li-analyzer

Viral Topic Breakdown and Classification System

Three operating modes:

Single-blogger analysis: Analyze all videos of a specified blogger and generate a classified organization report
Full benchmark analysis: Read all existing viral topic analysis reports, perform cross-blogger horizontal comparison, and generate a topic selection decision framework

Core Features

1. Automatic Data Collection

Read all video scripts of a specified blogger
Extract key data: likes, saves, views, follower growth, impressions
Calculate key metrics: save/like ratio, follower growth rate, completion rate

2. Viral Tier Classification

✈️ Super viral tier (5k+ likes)
🚗 Major viral tier (2k-5k likes)
🛵 Medium viral tier (1k-2k likes)
🚲 Small viral tier (500-1k likes)
🚶 Ordinary tier (<500 likes)

3. Topic Type Summarization

Tool-type topics (first tests of new tools, beginner introductions, tool breakdowns, money-saving guides)
Celebrity-type topics (expert interview interpretations, celebrity viewpoints)
Opinion-type topics (counterintuitive ideas, philosophical reflections)
Experience-type topics (hackathons, from 0 to 1, weekly vlogs)
Method-type topics (demand mining, open-source projects)

4. Viral Element Extraction

Title formula templates
Opening script routines
Content structure patterns
Data presentation strategies
Cover text styling

5. Classified Organization Output

Display grouped by viral tier
Display categorized by topic type
Extract replicable creation templates
Generate a topic inspiration list

Analysis Dimensions

Data Dimension

Absolute data: likes, saves, views, follower growth, impressions
Relative metrics:
- Save/like ratio (>1.3 = high practical value, <1.0 = high emotional value)
- Follower growth rate (follower growth/views)
- Cover click-through rate
- Completion rate
Viral tier: classified according to like count

Content Dimension

Topic type: tool/celebrity/opinion/experience/method
Title structure: time promise, specific numbers, lowering the barrier, counterintuitive
Content characteristics: number of cases, data precision, tutorial depth
Emotional value: surprise, anxiety, achievement, curiosity

Dissemination Dimension

Dissemination type (likes > saves): driven by emotional value, suitable for traffic generation
Save type (saves > likes): driven by practical value, suitable for conversion
Balanced type (likes ≈ saves): balanced emotional + practical, optimal state

Analysis Process

Mode Determination

用户触发词 → 判断运行模式：
- "分析[博主名]" / "拆解[博主名]" → 单博主分析
- "全对标分析" / "跨博主分析" / "总结爆款规律" → 全对标分析
- "分析我的内容" / "数据复盘" → 单博主分析（自己账号）

Single Blogger Analysis (Step 1-5)

Step 0: Read account-specific accumulated insights (required reading)

Read 04-方法论沉淀/INDEX.md, and use the "decision rules" to determine whether there are relevant accumulated insights;At minimum, required reading 04-方法论沉淀/自媒体思考/完播率悖论.md, to understand the counterintuitive rule verified on Li Ge's account that "a completion rate ≤ 5% actually makes it easier to go viral."

⚠️ Do not mark a low completion rate (<5%) as a problem; instead, it is a signal of high interaction potential. During analysis, treat "cover click-through rate" and "save/like ratio" as more important levers.

Step 1: Data Collection

1. 优先读取 `03-数据统计/数据统计表.md`，获取全局数据索引
   - 可直接得到所有内容的点赞/收藏/观看/涨粉/完播率
   - 跳过逐文件扫描，效率更高
2. 如需读取完整脚本内容（分析标题公式/开场结构等），
   再按需读取对应的 `02-已发布内容/` 具体文件
3. 统计表中标注"待填入"的条目，跳过数据分析，仅做内容维度分析

Step 2: Data Analysis

1. 按点赞数排序
2. 计算收藏/点赞比
3. 计算涨粉率
4. 分级标注(✈️🚗🛵🚲🚶)
5. 识别数据异常值

Step 3: Topic Summarization

1. 分析标题关键词
2. 识别话题模式
3. 归类到5大类型
4. 统计各类型数据表现
5. 提取高频要素

Step 4: Formula Extraction

1. 超级爆款的共同特征
2. 标题公式模板
3. 内容结构模式
4. 数据呈现方式
5. 封面花字策略

Step 5: Generate Report

1. 数据对比表格
2. 话题分类清单
3. 爆款公式总结
4. 创作灵感库
5. 选题决策树

Output Format

Categorized Organization Structure

# [博主名称]_爆款分类整理

生成日期: YYYY-MM-DD
分析样本: X个视频

---

## 📊 按爆款等级分类

### ✈️ 超级爆款 (5千+点赞) - X个

1. **[标题]** (点赞X | 收藏X | 观看X)
   - 话题类型: [工具/名人/观点/经验/方法]
   - 标题公式: [提取的公式]
   - 核心要素: [关键成功要素]

2. ...

### 🚗 大爆款 (2千-5千点赞) - X个

1. **[标题]** (点赞X | 收藏X | 观看X)
   - 话题类型: [类型]
   - 标题公式: [公式]
   - 核心要素: [要素]

[其他等级...]

---

## 🏷️ 按话题类型分类

### 工具类话题 (X个)

#### ✈️ 超级爆款
- **[标题]** - [一句话总结成功要素]

#### 🚗 大爆款
- **[标题]** - [一句话总结]

[其他等级...]

### 名人类话题 (X个)
[同上结构]

### 观点类话题 (X个)
[同上结构]

### 经验类话题 (X个)
[同上结构]

### 方法类话题 (X个)
[同上结构]

---

## 💡 可复制的爆款模板

### 工具类模板
1. **[工具名]深度实测** - 适用于新工具发布
   - 标题公式: [工具名] + 实测/首测 + [惊人数字]
   - 成功案例: [具体视频]

2. **[工具名]小白入门** - 适用于热门工具
   - 标题公式: X分钟上手[工具名] | 小白也能学会
   - 成功案例: [具体视频]

### 名人类模板
[同上结构]

### 观点类模板
[同上结构]

---

## 🎯 选题灵感清单

### 待创作选题（基于爆款规律）
- [ ] [选题1] - 参考: [成功案例]
- [ ] [选题2] - 参考: [成功案例]
...

---

**整理完成**

Usage Method

User says: "Break down the viral points of The Fourth Chimpanzee"

System executes:

Search the "The Fourth Chimpanzee past video scripts" folder
Read all .md files
Extract data and titles
Classify by viral level and topic type
Extract replicable templates
Save as: 爆款话题分析报告/第四种黑猩猩_爆款分类整理_YYYY-MM-DD.md

Full Benchmark Analysis Mode

Trigger words: "full benchmark analysis", "cross-blogger analysis", "which type of topic is most likely to go viral", "summarize viral patterns"

Data Sources

1. Glob 扫描 `05-选题研究/爆款话题分析报告/*.md`，Read 所有报告
2. 需要验证标题结构或内容特征时，按需读取对应逐字稿（不默认全读）

Analysis Process

Step 1：汇总各报告中所有视频条目
Step 2：按话题类型横向合并（打破博主边界）
Step 3：计算每类话题的爆款命中率、平均点赞、平均收/赞比
Step 4：找出在 2 个及以上博主账号上都有效的标题结构
Step 5：生成选题优先级排名和可套用模板

Output Format

# 全对标横向分析报告

生成日期：YYYY-MM-DD
分析来源：[列出读取的报告文件名]
覆盖样本：X 条视频，X 位博主/来源

---

## 一、话题类型爆款成功率

> 爆款 = ✈️ + 🚗，命中率 = 爆款数 / 该类型总数

| 话题类型 | 样本数 | 爆款命中率 | 平均点赞 | 平均收/赞比 | 代表案例 |
|---------|-------|-----------|---------|-----------|---------|
| 工具类  | X     | XX%       | X       | X.X       | [标题]  |
| 名人类  | X     | XX%       | X       | X.X       | [标题]  |
| 观点类  | X     | XX%       | X       | X.X       | [标题]  |
| 经验类  | X     | XX%       | X       | X.X       | [标题]  |
| 方法类  | X     | XX%       | X       | X.X       | [标题]  |

**结论：优先做 [类型]，因为 [数据理由]**

---

## 二、跨博主通用标题公式

> 仅收录在 2 个及以上来源上验证有效的公式

### 公式：[公式名]
- **模板**：[具体格式]
- **验证案例**：
  - [@来源] [标题] — 点赞X
  - [@来源] [标题] — 点赞X
- **适用话题类型**：[类型]

[更多公式...]

---

## 三、收/赞比规律

| 内容特征 | 典型收/赞比 | 典型代表 | 适合目标 |
|---------|-----------|---------|---------|
| 教程型  | 1.5+      | [标题]  | 转化留粉 |
| 工具测评 | 1.0-1.5  | [标题]  | 平衡    |
| 观点型  | <1.0      | [标题]  | 引流    |

---

## 四、选题优先级建议

1. **首选**：[话题类型] × [标题公式] — [数据支撑]
2. **次选**：[话题类型] × [标题公式] — [数据支撑]
3. **谨慎**：[话题类型] — [理由]

---

## 五、可立即使用的选题模板

- [ ] [选题] — 参考：[公式名] | 案例：[成功视频]
- [ ] [选题] — 参考：[公式名] | 案例：[成功视频]
...

---

**报告完成**

Save Path

05-选题研究/全对标分析报告/全对标分析_YYYY-MM-DD.md

Analysis Standards

Viral Level Determination Standards

Level	Like Count Range	Mark	Characteristics
Super Viral	5,000+	✈️	Phenomenal spread, massive follower growth
Major Viral	2,000-5,000	🚗	Strong spread, steady follower growth
Medium Viral	1,000-2,000	🛵	Moderate spread, moderate follower growth
Minor Viral	500-1,000	🚲	Small-scale spread, slight follower growth
Ordinary	<500	🚶	Basic traffic, viewed by followers

Interpretation of Save/Like Ratio

Ratio	Interpretation	Strategy
>1.5	Extremely high practical value	Tutorial-style content, suitable for conversion
1.3-1.5	High practical value	Balance of practical and emotional
1.0-1.3	Balanced type	Optimal state
0.8-1.0	Emotional value dominant	Suitable for spread and traffic generation
<0.8	Pure emotional value	Opinion-style content

Topic Type Determination Rules

Tool Type

Title contains a specific tool name (Claude, Gemini, Kimi, etc.)
Contains keywords such as "hands-on test", "getting started", "breakdown", "cheap"
Content is mainly operational steps

Celebrity Type

Title contains a person's name (Karpathy, Ilya, Gavin Baker, etc.)
Contains keywords such as "interview", "podcast", "opinion"
Content is mainly extraction of viewpoints

Opinion Type

Title contains "why", "the most important thing is", etc.
Contains counterintuitive viewpoints
Content is mainly argumentation

Experience Type

The title contains "growth diary", "participate", "from 0 to 1", etc.
Contains personal experience
The content is mainly narrative

Method type

The title contains "share", "how to", "method", etc.
Contains methodology
The content is mainly steps

Save rules

File naming

[博主名称]_爆款分类整理_YYYY-MM-DD.md

Save path

爆款话题分析报告/[博主名称]_爆款分类整理_YYYY-MM-DD.md

Core principles

Data-driven - Based on real data, let the numbers speak
Replicability - Provides specific formulas and templates that can be applied directly
Action-oriented - Provides creative inspiration and topic suggestions
Systematic - Comprehensively covers all videos, presented in a structured way

About this skill

Core Features

1. Automatic Data Collection

2. Viral Tier Classification

3. Topic Type Summarization

4. Viral Element Extraction

5. Classified Organization Output

Analysis Dimensions

Data Dimension

Content Dimension

Dissemination Dimension

Analysis Process

Mode Determination

Single Blogger Analysis (Step 1-5)

Step 0: Read account-specific accumulated insights (required reading)

Step 1: Data Collection

Step 2: Data Analysis

Step 3: Topic Summarization

Step 4: Formula Extraction

Step 5: Generate Report

Output Format

Categorized Organization Structure

Usage Method

Full Benchmark Analysis Mode

Data Sources

Analysis Process

Output Format

Save Path

Analysis Standards

Viral Level Determination Standards

Interpretation of Save/Like Ratio

Topic Type Determination Rules

Tool Type

Celebrity Type

Opinion Type

Experience Type

Method type

Save rules

File naming

Save path

Core principles