Last updated 2 October 2026
How a $99 Mac app updates itself: notarisation, Sparkle and a signed feed
Updates are the part of indie Mac software most likely to be done badly. Ours took a few tries to get right, and this is what the final version looks like.
Build and notarise
Each release is built as a universal binary, signed with an Apple Developer ID and sent to Apple for notarisation. Apple scans it and staples a ticket to the disk image. Gatekeeper on your Mac checks that ticket before the app opens.
Sign the feed
The app checks an appcast file on adhdterminal.com once a day. Each entry carries an EdDSA signature made with a private key that never leaves our build machine. The matching public key is compiled into the app. A download whose signature does not verify is thrown away.
Install in place
Sparkle replaces the app in Applications and relaunches it. This is why we warn when the app is run from the disk image: a mounted image is read-only, and there is nothing to replace.
What is sent
The update check sends the app name and version. No identifier, no system profile, no usage data. We can count checks on our server, and that is the only number we see.
Read next
- How to run five Claude Code sessions at once without losing oneParallel agent work fails at the attention layer, not the compute layer. A practical setup for keeping every running session visible.
- Claude Code resume: picking a session back up after a restartWhat --resume and --continue actually do, where the history lives, and how to make resuming automatic.
- The restart tax: why tabs are the wrong unit for agent workA tab you cannot see is a job you will restart. The hidden cost of hiding running work behind a title bar.
- Naming agent sessions: why the agent should name itselfFolder names describe projects, not jobs. Letting the session write its own title keeps the wall honest.