SuperAPI AI Directory

mcp_server_notify

Send desktop notifications with a sound effect when agent tasks complete.

Open website

About this tool

mcp_server_notify is an MCP server that sends a desktop notification with a sound effect when an Agent task is completed.

Purpose: After an agent finishes executing a task, it reminds the user via a desktop system notification and a prompt sound that the task is complete, making it convenient for users to learn the result promptly without having to keep staring at the screen waiting.

Main capabilities:

  • Sends a system desktop notification after the agent task is completed.
  • Plays a prompt sound to attract the user's attention, and includes sound files.
  • Cross-platform support (Windows, macOS, Linux).
  • Based on the standard MCP protocol and can be integrated with a variety of LLM clients.

How to use:

  • Install using the uv package manager.
  • After installation, you can directly call the module to check whether the installation was successful.
  • The module accepts the --debug or --file options.

Using with Claude Desktop:

  • Find the configuration file claude_desktop_config.json.
  • If it is a global installation, you can also use the python command.

Using with Cursor:

  • Find the configuration file ~/.cursor/mcp.json or your_project/.cursor/mcp.json.
  • After configuration, simply add a prompt like "finally, send me a notification when task finished." at the end of the task input given to the AI to trigger the notification.
  • In Cursor, you can add this prompt as a rule in Cursor Settings -> Rules so that you do not need to enter it manually every time.

Using with VSCode + Copilot:

  • Install the service manager uv/uvx: pip install uv.
  • Add the service to the VSCode settings: on Windows it is %APPDATA%\Code\User\settings.json, on macOS it is $HOME/Library/Application Support/Code/User/settings.json, and on Linux it is $HOME/.config/Code/User/settings.json, with the configuration content being "mcp" : { "servers" : { "notifier" : { "command" : "uvx", "args" : [ "mcp-server-notify" ], "env" : {} } } }.
  • Make sure to use the latest version of VSCode, which will automatically run the MCP service.
  • Open VSCode → enable Copilot → switch to agent mode.
  • Enter # → you will see the #send_notification option.
  • Have the agent run #send_notification (it will automatically handle the notification).
  • At this point, Copilot in agent mode can send desktop notifications.

Special requirements:

  • Sending desktop notifications uses the Apprise API, so some special dependencies need to be installed on the desktop side.

Docker support:

  • Currently unavailable due to environment compatibility issues. Regardless of whether the host operating system is Windows, macOS, or Linux, if a Docker container needs to trigger a host notification, the solution becomes much more complex, and it is usually not feasible to use native notifications directly. Specific reasons include: each operating system (Windows, macOS, Linux) has its own unique notification mechanism; the isolation of Docker containers limits their ability to directly access host operating system resources; and different notification libraries and dependencies for each operating system need to be handled.

Applicable scenarios:

  • Suitable for users who use LLM clients that support the MCP protocol, such as Claude Desktop, Cursor, and VSCode + Copilot, and who need to receive completion reminders promptly after an agent runs a task for a long time.

Limitations:

  • Running in Docker is currently unavailable.
  • Special dependencies related to the Apprise API need to be installed additionally.
  • Information such as pricing and charging models has not yet been provided.