StreakFocus

Context: People with ADHD, and anyone who shuts down when there's too much to track, stall on long task lists. Too many items create paralysis; too few feel trivial.

Decision: Hardcode exactly 3 tasks. No setting to change it. Task IDs are canonical: task-1, task-2, task-3.

Rationale: 3 is the smallest number that feels like "enough" while being completable in any day. The constraint is the feature: it forces prioritization.

Trade-off: Power users who want more will need a different app. That's intentional.

Context: People who work late finish tasks after midnight. A midnight cutoff would count that work toward the wrong day and break a streak they kept.

Decision: getDayKey() subtracts dayStartHour (default 3) hours before extracting YYYY-MM-DD.

Rationale: A task finished at 1 AM counts toward the day it was planned for, not the next one.

Trade-off: A person whose day starts before 3 AM has those early completions counted toward the day before. One fixed boundary keeps the rule explainable in a sentence.

Context: Users distrust habit apps that require accounts. Habit data feels private: sharing it with a server creates friction and trust barriers.

Decision: Data stays on the phone, in Capacitor Preferences and the app's local storage. It leaves only when the user exports a JSON backup through the share sheet, and importing that file restores it.

Rationale: Zero infrastructure cost, zero privacy liability, instant startup. The JSON backup covers the "new phone" scenario.

Trade-off: No cross-device sync. Deleting the app without exporting first loses the streak, so a backup reminder banner prompts an export.

Context: Even three tasks shown side by side invite re-reading, re-sorting and second-guessing. People who struggle to start need the next action, not a menu of options.

Decision: The app is one card beneath the header. The same card holds writing the three, doing them one at a time, the finished day, and the start of a new day. Each state shows one primary action.

Rationale: Every screen presents a single decision, so starting costs one tap. Large task text and three progress dots give structure without adding choices.

Trade-off: The three appear together only once the day is done. Seeing another task mid-day takes a swipe or a tap on its dot, which is the point: the other two wait their turn.

Context: Completing all 3 tasks is the app's core moment. It deserves more than a checkbox flip.

Decision: 5 user-selectable celebrations. Full-screen, brief (3s), automatic. Under reduced motion, Confetti, Stars and Fireworks give way to Minimal, and Focus keeps its screen with fades only.

Rationale: Positive feedback loops without points, badges, or leaderboards. The celebration is personal, not competitive. User choice of style makes it feel owned.

Trade-off: Could feel juvenile to some. Minimal, a brief brand tint on milestone days only, provides a quiet alternative.

Context: App lifecycle events (resume, visibility change, the 3 AM rollover check) can trigger revalidation multiple times per day. Without guards, streaks could double-increment.

Decision: Two idempotency keys: lastCountedDayKey prevents double streak increment, lastEvaluatedDayKey prevents redundant revalidation within a day.

Rationale: Belt and suspenders. Streak integrity is the app's most critical invariant: a streak that randomly increments or resets destroys user trust.

Trade-off: More complex state, harder to reason about. The App Info screen shows the stored streak state, which keeps it inspectable.

Context: People set their three early and then notice one is worded badly or is bigger than it looked. Retyping all three to fix one adds friction at the moment someone is trying to start.

Decision: On the do state a pencil sits beside "Task n of 3", and tapping the task text does the same. The text becomes a field: Enter saves, Escape cancels. Completion state is untouched.

Rationale: The pencil reads as belonging to the task it edits, and it adds no row and no words to the card. An Edit button under the task would put a second action in the primary action's path.

Trade-off: Editing closes once all three are done. The finished day is a record of what was completed, so its three stay as they were.

Context: A card sized to its content shrinks as the day goes on, and the finished day would be the smallest, lowest box on screen. The payoff moment would read as the app deflating.

Decision: Every state renders in one card with a minimum height of 33rem, centred in a container of constant height, measured against the tallest state at 320, 390 and 430px. It is a minimum rather than a fixed height, so a long task title still grows the card.

Rationale: The card is the fixed point of the app. Content changes inside it while the frame holds still, so completing the last task is an arrival rather than a jump.

Trade-off: Short states such as the new-day card carry more empty space than their content needs. Accepted so that no state sits smaller or lower than another.