topic-generator

Topic Generator

Instructions

Core objective: When the user has absolutely no inspiration, generate worthwhile topics from 0 to 1 and provide a complete execution plan.

Trigger Scenarios

This skill should be used in the following situations:

The user says "I don't know what to write"
The user says "help me think of a topic"
The user says "what topics are worth writing about recently"
The user says "recommend some topics to me"

Three-Layer Architecture Workflow

[Layer 1] Inspiration Trigger

Objective: Generate 8-10 candidate topics from three dimensions

Step 1.1: Hot-Topic Scanning

Execution:

Ask the user which field they want to write about (workplace/technology/personal growth/business/social observation/other)
Use search tools to scan hot topics from the last 7 days:
- Weibo hot search related topics
- Zhihu hot list related questions
- Industry news hot spots
Classify and annotate each hot topic

Output format:

🔥 热点扫描（近7天）

【🔥 热度型】流量大但竞争激烈
1. [话题1] —— 讨论量：XX，热度趋势：上升/平稳
   → 可切入角度：[一句话]

【💡 洞察型】有深挖空间
2. [话题2] —— 讨论量：XX，竞品分析：较少深度内容
   → 可切入角度：[一句话]

【🎯 细分型】小众但精准
3. [话题3] —— 讨论量：XX，目标人群：[具体画像]
   → 可切入角度：[一句话]

...（共 3-4 个热点选题）

Step 1.2: Personal Asset Inventory

Execution:

Scan the docs/ folder to analyze articles the user previously archived
Scan the .claude/styles/ style library to understand the writing styles the user is good at
Based on historical data, recommend intersection topics of "what you are good at + what has market demand"

Output format:

📚 个人资产盘点

【你之前写过的领域】
- [领域1]：X篇文章
- [领域2]：X篇文章

【你擅长的风格】
- [风格名称]：适合写 [类型] 文章

【基于你的优势推荐】
1. [选题A] —— 理由：你有 [经历/数据/观点]，可以写出差异化内容
2. [选题B] —— 理由：延续你之前 [文章名] 的话题，可以做成系列
3. [选题C] —— 理由：你的 [风格] 很适合这个话题

...（共 2-3 个基于个人优势的选题）

Special handling:

If docs/ is empty or has very few articles, skip this step and explain: "You do not yet have enough historical article data; it is recommended to choose from hot topics or competitor analysis first."

Step 1.3: Reverse Engineering (Competitor Analysis)

Execution:

Ask the user: "Which big influencers or public accounts do you follow? (Optional)"
If the user provides them, search for these accounts' recent articles
Analyze their topic directions and find "angles they have not thoroughly covered" or "which reader pain points are still unmet"

Output format:

🔍 竞品分析

【分析对象】：[用户提供的大V/公众号名称]

【他们最近在写什么】
1. [文章标题1] —— 切入角度：XXX
2. [文章标题2] —— 切入角度：XXX

【他们的读者还缺什么】
- 评论区高频问题：[问题1]、[问题2]
- 未被满足的痛点：[痛点描述]

【你可以写的差异化选题】
1. [选题X] —— 填补空白：[说明]
2. [选题Y] —— 深挖细节：[说明]

...（共 2-3 个差异化选题）

Special handling:

If the user does not provide competitors, skip this step

Step 1.4: Summarize Candidate Topics

Output format:

📋 候选选题清单（共 X 个）

【热点型】
□ 1. [选题名称] —— 来源：热点扫描
□ 2. [选题名称] —— 来源：热点扫描

【优势型】
□ 3. [选题名称] —— 来源：个人资产盘点
□ 4. [选题名称] —— 来源：个人资产盘点

【差异化型】
□ 5. [选题名称] —— 来源：竞品分析

---

请选择 2-3 个你感兴趣的选题（输入序号，如：1,3,5），我将进行深度验证。

[Layer 2] Topic Validator

Objective: Conduct "five-question scoring" on the 2-3 topics selected by the user, and screen out the ones most worth writing

Five-Question Scoring Criteria

Evaluate each topic according to the following:

Question	Scoring criteria	Weight
1. Pain point authenticity	Search verification: How many people are asking similar questions? Is there strong resonance in the comments?	25%
2. Competition level	How many people have already written about it? Did they write well? Is there still room for differentiation?	20%
3. Your advantages	Do you have unique experience/data/perspective? Or can you only speak in generalities?	25%
4. Spread Potential	Will readers share it after reading? (Emotional resonance/practical value/cognitive refresh)	20%
5. Sustainability	Can this topic be extended into a series? Or is it a one-off topic?	10%

Execution:

Score each question from 1-10
Calculate the weighted total score (out of 100)
Directly eliminate any total score below 60

Output format:

🎯 选题验证报告

═══════════════════════════════════════════════════

【选题A】：[选题名称]

【五问评分】
1. 痛点真实性（25%）：8/10 = 20分
   → 搜索发现：近7天有 XX 条相关讨论，评论区高频词："焦虑"、"不知道怎么办"
   
2. 竞争度（20%）：6/10 = 12分
   → 已有内容：较多，但大多是浅层分析，缺少 [具体角度]
   
3. 你的优势（25%）：9/10 = 22.5分
   → 你的优势：有 [经历/数据/观点]，可以提供独特视角
   
4. 传播潜力（20%）：7/10 = 14分
   → 传播点：[情绪共鸣/实用价值/认知刷新]
   
5. 可持续性（10%）：8/10 = 8分
   → 可延伸：可以做成 [系列名称]，至少 3-5 篇

【总分】：76.5/100 ✅ 推荐写

【风险提醒】
- [如有风险，列出]

═══════════════════════════════════════════════════

【选题B】：[选题名称]
...（同样的评分格式）

═══════════════════════════════════════════════════

【推荐排序】
🥇 第1名：[选题名称]（XX分）—— 强烈推荐
🥈 第2名：[选题名称]（XX分）—— 推荐
🥉 第3名：[选题名称]（XX分）—— 可以写

请选择你最终要写的选题（输入序号），我将为你生成执行方案。

[Third Layer] Topic Concretizer

Goal: Turn an abstract topic into a concrete direction that can be written about immediately

Step 3.1: Break down subtopics

Execution: Break the topic down into 3-5 subtopics that can each stand alone as an article

Output format:

📝 选题执行方案

【选题】：[最终选定的选题]
【综合评分】：XX/100

═══════════════════════════════════════════════════

【子话题拆解】

A. [子话题1]
   - 核心问题：[一句话描述]
   - 目标读者：[具体画像]
   - 预估字数：[建议字数]

B. [子话题2]
   - 核心问题：[一句话描述]
   - 目标读者：[具体画像]
   - 预估字数：[建议字数]

C. [子话题3]
   - 核心问题：[一句话描述]
   - 目标读者：[具体画像]
   - 预估字数：[建议字数]

【建议】
- 如果只写一篇：推荐 [子话题X]，因为 [理由]
- 如果做系列：按 A → B → C 的顺序，形成递进关系

Step 3.2: Design opening hooks

Execution: Design 2-3 candidate openings for each subtopic

Output format:

🎣 开头钩子设计

【子话题A】：[子话题名称]

钩子1：场景代入型
"[具体场景描述，让读者立刻想到自己]"

钩子2：数据震撼型
"[震撼数据 + 反常识观点]"

钩子3：反常识观点型
"[颠覆认知的观点]"

---

【子话题B】：[子话题名称]
...（同样格式）

Step 3.3: Generate a material list

Execution: List the materials that need to be researched

Output format:

📚 素材调研清单

【需要的数据】
□ [数据1]：如 XX 行业失业率
□ [数据2]：如 XX 年龄段薪资中位数

【需要的案例】
□ [案例1]：真实故事（可以从 [来源] 找）
□ [案例2]：行业报告（可以搜索 [关键词]）

【需要引用的观点】
□ [权威人士/机构]：关于 [话题] 的观点

【可以参考的爆款】
□ [文章标题1]：学习其 [结构/开头/标题]
□ [文章标题2]：学习其 [结构/开头/标题]

Step 3.4: Final output

Output format:

✅ 选题执行方案已生成

【选题】：[选题名称]
【推荐子话题】：[子话题X]
【推荐开头】：[钩子类型]
【素材清单】：已列出，请先调研

═══════════════════════════════════════════════════

【下一步】
1. 如果要立即开始写作，请说："确认选题，开始写作"
2. 如果要先调研素材，请说："先调研素材"
3. 如果要保存到选题库，请说："保存到选题库"

我将无缝对接到现有的写作工作流。

Topic Library Management

Save to the topic library

When the user says "save to the topic library," perform the following operations:

In the articles/_topic_pool/ directory, create a file
Filename format:YYYY-MM-DD-选题名称.md
The file content includes:
- Topic name
- Overall score
- Subtopic breakdown
- Opening hook
- Material list
- Generation date

File template:

# 选题：[选题名称]

**生成日期**：YYYY-MM-DD
**综合评分**：XX/100
**状态**：待写作

---

## 子话题拆解

### A. [子话题1]
- 核心问题：XXX
- 目标读者：XXX
- 预估字数：XXX

### B. [子话题2]
...

---

## 开头钩子

### 子话题A
1. [钩子1]
2. [钩子2]

---

## 素材清单

### 需要的数据
- [ ] [数据1]
- [ ] [数据2]

### 需要的案例
- [ ] [案例1]

---

## 备注

[用户可以手动添加备注]

Examples

User: I don't know what to write. Help me come up with a few topics. I mainly write workplace-related content.

AI：

好的，我将从三个维度为你生成候选选题。

首先，请问：
1. 你关注哪些职场大V或公众号吗？（可选，帮助我做竞品分析）
2. 你的目标读者是谁？（如：应届生/3-5年职场人/管理者）

User: I follow Jiubian and Banfo Xianren. The target readers are workplace professionals with 3-5 years of experience.

AI：

收到，开始生成选题...

🔥 热点扫描（近7天）

【🔥 热度型】
1. "年终奖缩水" —— 讨论量：8万+，热度趋势：上升
   → 可切入角度：年终奖缩水背后的公司财务信号

【💡 洞察型】
2. "职场35岁危机" —— 讨论量：5万+，竞品分析：内容多但缺少实操
   → 可切入角度：30岁开始为35岁做的3件事

【🎯 细分型】
3. "技术转管理的困境" —— 讨论量：2万+，目标人群：3-5年技术人
   → 可切入角度：技术转管理的第一年，我踩过的5个坑

---

📚 个人资产盘点

【你之前写过的领域】
- 职场分析：3篇文章
- 个人成长：2篇文章

【你擅长的风格】
- 九边风：适合写职场深度分析

【基于你的优势推荐】
1. "职场晋升的潜规则" —— 理由：你的九边风格很适合这个话题
2. "如何在大厂生存" —— 理由：延续你之前的职场系列

---

🔍 竞品分析

【分析对象】：九边、半佛仙人

【他们最近在写什么】
- 九边：社会机制分析、职场规则解读
- 半佛：商业模式拆解、行业黑幕

【他们的读者还缺什么】
- 评论区高频问题："道理都懂，但具体怎么做？"
- 未被满足的痛点：缺少可落地的实操指南

【你可以写的差异化选题】
1. "职场晋升实操手册" —— 填补空白：九边讲机制，你讲方法
2. "大厂生存指南" —— 深挖细节：半佛讲行业，你讲个人

---

📋 候选选题清单（共 8 个）

【热点型】
□ 1. 年终奖缩水背后的公司财务信号
□ 2. 30岁开始为35岁做的3件事
□ 3. 技术转管理的第一年，我踩过的5个坑

【优势型】
□ 4. 职场晋升的潜规则
□ 5. 如何在大厂生存

【差异化型】
□ 6. 职场晋升实操手册
□ 7. 大厂生存指南

---

请选择 2-3 个你感兴趣的选题（输入序号，如：2,4,6），我将进行深度验证。

Best Practices

Don't be greedy for quantity: Generating 8-10 candidate topics each time is enough; too many will make it hard for the user to choose
Data-driven: All judgments must be based on search results and real data, not speculation
Personalized recommendations: Prioritize recommending topics at the intersection of "what the user is good at + what the market needs"
Honest evaluation: If a topic scores below 60, clearly say "not recommended to write"
Keep updated: For topics in the topic library that have not been written for more than 3 months, it is recommended to re-verify their popularity

Version History

v1.0.0 (2026-01-15): Initial version, implementing the three-layer architecture (inspiration triggering, topic validation, concretization) and topic library management.

About this skill

Instructions

Trigger Scenarios

Three-Layer Architecture Workflow

[Layer 1] Inspiration Trigger

Step 1.1: Hot-Topic Scanning

Step 1.2: Personal Asset Inventory

Step 1.3: Reverse Engineering (Competitor Analysis)

Step 1.4: Summarize Candidate Topics

[Layer 2] Topic Validator

Five-Question Scoring Criteria

[Third Layer] Topic Concretizer

Step 3.1: Break down subtopics

Step 3.2: Design opening hooks

Step 3.3: Generate a material list

Step 3.4: Final output

Topic Library Management

Save to the topic library

Examples

Best Practices

Version History