YT-NINJA
Enables AI-powered YouTube video analysis including transcript management, video summaries, chapter generation, keyword extraction, and playback control. Supports searching videos, retrieving channel/playlist information, and translating transcripts using Google Gemini AI.
README
YT-NINJA 🥷
A comprehensive YouTube MCP (Model Context Protocol) server that provides AI-powered video analysis, playback control, transcript management, and advanced content processing capabilities.
Features
🎬 Video Playback
- Play videos in browser or VLC player
- Audio-only playback with ffplay
- Video segment playback with timestamp control
- Active playback session management
📊 Data Retrieval
- Get detailed video information (title, views, likes, duration, etc.)
- Fetch playlist details with all videos
- Retrieve channel information and statistics
- Search YouTube videos and music
- Download video thumbnails in multiple qualities
📝 Transcript Management
- Get official video transcripts
- AI-powered transcript generation (when official unavailable)
- Translate transcripts to any language
- Format transcripts with or without timestamps
🤖 AI-Powered Analysis
- Generate video summaries with key points
- Auto-generate chapter markers
- Extract relevant keywords with relevance scores
- Detect topics and categories
- Create AI-powered video highlights
Installation
Prerequisites
- Node.js >= 18.0.0
- npm >= 9.0.0
- Google Gemini API key (required for AI features)
- Optional: VLC Media Player (for VLC playback)
- Optional: FFmpeg (for audio playback and processing)
Setup
- Clone the repository:
git clone <repository-url>
cd yt-ninja
- Install dependencies:
npm install
- Configure environment variables:
cp .env.example .env
Edit .env and add your configuration:
# Required
GEMINI_API_KEY=your-google-gemini-api-key
# Optional
DOWNLOAD_DIR=./downloads
TEMP_DIR=./temp
MAX_CONCURRENT_DOWNLOADS=3
LOG_LEVEL=info
- Build the project:
npm run build
Configuration
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
GEMINI_API_KEY |
Yes | - | Google Generative AI API key for AI features |
DOWNLOAD_DIR |
No | ./downloads |
Directory for downloaded files |
TEMP_DIR |
No | ./temp |
Temporary files directory |
MAX_CONCURRENT_DOWNLOADS |
No | 3 |
Maximum concurrent downloads |
LOG_LEVEL |
No | info |
Logging level (error, warn, info, debug) |
Getting a Gemini API Key
- Visit Google AI Studio
- Sign in with your Google account
- Click "Create API Key"
- Copy the key and add it to your
.envfile
MCP Configuration
Add to your MCP settings file (mcp.json):
{
"mcpServers": {
"yt-ninja": {
"command": "node",
"args": ["/path/to/yt-ninja/dist/index.js"],
"env": {
"GEMINI_API_KEY": "your-api-key-here"
},
"disabled": false
}
}
}
Available Tools
Playback Tools
play_youtube_video
Play a YouTube video in browser or VLC player.
Parameters:
url(string, required): YouTube video URLplayer(string, optional): Player type -browserorvlc(default:browser)
Example:
{
"url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
"player": "browser"
}
Data Retrieval Tools
get_video_info
Get comprehensive information about a YouTube video.
Parameters:
url(string, required): YouTube video URL
Returns: Video title, description, channel, views, likes, duration, tags, thumbnail, etc.
get_playlist_info
Get information about a YouTube playlist.
Parameters:
url(string, required): YouTube playlist URL
Returns: Playlist title, description, video count, total duration, list of videos
get_channel_info
Get information about a YouTube channel.
Parameters:
channelId(string, required): Channel ID or URL
Returns: Channel name, description, subscriber count, total views, video count
search_youtube
Search for videos on YouTube.
Parameters:
query(string, required): Search querymaxResults(number, optional): Maximum results (1-50, default: 10)
Returns: Array of search results with video details
search_music
Search specifically for music on YouTube.
Parameters:
query(string, required): Music search querymaxResults(number, optional): Maximum results (1-50, default: 10)
Returns: Array of music search results
download_thumbnail
Download a video thumbnail image.
Parameters:
url(string, required): YouTube video URLoutputPath(string, optional): Output file pathquality(string, optional): Quality -maxres,high,medium,default(default:maxres)
Transcript Tools
get_transcript
Get the transcript/subtitles of a video.
Parameters:
url(string, required): YouTube video URLlanguage(string, optional): Language code (e.g., 'en', 'es', 'fr')
Returns: Transcript text, language, timestamps, source type
translate_transcript
Translate a video transcript to another language.
Parameters:
url(string, required): YouTube video URLtargetLanguage(string, required): Target language code
Returns: Translated transcript with original timestamps
AI Analysis Tools
summarize_video
Generate an AI-powered summary of a video.
Parameters:
url(string, required): YouTube video URLmaxWords(number, optional): Maximum words in summary (default: 200)
Returns: Summary text, key points, word count
generate_chapters
Auto-generate chapter markers for a video.
Parameters:
url(string, required): YouTube video URL
Returns: Array of chapters with timestamps, titles, and descriptions
get_keywords
Extract relevant keywords from a video.
Parameters:
url(string, required): YouTube video URLcount(number, optional): Number of keywords (default: 15)
Returns: Array of keywords with relevance scores and frequency
detect_topics
Detect topics and categories in a video.
Parameters:
url(string, required): YouTube video URL
Returns: Array of topics with confidence scores and categories
generate_video_highlights
Generate AI-powered video highlights.
Parameters:
url(string, required): YouTube video URLcount(number, optional): Number of highlights (5-10, default: 7)
Returns: Array of highlight moments with timestamps, descriptions, reasons, and scores
Usage Examples
Using with\
AI
Once configured as an MCP server, you can use YT-NINJA through natural language:
"Get information about this video: https://www.youtube.com/watch?v=dQw4w9WgXcQ"
"Summarize this YouTube video in 150 words"
"Generate chapters for this tutorial video"
"Extract the top 20 keywords from this video"
"Get the transcript and translate it to Spanish"
Programmatic Usage
import { dataManager, aiAnalyzer, transcriptManager } from 'yt-ninja';
// Get video info
const videoInfo = await dataManager.getVideoInfo('https://youtube.com/watch?v=...');
// Generate summary
const summary = await aiAnalyzer.summarizeVideo('https://youtube.com/watch?v=...', 200);
// Get transcript
const transcript = await transcriptManager.getTranscript('https://youtube.com/watch?v=...');
Development
Scripts
npm run dev- Run in development mode with hot reloadnpm run build- Build for productionnpm start- Start the production servernpm run lint- Lint codenpm run format- Format code with Prettiernpm run type-check- Check TypeScript types
Project Structure
yt-ninja/
├── src/
│ ├── index.ts # Entry point
│ ├── server.ts # MCP server setup
│ ├── integrations/ # External service integrations
│ │ ├── youtube.ts # YouTube API client
│ │ ├── genai.ts # Google GenAI client
│ │ ├── ffmpeg.ts # FFmpeg integration
│ │ └── process.ts # Process management
│ ├── managers/ # Feature managers
│ │ ├── DataManager.ts # Data retrieval
│ │ ├── PlaybackManager.ts # Playback control
│ │ ├── TranscriptManager.ts # Transcript operations
│ │ ├── AIAnalyzer.ts # AI analysis
│ │ ├── MediaProcessor.ts # Media processing
│ │ └── AdvancedFeaturesManager.ts # Advanced features
│ ├── types/ # TypeScript type definitions
│ └── utils/ # Utility functions
├── dist/ # Compiled output
├── downloads/ # Downloaded files
├── .env # Environment configuration
└── package.json
Error Handling
YT-NINJA provides detailed error messages with suggestions:
{
"success": false,
"error": {
"code": "INVALID_URL",
"message": "Invalid YouTube video URL",
"details": { "url": "..." },
"suggestions": [
"Provide a valid YouTube video URL",
"Example: https://www.youtube.com/watch?v=VIDEO_ID"
]
}
}
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。