April
Vanderslice

The product promise: keep using Reminders. This is also the hardest engineering constraint in the app. I need typed numbers per item, and Reminders gives me strings. I can't add a column, I can't migrate anyone, and the user can edit the same list in Apple's app, on a Mac, or from Siri while my app is open. Three things followed from that:

1. 01The encoding has to be legible to humans, not just to me. Whatever I write into a reminder is text a user will see in Apple's own app forever. So the format reads as something a person would have typed, and an item that has never been touched by List Total is simply an item with no price. It is not an error state and not a migration prompt.
2. 02Parsing is forgiving; writing is strict. The reader accepts the sloppy variants people actually type and ignores anything it can't understand, rather than throwing. Every write goes back out in one canonical form. Unparseable text is never destroyed. Worst case an item contributes nothing to the total, which is a visible, correctable outcome instead of silent data loss in someone's shopping list.
3. 03Treat the store as the source of truth. Totals are computed from whatever EventKit currently reports and the app refreshes on external change rather than assuming it owns the list. That's what makes iCloud conflicts a non-event: I have no second copy that can disagree.

The payoff is the thing I'd defend in review: the app has no onboarding, no account, no sync layer, and no data of its own to lose.

• Write the encode/decode round-trip tests before the UI. It's the one piece where a bug corrupts data a user actually cares about, and it's trivially testable in isolation. I built it alongside the interface instead, which meant I was verifying parsing by shopping.
• Decide the permission story on day one. EventKit access is the entire app: without it there is nothing to show, so the denied and not-yet-asked states deserved to be designed screens from the start rather than something I handled once the happy path worked.
• Ship the paid tier with a way to try it. $2.99 with no free path means the App Store page has to do all the convincing, and I have no data on how many people bounced because they couldn't see a total on their own list first.