YouTube MCP Server
Provides distraction-free access to YouTube by allowing AI assistants to retrieve subscribed channels, latest videos, and transcripts without algorithmic recommendations. It enables searching within subscriptions and direct reading of video content through simple browser cookie authentication.
README
YouTube MCP Server
Distraction-free YouTube for AI assistants. Get your subscribed channels, latest videos, and transcripts without algorithm recommendations.
Features
- No Distractions: Lists only your subscribed content
- Direct Links: Every video has a clickable link
- Read Mode: Get transcripts to read instead of watching
- No OAuth: Uses browser cookies - much simpler
- Caching: Fast responses for repeated queries
Available Tools
| Tool | Description |
|---|---|
get_subscribed_channels |
List all your subscribed channels |
get_latest_videos |
Get latest videos from subscriptions with direct links |
get_video_transcript |
Get transcript of a video (read instead of watching) |
search_subscribed_channels |
Search within your subscriptions |
get_channel_videos |
Get videos from a specific channel |
setup_youtube_auth |
Get authentication setup instructions |
Setup
1. Install Dependencies
cd /home/naresh/Projects/youtube-mcp-python
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
2. Export YouTube Cookies
Using Chrome:
yt-dlp --cookies-from-browser chrome --cookies auth/cookies.txt
Using Firefox:
yt-dlp --cookies-from-browser firefox --cookies auth/cookies.txt
Note: You must be logged into YouTube in your browser first.
3. Add to OpenCode
Add to ~/.config/opencode/opencode.jsonc:
{
"mcp": {
"youtube": {
"type": "local",
"command": ["/home/naresh/Projects/youtube-mcp-python/.venv/bin/python", "/home/naresh/Projects/youtube-mcp-python/main.py"],
"enabled": true
}
}
}
Usage Examples
Get latest videos:
Use get_latest_videos with max_results=15
Read a video transcript:
Use get_video_transcript with video_id="dQw4w9WgXcQ"
Search your subscriptions:
Use search_subscribed_channels with query="python tutorial"
Get a specific channel's videos:
Use get_channel_videos with channel_id="UCwX6rVkOq0"
Project Structure
youtube-mcp-python/
├── main.py # Entry point
├── mcp_server.py # MCP tool definitions
├── config.py # Configuration
├── requirements.txt # Dependencies
├── opencode-mcp-config.json
├── auth/
│ └── __init__.py # Cookie validation
├── services/
│ ├── __init__.py
│ ├── subscriptions_service.py
│ ├── videos_service.py
│ ├── transcripts_service.py
│ └── search_service.py
├── utils/
│ ├── __init__.py
│ ├── ytdlp_wrapper.py
│ ├── cache.py
│ └── retry.py
└── data/
└── cache/ # Cached data
Troubleshooting
"Cookies not found" → Run the yt-dlp cookie export command
"401 Unauthorized" → Re-export cookies (they expire when you log out)
"No subscriptions found" → Make sure you're subscribed to channels on YouTube
Credits
Built with:
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。