No AI connection
The MCP endpoint that connects an assistant to the planning layer exists only in the Mac app. AI access is one of its three paid features.

ReminderMAP began as a Python program with a web interface. It uses a server built from one Python file and the standard library, plain HTML, CSS and JavaScript with no framework or build step, and a local SQLite file for the planning layer.
The open source edition is complete, MIT licensed and served as the design reference for the native Mac app.
01 / Contents
02 / The difference
The open source edition differs from the Mac app in two important ways.
The MCP endpoint that connects an assistant to the planning layer exists only in the Mac app. AI access is one of its three paid features.
It runs as a local server you open in a browser. There is no menu bar, no window restoration, no App Store updates, and permission prompts are attributed to the terminal you launched it from rather than to an app.
03 / Relationship
Same design, two implementations. The Mac app is a native rewrite rather than a wrapper around this code, and the two develop separately.
This edition stays public because it is worth having in the open. The code is small enough to read in an afternoon, and under MIT you can fork it and take it wherever you want.
04 / Audience
05 / Quick start
git clone https://github.com/Cogria-AI/ReminderMap
cd ReminderMap
python3 server.pyThen open 127.0.0.1:8788 in a browser. macOS asks for permission to access your reminders on the first run. Writes are real writes: completing or deleting a task here changes the actual reminder, exactly as the Mac app does.