music-assistant-mcp
Enables AI assistants to control Music Assistant for music search, library browsing, playback, queue management, and playlist curation.
README
music-assistant-mcp
An MCP server that wraps the Music Assistant client library, exposing music search, library browsing, playback control, queue management, and playlist curation to AI assistants.
Configuring and Installing
Generate a Long-Lived Access Token First
- Navigate to your Music Assistant instance at
http://<your-ma-host>:8095/#/settings/profile. - Scroll down to the Long-lived access token section and click + Create new token.
- Assign a name and submit.
- Copy the generated token value to a safe place — you will need it in the setup steps below.
Get Environment Variables Ready
To reach a Music Assistant instance, you'll need to set some environment variables:
| Variable | Required | Description |
|---|---|---|
MA_SERVER_URL |
Yes (or MA_HOST) |
Full URL like http://10.10.10.10:8095 |
MA_HOST |
No | Hostname, combined with MA_PORT |
MA_PORT |
No | Defaults to 8095 |
MA_TOKEN |
Yes | Bearer token generated by Music Assistant |
You don't necessarily have to set these in your shell in a persistent way. These variables can typically be injected into the MCP server's execution context as necessary as shown in the examples below.
Claude Integration
Many will want to use this MCP server from Claude Desktop, Claude Code, or another MCP host. You can add the MCP server to Claude Code with the CLI like this:
claude mcp add --transport stdio music-assistant \
--env MA_SERVER_URL=http://your-ma-host:8095 \
--env MA_TOKEN=your-token \
-- uvx music-assistant-mcp
Or add it to your Claude Code or Claude Desktop MCP configuration files as follows:
{
"mcpServers": {
"music-assistant": {
"command": "uvx",
"args": ["music-assistant-mcp@latest"],
"env": {
"MA_SERVER_URL": "http://your-ma-host:8095",
"MA_TOKEN": "your-token"
}
}
}
}
Hosting in a Container
When you run the Music Assistant MCP server in a Docker container, your MCP client can't get to the stdin and stdout file descriptors of the server. You'll have to use the streamable-http transport instead. This MCP server uses three more environment variables to configure for this case:
| Variable | Required | Description |
|---|---|---|
MA_MCP_TRANSPORT |
No | Transport protocol: stdio (default) or streamable-http |
MA_MCP_HOST |
No | IP address to bind (default 0.0.0.0) |
MA_MCP_PORT |
No | Port to listen on (default 8000) |
Check out the sample MA_MCP_Docker project that shows how to build a Docker container that uses them to host this server.
A Note on Authentication
The stdio transport is unauthenticated because only the spawning process has access to the server's stdin and stdout file descriptors. With HTTP however, the endpoint becomes network-accessible so you need to take care when running the server in any sort of container.
If this is on your home lab and isolated to a VLAN that you control, you're probably fine. But if you're exposing it more broadly, it is highly recommended that you put the service behind an nginx reverse proxy with bearer token authorization at a minimum.
Tools (19)
The tools exposed by the Music Assistant MCP server generally follow the pattern of the underlying music-assistant module upon which it is built. However, MCP clients often become overwhelmed by having too many choices, so some related functions have been grouped into tools like manage_favorites and player_control with a range of actions available on them.
Search & Browse
- search_music - Search across all providers
- browse_media - Browse provider folders
- get_item_by_name - Find item by name, artist, album
Library & Discovery
- get_library - Browse library by media type (artist, album, track, playlist) with search/filter/pagination
- get_item_children - Get child items: album tracks, artist albums, artist top tracks, or playlist tracks
- get_similar_tracks - Get tracks similar to a given track
- get_recommendations - Get personalized recommendations
- get_recently_played - Recently played items
- get_library_stats - Library counts (tracks, albums, artists, playlists)
- get_server_info - Server version, Music Assistant backend details, and connected providers
- manage_favorites - Add or remove items from favorites
Playback
- get_players - List all players and state
- player_control - Control a player: play, pause, stop, next, previous, seek, volume, mute, power, group, ungroup
Queue
- play_media - Play URIs on a queue
- get_queue - Queue state
- get_queue_items - Items in queue
- queue_control - Queue settings: clear, shuffle, repeat
Playlists
- create_playlist - Create new playlist
- manage_playlist_tracks - Add or remove tracks from a playlist
Resources (2)
As of this writing, Claude Desktop doesn't support MCP resources well. In a future release, that may change. For now, we've also made these resources available as tools as described.
ma://players- All players with current state (also available as the get_players tool)ma://library/stats- Library counts (also available as the get_library_stats tool)
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。