Get the FREE Ultimate OpenClaw Setup Guide →

notify-completion

The MCP server that provides notifications when AI agents complete long-running tasks. This allows you to leave your desktop without forgetting about tasks.

Installation
Run this command in your terminal to add the MCP server to Claude Code.
Run in terminal:
Command
claude mcp add --transport stdio nakat-t-notify-completion-mcp-server npx -y @nakat-t/notify-completion-mcp-server@latest \
  --env OVERDUE_COMMAND="<Command to execute when a task is overdue or complete, e.g. show notification>" \
  --env EXECUTE_IN_SERVER="<If using Claude Desktop set to "true", otherwise omit or leave unset>"

How to use

The notify-completion MCP server provides a lightweight task completion notifier. It integrates with clients that support MCP configurations (such as Cline, Claude Desktop, Cursor, or other compatible tools) to monitor the execution time of long-running AI tasks and trigger a user notification when tasks complete or exceed a specified threshold. By configuring the MCP server, you can start a timer at task initiation and then check whether the task is overdue; if overdue, a user-defined command executes (for example displaying a desktop notification or playing a sound). This empowers you to multitask without constantly watching your task status. To use it, set up the MCP server in your client configuration, supply an overdue command via OVERDUE_COMMAND, and ensure the client is configured to auto-approve the MCP tools so the timer and overdue checks can run automatically. The tools exposed are start-timer to begin tracking time, and check-overdue to determine if the task has exceeded the allowed duration.

How to install

Prerequisites:

  • Node.js and npm installed on your system.
  • Access to the npm registry (public or private as configured).

Installation steps:

  1. Prepare your client configuration file (e.g., cline_mcp_settings.json) and add the MCP server block as shown:
{
  "mcpServers": {
    "notify-completion": {
      "command": "npx",
      "args": ["-y", "@nakat-t/notify-completion-mcp-server@latest"],
      "env": {
        "OVERDUE_COMMAND": "<Command to execute when the task is overdue/complete>",
        "EXECUTE_IN_SERVER": "<true|false>"
      },
      "autoApprove": [ "start-timer", "check-overdue" ]
    }
  }
}
  1. Install and start using the MCP server via npx as configured above. The package will be fetched from npm and loaded by your client runtime.

  2. Customize environment variables:

  • OVERDUE_COMMAND: the command to run when the task is overdue or completed (e.g., a desktop notification command).
  • EXECUTE_IN_SERVER: if using Claude Desktop, set to "true" to execute inside the MCP server; otherwise omit or set to false.
  1. Ensure your client is configured to auto-approve the MCP tools so the start-timer and check-overdue steps execute without user intervention.

Additional notes

Notes and tips:

  • OVERDUE_COMMAND controls what happens when a task is complete or overdue. Use a notification command suitable for your OS (e.g., display notification, play a sound).
  • If you are using Claude Desktop, you may set EXECUTE_IN_SERVER to true, which changes where the overdue command runs.
  • The MCP tools exposed are:
    • start-timer: returns a startTime in ISO format to mark when the task began.
    • check-overdue: compares elapsed time against a threshold; if overdue, it returns a command to run via OVERDUE_COMMAND.
  • For best results, keep your environment variables and auto-approve settings in sync across client configurations to prevent prompts during automation.
  • This MCP server is ideal for notifying you about long-running AI tasks so you can proceed with other work without manual monitoring.

Related MCP Servers

Sponsor this space

Reach thousands of developers