面友好。"

2. 统一用户反馈的表达
   建议: 统一使用"用户反馈非常好"或"用户的评价很高",避免混用

Example 2: Check and automatically repair

Input:

texts: [
  "产品功能包括:注册,登录,设置",
  "产品功能包括:注册,登录,设置",
  "产品功能包括: 注册、登录、设置 "
]
auto_fix: true

Output:

【风格一致性检查报告】

总体评分: 3/10 (一致性较差,需要较大改进)

---

主要问题:
1. 标点符号不一致 (使用了半角和全角混用)
2. emoji使用不一致
3. 空格使用不一致

---

【自动修复中...】
检测到风格不一致,已自动调用content_imitation进行风格统一。

【修复后的内容】

文本1: "产品功能包括:注册,登录,设置"
文本2: "产品功能包括:注册,登录,设置"
文本3: "产品功能包括:注册,登录,设置"

Example 3: Use reference text for unification

Input:

texts: [
  "我们的产品功能强大,操作简单,界面友好。欢迎您体验!",
  "这款产品功能丰富,易于使用,界面美观。希望您喜欢!",
  "本产品性能卓越,操作便捷,用户界面友好。期待您的使用!"
]
reference_text: "我们的产品功能强大,操作简单,界面友好。欢迎您体验!"
auto_fix: true

Output:

【风格一致性检查报告】

总体评分: 6/10 (部分一致,需要调整)

---

【自动修复中...】
检测到风格不一致,已使用参考文本进行风格统一。

参考文本: "我们的产品功能强大,操作简单,界面友好。欢迎您体验!"

【修复后的内容】

文本1: "我们的产品功能强大,操作简单,界面友好。欢迎您体验!" (保持不变)
文本2: "我们的产品功能强大,操作简单,界面友好。欢迎您体验!" (已统一)
文本3: "我们的产品功能强大,操作简单,界面友好。欢迎您体验!" (已统一)

Notes

Style consistency checking is based on statistics and rules, and the possibility of misjudgment cannot be ruled out
The strictness setting affects the sensitivity of the check, so choose according to actual needs
The automatic repair feature will call content_imitation; it is recommended to check first and then decide whether to repair
Some inconsistencies may be intentional, such as different scenarios requiring different tones

Parameter Adjustment Tips

After generating content, if you are not satisfied with the current result, you can try adjusting the following parameters:

💡 Parameter adjustment options:

Check scope: single_text_internalConsistency within a single text | multiple_texts_crossConsistency among multiple texts | bothCheck both within a single text and among multiple texts
Check dimensions: toneTone consistency | vocabularyVocabulary consistency | sentence_structureSentence structure consistency | punctuationPunctuation consistency | emojiEmoji consistency
Strictness level: lowLoose check - only check obvious inconsistencies | mediumModerate check - check major inconsistencies | highStrict check - check all minor inconsistencies
Whether to automatically repair: falseCheck only, do not automatically repair | trueAutomatically unify style after checking

Usage: Reply directly with something like "change to strict check" or "automatically unify style after checking"

Relationship with other skills

content_imitation: When auto_fix=true, automatically call this skill for style unification
rewrite_content: If inconsistent text needs to be rewritten, this can be used together
review_text: If a comprehensive review of text quality is needed, this skill should be used

Configuration file reference

intent_keywords: ["style consistency", "consistency check", "style check", "unified style", "style unification"]
style_consistency_rules: Style consistency check rules are defined in config/style_consistency_rules.json

Feature 3: Content Imitation

Content Imitation - Content imitation

Overview

Imitate the style of the source content according to the target content, achieving style transfer and style unification. Supports multi-dimensional imitation and controllable imitation depth, suitable for unifying the style of multiple pieces of content or imitating a specific style. After generation is complete, automatically call humanize_text to remove AI traces.

Use cases

Unify the style of multiple pieces of content (such as keeping multiple Xiaohongshu copywriting pieces in a consistent style)
Imitate the style of an excellent work
Integrate new content into an existing content style system
Style transfer and style conversion
Unification and inheritance of brand tone

Input Parameters

Required Parameters

target_content (string): Target content (the content to be modified)
source_content (string): Source content (the content used to provide the imitation style)

Optional Parameters

imitation_dimensions (string[]): Imitation dimensions (multiple selections allowed)
- tone - Tone
- vocabulary - Vocabulary
- sentence_structure - Sentence structure
- punctuation - Punctuation
- emoji - Emojis
- format - Format
- all - All dimensions (recommended)
- Default value: all
imitation_depth (string): Imitation depth
- light - Light imitation (retain the core expression of the target content, adjust style only)
- medium - Medium imitation (retain core information, substantially adjust style)
- deep - Deep imitation (fully imitate the source content style, may involve considerable rewriting)
- Default value: medium
output_count (integer): Output quantity
- Generate 1-3 imitation results for selection
- Default value: 1
retain_key_info (boolean): Whether to retain key information
- true - Retain the core information and data of the target content
- false - Secondary information may be adjusted or omitted
- Default value: true

Output

The imitated content, with a style close to the source content while retaining the core information of the target content. Automatically call humanize_text to remove AI traces.

Processing Rules

Detailed Explanation of Imitation Dimensions

1. Tone (tone)

Imitation points:
- Tone color (friendly/serious/lively/steady)
- Emotional expression (positive/neutral/negative)
- Person usage (I/we/you/everyone)
- Use of modal particles (ne/a/ba, etc.)
Example:
- Source content tone: friendly, cordial, enthusiastic
- Before adjustment of target content: professional, rigorous, objective
- After adjustment of target content: friendly, cordial, enthusiastic

2. Vocabulary (vocabulary)

Imitation points:
- Vocabulary choice (colloquial/written/professional)
- Vocabulary complexity (simple/medium/complex)
- Use of internet slang (whether to use internet buzzwords)
- Vocabulary style (warm/lively/professional, etc.)
Example:
- Source content vocabulary: colloquial, lively, with internet slang
- Before adjustment of target content: written, formal, no internet slang
- After adjustment of target content: colloquial, lively, with internet slang

3. Sentence Structure (sentence_structure)

Imitation points:
- Sentence length (short/medium/long)
- Sentence type (simple sentence/compound sentence/compound-complex sentence)
- Proportion of active and passive voice usage
- Rhetorical devices (parallelism/antithesis/metaphor, etc.)
Example:
- Source content sentence structure: mainly short sentences, parallel structures, with parallelism
- Target content before adjustment: mainly long sentences, compound sentence structures, no rhetoric
- Target content after adjustment: mainly short sentences, parallel sentence structures, with parallelism

4. Punctuation

Imitation points:
- Use of Chinese and English punctuation
- Punctuation type (half-width/full-width)
- Punctuation usage frequency
- Use of special punctuation (e.g., ellipsis, dash)
Example:
- Source content punctuation: full-width punctuation, frequent use of exclamation marks
- Target content before adjustment: mixed Chinese and English, mainly periods
- Target content after adjustment: full-width punctuation, frequent use of exclamation marks

5. Emoji

Imitation points:
- Emoji usage frequency
- Emoji type (emoticons/objects/symbols, etc.)
- Emoji usage position (beginning/middle/end of sentence)
- Emotional tone of emoji
Example:
- Source content emoji: high-frequency use, mainly 🔥✨💯, positioned at the end of sentences
- Target content before adjustment: no emoji used
- Target content after adjustment: high-frequency use, mainly 🔥✨💯, positioned at the end of sentences

6. Format

Imitation points:
- Paragraph structure (single paragraph/multiple paragraphs)
- List format (numbered list/bulleted list/no list)
- Heading usage (whether there are headings, heading hierarchy)
- Indentation and spacing usage
Example:
- Source content format: multiple paragraphs, numbered lists, main heading and subheadings
- Target content before adjustment: single paragraph, no list, no headings
- Target content after adjustment: multiple paragraphs, numbered lists, main heading and subheadings

Imitation depth control

1. Light imitation (light)

Features: retain the core expression of the target content, only adjust the style
Scope of changes: 10-30%
Retain: core expression, main sentence structures, key information
Adjust: word choice, tone and color, punctuation usage, emoji
Applicable scenarios:
- The target content is already close to the source content style, only minor adjustments are needed
- Retain the unique expression of the target content

2. Medium imitation (medium)

Features: retain core information, significantly adjust style
Scope of changes: 30-60%
Retain: core information, key data, main viewpoints
Adjust: vocabulary, sentence structures, tone, punctuation, emoji, format
Applicable scenarios:
- The target content differs greatly in style from the source content
- Need to significantly change the style

3. Deep imitation (deep)

Features: comprehensively imitate the source content style, possibly rewriting to a greater extent
Scope of changes: 60-90%
Retain: core information (when retain_key_info=true)
Adjust: all-round style adjustment, may change the angle of expression
Applicable scenarios:
- Completely change the style to move closer to the source content
- Pursue a high degree of stylistic consistency

Imitation Process

1. Style Analysis

Analyze the style characteristics of the source content
Identify the style features of each dimension
Build a style profile of the source content

2. Difference Identification

Compare the target content with the source content
Identify style differences
Determine the dimensions that need adjustment

3. Imitation Execution

Adjust according to the selected imitation_dimensions
Control the degree of modification according to imitation_depth
Generate the imitated content

4. Quality Check

Check whether the imitation effect meets expectations
Ensure core information is retained (when retain_key_info=true)
Ensure the content is natural and fluent

5. Optimize Output

Adjust unnatural parts
Optimize expression fluency
Generate the final version

Style Profile Building

Style Profile Elements

Tone Characteristics
- Tone Color
- Emotional Expression
- Person Usage
- Modal Particle Usage
Vocabulary Characteristics
- Vocabulary Selection
- Vocabulary Complexity
- Internet Slang Usage
- Vocabulary Style
Sentence Pattern Characteristics
- Sentence Length
- Sentence Type
- Active/Passive Voice
- Rhetorical Devices
Punctuation Characteristics
- Chinese and English Punctuation
- Punctuation Type
- Punctuation Frequency
- Special Punctuation
Emoji Characteristics
- emoji Frequency
- emoji Type
- emoji Position
- emoji Emotional Color
Format Characteristics
- Paragraph Structure
- List Format
- Heading Usage
- Indentation and Spaces

Quick Mode

When the user input contains complete target_content and source_content, directly execute using the default parameters (all dimensions, medium imitation, output 1 result, retain key information).

Interactive Question Template (only displayed when the user does not provide the necessary parameters)

我将为您进行内容模仿，请确认以下要求：

1. 目标内容（必选）：
   请提供需要被修改的内容

2. 源内容（必选）：
   请提供用于提供模仿风格的内容

3. 模仿维度（可多选，可选）：
   [1] 语气
   [2] 词汇
   [3] 句式结构
   [4] 标点符号
   [5] 表情符号
   [6] 格式
   [7] 全部维度（推荐）

4. 模仿深度（可选）：
   [1] 轻度模仿（保留核心表达，仅调整风格）
   [2] 中度模仿（保留核心信息，大幅调整风格）（推荐）
   [3] 深度模仿（全面模仿，可能较大程度改写）

5. 输出数量（1-3，可选）：
   默认生成1个结果

6. 是否保留关键信息（可选）：
   [1] 保留核心信息和数据（推荐）
   [2] 可以调整或省略次要信息

您可以：
- 一次性回复所有信息
- 回复部分参数，我再确认是否需要补充
- 回复"模仿"或"好的"，使用推荐设置执行（推荐：全部维度，中度模仿，输出1个，保留关键信息）
- 如果你有其他想法，可在回复的时候一并描述告知

请回复：

Examples

Example 1: Xiaohongshu Style Imitation

Input：

target_content: "这款产品功能强大，操作简单，界面友好，欢迎大家体验。"
source_content: "姐妹们！这个产品真的绝了🔥 功能超强，好用到飞起！界面也超好看！必须安利给大家～ ✨"
imitation_dimensions: ["tone", "vocabulary", "emoji"]
imitation_depth: "medium"
output_count: 1

Output：

【模仿结果】

原内容：
这款产品功能强大，操作简单，界面友好，欢迎大家体验。

模仿后：
姐妹们！这款产品功能超强，好用到飞起！界面也超好看！必须安利给大家～ ✨

---

【模仿说明】
- 语气：从专业正式变为亲切热情，使用"姐妹们"、"必须安利"
- 词汇：从"功能强大"变为"功能超强"，从"操作简单"变为"好用到飞起"，从"界面友好"变为"界面也超好看"
- emoji：新增🔥和✨，位置在句末

Example 2: Multi-dimensional Deep Imitation

Input：

target_content: "本产品采用人工智能技术，能够为用户提供个性化服务。我们的团队由多位专家组成，拥有丰富的行业经验。"
source_content: "这款产品AI加持，智能化超强！给你个性化定制体验！团队大牛云集，行业经验满满！"
imitation_dimensions: "all"
imitation_depth: "deep"
output_count: 1

Output：

【模仿结果】

原内容：
本产品采用人工智能技术，能够为用户提供个性化服务。我们的团队由多位专家组成，拥有丰富的行业经验。

模仿后：
这款产品AI加持，智能化超强！给你个性化定制体验！团队大牛云集，行业经验满满！

---

【模仿说明】
- 语气：从正式变为活泼热情，使用感叹号
- 词汇："本产品"→"这款产品"，"采用人工智能技术"→"AI加持"，"能够"→"给你"，"专家"→"大牛"
- 句式结构：从长句变为短句，节奏明快
- 标点符号：使用更多感叹号
- 格式：从单段落变为短句式表达

Example 3: Light Imitation (Retain Core Expression)

Input：

target_content: "我们的产品具有以下功能：1. 注册登录；2. 数据管理；3. 分析报表。"
source_content: "产品超好用，功能超强！注册登录、数据管理、分析报表，啥都有！"
imitation_dimensions: ["tone", "vocabulary"]
imitation_depth: "light"
output_count: 1

Output：

【模仿结果】

原内容：
我们的产品具有以下功能：1. 注册登录；2. 数据管理；3. 分析报表。

模仿后：
我们的产品功能超强！包括：1. 注册登录；2. 数据管理；3. 分析报表。

---

【模仿说明】
- 语气：从正式变为活泼，增加感叹号
- 词汇："具有以下功能"→"功能超强"
- 句式结构和格式基本保持不变

Example 4: Multiple Output Results

Input：

target_content: "产品功能强大，操作简单，欢迎大家体验。"
source_content: "姐妹们！这个产品真的绝了🔥 必须安利！"
imitation_dimensions: ["tone", "vocabulary", "emoji"]
imitation_depth: "medium"
output_count: 3

Output：

【模仿结果1】
姐妹们！这款产品功能超强，操作超简单，必须安利给大家～ ✨

【模仿结果2】
姐妹们！产品功能太强了，操作也超简单，赶紧来试试吧！🔥

【模仿结果3】
姐妹们！这款产品功能爆表，好用到飞起！必须体验一下！✨

---

【模仿说明】
三个版本都模仿了源内容的亲切热情语气、口语化词汇和emoji使用
- 版本1："功能超强"、"超简单"、"必须安利"
- 版本2："太强了"、"超简单"、"赶紧来试试"
- 版本3："功能爆表"、"好用到飞起"、"必须体验"
三个版本略有差异，供选择使用

Example 5: Style Consistency (Make Multiple Xiaohongshu Copies Maintain a Consistent Style)

Input：

target_content: "这款护肤品补水效果很好，适合干性皮肤使用。"
source_content: "姐妹们！这个产品真的绝了🔥 亲测有效！必须安利给大家～ ✨"
imitation_dimensions: "all"
imitation_depth: "medium"
output_count: 1

Output：

【模仿结果】

原内容：
这款护肤品补水效果很好，适合干性皮肤使用。

模仿后：
姐妹们！这款护肤品补水效果真的绝了🔥 亲测有效！干性皮肤姐妹必须安排上～ ✨

---

【模仿说明】
- 语气：增加"姐妹们"，亲切热情
- 词汇："很好"→"真的绝了"，增加"亲测有效"、"必须安排上"
- emoji：新增🔥和✨
- 格式：更加活泼，节奏明快

这样处理后，该文案与已有的小红书文案风格保持一致

Notes

Imitation will automatically call humanize_text, so there is no need to manually remove the AI flavor again
Imitation depth controls the degree of modification; please choose according to actual needs
When retain_key_info=true, core information and data will be retained, but the expression may be adjusted
When the output quantity exceeds 1, multiple slightly different versions will be generated for selection
Imitation is based on style analysis, and unnatural elements may exist; manual review is recommended.
If multiple pieces of content need to have a unified style, you can use this skill in batch with the same source content.

Parameter Adjustment Tips

After generating content, if you are not satisfied with the current result, you can try adjusting the following parameters:

💡 Parameter adjustment options:

Imitation dimensions: toneTone | vocabularyVocabulary | sentence_structureSentence structure | punctuationPunctuation | emojiEmoji | formatFormat | allAll dimensions
Imitation depth: lightLight imitation (retain core expression, adjust style only) | mediumModerate imitation (retain core information, substantially adjust style) | deepDeep imitation (comprehensive imitation, may involve a relatively large degree of rewriting)
Output quantity: 1-3 imitation results to choose from
Retain key information: trueRetain core information and data | falseSecondary information can be adjusted or omitted

How to use: Reply directly, for example, "Deep imitation of all dimensions" or "Generate 3 versions"

Relationship with other skills

humanize_text: After imitation is complete, automatically call this skill to remove AI traces
check_style_consistency: If you need to check the style consistency of multiple pieces of content, you should use this skill
rewrite_content: If you need to perform style transfer but do not need to precisely imitate a certain source content, you can use this skill
generate_titles/generate_slogan: If you need to make titles/slogans consistent with the style of existing content, you can use it together

Configuration file reference

intent_keywords: ["content imitation", "style imitation", "style unification", "style transfer", "imitate style"]
style_analysis_rules: Style analysis rules are defined in config/style_analysis_rules.json

About this skill

Overview

Function 1: Rewrite Content

Overview

Use Cases

Input Parameters

Required Parameters

Optional Parameters

Podcast mode parameters (effective when target_channel is a podcast platform)

Output

Processing Rules

Multi-channel rewriting strategy

1. Multi-channel mode activation

2. Channel characteristic adaptation

WeChat Official Account (wechat_article)

Weibo (weibo)

Zhihu (zhihu)

Xiaohongshu (xiaohongshu)

Douyin short video (douyin_short)

E-commerce (ecommerce)

Official website/portal (official_website)

Email (email)

Report (report)

Speech (speech)

3. Style consistency control

maintain_style_consistency=true (maintain consistency)

maintain_style_consistency=false (full adaptation)

4. Multi-channel output format

Single-channel rewriting strategy

Style characteristic analysis

Rewriting strategy

Podcast content generation strategy

Quality control

Quick mode

Interaction question template (only shown when the user has not provided the necessary parameters)

Examples

Example 1: Multi-channel rewriting (maintain style consistency)

Example 2: Multi-channel rewriting (fully adapt to each channel's style)

Example 3: Change to social media style

Example 2: Change to academic style

Example 3: Change to humorous style

Example 4: Business style rewriting

Notes

Parameter adjustment tips

Relationship with Other Skills

Configuration File References

Feature 2: Check Style Consistency

Overview

Use Cases

Input Parameters

Required Parameters

Optional Parameters

Output

Quick Mode

Interactive Question Template (Simplified Version)

Interaction with content_imitation

Example

Example 1: Check only

Example 2: Check and automatically repair

Example 3: Use reference text for unification

Notes

Parameter Adjustment Tips

Relationship with other skills

Configuration file reference

Feature 3: Content Imitation

Overview

Use cases

Input Parameters

Required Parameters

Optional Parameters

Output

Processing Rules

Detailed Explanation of Imitation Dimensions

1. Tone (tone)

2. Vocabulary (vocabulary)

3. Sentence Structure (sentence_structure)

4. Punctuation

5. Emoji

6. Format