sound-mcp
A server that enables agents to play notification sounds and desktop toasts, useful for notifying users when background tasks complete.
README
sound-mcp
<p align="center"> <img src="assets/banner.png" alt="sound-mcp banner" width="600"> </p>
<p align="center"> <a href="https://www.python.org/"><img src="https://img.shields.io/badge/python-%E2%89%A5%203.11-3776AB?logo=python&logoColor=white" alt="Python"></a> <a href="#requirements"><img src="https://img.shields.io/badge/platform-linux-555?logo=linux&logoColor=white" alt="Platform"></a> <a href="#license"><img src="https://img.shields.io/badge/license-MIT-blue" alt="License"></a> </p>
A small FastMCP server that lets the agent play a notification sound, and optionally pop a desktop toast, when something finishes. Useful for "poll X until Y, then ping me" workflows.
Tools
list_sounds(): show every available sound name and the current default.play_sound(name=None): play a registered sound, an absolute file path, or the default.set_default_sound(name_or_path): persist a new default (registered name or absolute path) to~/.config/sound-mcp/default. Survives restarts. Overridden at runtime bySOUND_MCP_DEFAULTif it is set inmcp.json.notify(message, title="Claude", sound=None, urgency="normal", expire_ms=0): desktop toast + sound.urgency=critical: makes the toast sticky on most desktops.
Requirements
Linux with paplay (PulseAudio/PipeWire) and notify-send (libnotify-bin).
Builtin sounds come from sound-theme-freedesktop, preinstalled on Ubuntu.
You only need uv and Python ≥ 3.11.
Quick start
Claude Code
claude mcp add sound-mcp -- uvx --from git+https://github.com/giuliocapecchi/sound-mcp sound-mcp
Other MCP clients (Claude Desktop, Codex, OpenCode, …)
Add this block to your client's MCP config file:
{
"mcpServers": {
"sound-mcp": {
"command": "uvx",
"args": ["--from", "git+https://github.com/giuliocapecchi/sound-mcp", "sound-mcp"]
}
}
}
Common config locations:
| Client | Config file |
|---|---|
| Claude Desktop | claude_desktop_config.json |
| Codex (OpenAI) | ~/.codex/config.json |
| OpenCode | opencode.json |
Configuration
All settings are read from environment variables, so you set them in the env
block of your mcp.json (or equivalent):
{
"mcpServers": {
"sound-mcp": {
"command": "uvx",
"args": ["--from", "git+https://github.com/giuliocapecchi/sound-mcp", "sound-mcp"],
"env": {
"SOUND_MCP_DEFAULT": "warning",
"SOUND_MCP_SOUNDS_DIR": "/home/you/.config/sound-mcp/sounds"
}
}
}
}
| Variable | Default | Purpose |
|---|---|---|
SOUND_MCP_DEFAULT |
warning |
Sound played when play_sound/notify is called with no name. |
SOUND_MCP_SOUNDS_DIR |
~/.config/sound-mcp/sounds |
Directory scanned for user sound files. |
Custom sounds
Builtin names: complete, bell, warning, error, alarm.
Drop audio files (.oga, .ogg, .wav, .mp3, .flac) into your sounds
directory and each becomes a sound keyed by its filename stem. For example
tada.wav → play_sound("tada"). User files override builtins on name collision.
You can also pass an absolute path directly: play_sound("/tmp/horn.wav").
Example prompt
Poll
tailscale statusevery 10 seconds. When nodemyservershows as online, call thenotifytool withtitle="Tailscale"andmessage="myserver is back".
Local development
git clone https://github.com/giuliocapecchi/sound-mcp
cd sound-mcp
uv venv && uv pip install -e .
uv run fastmcp dev src/sound_mcp/server.py # opens MCP Inspector
License
MIT.
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。
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。
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。
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。
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。
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。
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。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。