MCP Connector

Connect Routine to Productivity Tools

Routine MCP Server bridges the Routine task and calendar app to MCP clients like Claude Desktop via a local npx server.

Works with github

77
Spark score
out of 100
Updated Jun 2025
Source checked Sep 10, 2026
Version 1.0.0
Models
universal

Add to Favorites

Why it matters

Integrate your Routine productivity suite with external applications. This asset enables seamless interaction with calendars, tasks, and notes via the Model Context Protocol.

Outcomes

What it gets done

01

Facilitate communication between Routine and external services.

02

Enable access to calendars, tasks, and notes.

03

Operate as a server for the Model Context Protocol.

Source

Get it from source

Spark does not host a copy of it.

Open source

Reports

Agent outcome reports

No reports yet

Overview

Routine MCP Server

Routine MCP Server is a stdio bridge between the Routine desktop task and calendar app and MCP-compatible AI assistants, letting an assistant act on a locally running Routine instance. Use it if you already run Routine locally and want an assistant like Claude Desktop to interact with it via natural language rather than the app's own UI.

What it does

The Routine MCP server is a bridge between the Routine desktop application (routine.co) and MCP-compatible AI assistants. It lets an assistant interact with a running Routine instance through the Model Context Protocol, communicating over stdin/stdout in line with the MCP protocol. The server itself does not replace Routine - it requires the Routine desktop app to be running locally for the MCP tools to have anything to act on.

When to use - and when NOT to

Use this when you already run the Routine app for task and calendar management and want an AI assistant such as Claude Desktop to read from or act on that data through natural-language requests, without switching to the Routine UI. It is not useful without Routine installed and running, since the server is purely a protocol bridge with no standalone functionality of its own.

Inputs and outputs

The server is started either via npx routine-mcp-server or by pointing an MCP client directly at a built node/dist/index.js entry point. Communication follows the standard MCP request/response pattern over stdio: an MCP client sends JSON requests to the server's stdin and reads JSON responses from stdout, with the server translating those into operations against the local Routine application.

Integrations

Built with a standard Node.js/TypeScript toolchain (yarn for install and build), it integrates with any MCP client that supports stdio transport, with Claude Desktop given as the primary configuration example via claude_desktop_config.json.

Who it's for

Routine app users who want an AI assistant to interact with their tasks and calendar through natural language instead of the Routine interface directly.

Source README

Routine Model Context Protocol (MCP) Server

This is the Routine Model Context Protocol (MCP) server.

Usage

  1. Run the Routine application for the MCP server to work.
  2. Run this MCP server with npx routine-mcp-server or configure it in your favorite MCP client.

Claude Desktop

For Claude Desktop, refer to https://modelcontextprotocol.io/quickstart/user

In particular, your file claude_desktop_config.json should look something like that:

{
  "mcpServers": {
    "routine": {
      "command": "npx",
      "args": ["routine-mcp-server"]
    }
  }
}

Development

# Install dependencies
yarn

# Build the project
yarn build

Then install the MCP server:

  • Command: full path to node executable
  • Arguments: full path to ./dist/index.js

Claude Desktop

For Claude Desktop, refer to https://modelcontextprotocol.io/quickstart/user

In particular, your file claude_desktop_config.json should look something like that:

{
  "mcpServers": {
    "routine": {
      "command": "/absolute/path/to/bin/node",
      "args": ["/absolute/path/to/mcp-server/dist/index.js"]
    }
  }
}

Running the MCP Server (development)

# Start the server
yarn start

The server communicates via stdin/stdout, following the MCP protocol. You can interact with it by sending JSON requests to its stdin and reading responses from stdout.

FAQ

Common questions

Discussion

Questions & comments · 0

Sign In Sign in to leave a comment.