YouTube Music MCP
# YouTube Music MCP 🎵
This is a simple MCP server that allows you to search for and play tracks on YouTube Music directly from your AI assistant like Cursor or Claude Desktop.
Built with:
- [YouTube Music](https://music.youtube.com/)
- [Anthropic MCP](https://docs.anthropic.com/en/docs/agents-and-tools/mcp)
- [Cursor](https://cursor.so/)
## Available Tools
- `searchTrack`: Search for tracks on YouTube Music by name.
- `playTrack`: Play tracks directly by searching and opening them in your default browser.
# Installation
## 1. Get a key
To make this work you need a valid [Google Youtube API Key](https://console.cloud.google.com/marketplace/product/google/youtube.googleapis.com)
## 2. Add to cursor
Add the following MCP configuration to your Cursor `.cursor/mcp.json` settings:
```json
{
"mcpServers": {
"youtube-music-mcp": {
"command": "npx",
"args": ["-y", "@instructa/mcp-youtube-music"],
"env": {
"YOUTUBE_API_KEY": "<INSERT_API_KEY_HERE>"
}
}
}
}
```
**Develop**
This MCP is typically run directly using `npx` and doesn't require local installation or building unless you intend to modify the source code. If you want to develop it locally, you would typically clone the source repository (if available) and follow its specific contribution guidelines.
## License
This project is licensed under the MIT License - see the LICENSE file for details.
---
## Links
- X/Twitter: [@kregenrek](https://x.com/kregenrek)
- Bluesky: [@kevinkern.dev](https://bsky.app/profile/kevinkern.dev)
## Courses
- Learn Cursor AI: [Ultimate Cursor Course](https://www.instructa.ai/en/cursor-ai)
- Learn to build software with AI: [instructa.ai](https://www.instructa.ai)
## See my other projects:
* [AI Prompts](https://github.com/instructa/ai-prompts/blob/main/README.md) - Curated AI Prompts for Cursor AI, Cline, Windsurf and Github Copilot
* [codefetch](https://github.com/regenrek/codefetch) - Turn code into Markdown for LLMs with one simple terminal command
* [aidex](https://github.com/regenrek/aidex) A CLI tool that provides detailed information about AI language models, helping developers choose the right model for their needs.
* [codetie](https://github.com/codetie-ai/codetie) - XCode CLITDQS
Scored across 2 tools
The two tools have overlapping purposes that could cause confusion. Both 'playTrack' and 'searchTrack' involve searching for tracks on YouTube Music, with 'playTrack' additionally opening the top result in a browser. An agent might misselect between them when the intent is ambiguous, such as wanting to search without playing or vice versa, as the descriptions don't clearly delineate distinct use cases.
The tool names follow a consistent pattern throughout. Both use camelCase with a verb-noun structure ('playTrack' and 'searchTrack'), making them predictable and readable. There are no deviations or mixed conventions, which aids in clarity and usability.
The tool count of 2 feels too thin for a YouTube Music server's apparent scope. With only search and play functionality, there are obvious gaps like managing playlists, controlling playback, or accessing user data. This limited set may hinder agents from performing common music-related tasks, indicating an incomplete surface for the domain.
The tool surface is severely incomplete for a YouTube Music domain. It only covers searching and playing tracks, missing essential operations such as pausing, skipping, volume control, playlist management, or user authentication. These gaps will likely cause agent failures when attempting broader music interactions, as the current tools don't support a full lifecycle or common workflows.