youtube-video-analyzer

YouTube Video Intelligence Mining

Quickly get the core content, key viewpoints, and structured summary of a YouTube video without watching the full video.

Trigger Conditions

Start this skill when the user says the following:

• "Analyze this YouTube video"
• "Summarize this video"
• "Help me see what this video is talking about"
• "YouTube video summary"
• Or directly provide a YouTube link

Workflow

┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│  解析视频URL     │───▶│  获取视频信息    │───▶│  提取字幕/描述   │
│  提取Video ID   │    │  标题/作者/时长  │    │  (WebFetch)     │
└─────────────────┘    └─────────────────┘    └─────────────────┘
                                                      │
                                                      ▼
                              ┌─────────────────────────────────────┐
                              │            AI 分析生成               │
                              │  • 内容摘要                         │
                              │  • 核心观点（按主题分类）            │
                              │  • 关键术语解释                      │
                              │  • 实用建议/行动项                   │
                              └─────────────────────────────────────┘

Execution Steps

Step 1: Parse the YouTube URL

Extract the Video ID from the URL provided by the user. Multiple formats are supported:

• https://www.youtube.com/watch?v=VIDEO_ID
• https://youtu.be/VIDEO_ID
• https://youtube.com/embed/VIDEO_ID

Extraction rule (regex):

(?:youtube\.com\/(?:[^\/]+\/.+\/|(?:v|e(?:mbed)?)\/|.*[?&]v=)|youtu\.be\/)([^"&?\/\s]{11})

Step 2: Get basic video information

Use WebFetch to access the video page and extract:

• Video title
• Channel name
• Publish time
• Video duration
• Video description

Step 3: Get video content

Method A(Recommended): Via a third-party transcription service

• Try to access https://youtubetranscript.com/?v=VIDEO_ID
• Or use other subtitle extraction APIs

Method B: Analyze the video description and comments

• If subtitles cannot be obtained, analyze based on the video description and title

Step 4: AI structured analysis

Analyze the obtained content and output according to the following structure:

## 视频概要

**标题**: [视频标题]
**频道**: [频道名称]
**时长**: [时长]
**链接**: [原始链接]

## 内容摘要

[2-3 句话概括视频主要内容]

## 核心观点

### 主题 1: [主题名称]
- 要点 1
- 要点 2

### 主题 2: [主题名称]
- 要点 1
- 要点 2

## 关键术语

| 术语 | 解释 |
|------|------|
| **术语1** | 简要解释 |
| **术语2** | 简要解释 |

## 实用建议

- [ ] 行动项 1
- [ ] 行动项 2

## 适合人群

- 人群 1
- 人群 2

Usage Examples

Example 1: Analyze a single video

用户: 帮我分析这个视频 https://www.youtube.com/watch?v=dQw4w9WgXcQ

Claude:
1. 解析 URL，提取 Video ID: dQw4w9WgXcQ
2. 获取视频信息和描述
3. 尝试获取字幕内容
4. 生成结构化分析报告

Example 2: Quickly understand a long video

用户: 这个2小时的演讲视频讲了什么？[URL]

Claude:
[生成简洁的内容摘要，突出核心观点]

Data Storage

Analyzed videos are cached in:~/.claude/cache/youtube-video-analyzer/

File naming:

• {VIDEO_ID}-{YYYYMMDD}.md: Video analysis report
• cache.json: Index cache of analyzed videos

{
  "VIDEO_ID": {
    "title": "视频标题",
    "analyzed_at": "2025-12-29T10:00:00Z",
    "summary": "...",
    "key_points": [...]
  }
}

Dependent Tools

• WebFetch: Get video pages and subtitles
• Read/Write: Cache analysis results

Limitations

• Videos without subtitles can only be analyzed based on the description, and the information may be incomplete
• Private videos or region-restricted videos cannot be accessed
• Very long videos (>2 hours) may only have part of their content analyzed