Understand how Things Cloud MCP connects your AI to Things 3 from anywhere.
Most MCP servers for Things 3 run locally on your Mac, communicating directly with the Things app via AppleScript or URL schemes.
This means you can only manage tasks from the same machine where Things is running. If you want to use a web-based AI like Claude.ai or ChatGPT, or manage tasks from a Linux server, you are stuck.
Things Cloud MCP takes a different approach: it talks directly to Things Cloud — the sync service that keeps your tasks in sync across all your Apple devices.
We use a reverse-engineered Things Cloud SDK to read and write your tasks over the internet. This means the MCP server can run anywhere — on a remote server, in a Docker container, on a Linux machine — and any MCP-compatible AI client can connect to it.
The server implements the Streamable HTTP transport, so web-based clients like Claude.ai and ChatGPT can connect directly.
This project uses a reverse-engineered, unofficial API to communicate with Things Cloud. While we have tested it thoroughly, there is always a risk when using unofficial APIs.
Before connecting: Make sure your Things 3 data is backed up. Things automatically syncs to iCloud, but you may also want to export your data via File → Export in Things for Mac.
The authors are not responsible for any data loss.