A first cut of a one-tap Tune-Up run-all shipped on `feat/ui-followups-2026-06-14` and was reverted before 0.7.2 — it was too thin. Tracking the real design here.
Problems with the first cut
- State didn't persist. Closing the Tune-Up sheet and reopening it reset everything — spinners/progress gone, run state lost. It must survive navigation and ideally app relaunch.
- It was a popup sheet. It should be its own screen/pane, not a modal that loses context.
- It only sequenced Clean + Optimize. It should be a proper "review everything, then act" surface.
Desired design
A persistent Tune-Up pane that, on entry, auto-runs the read-only/preview scans and flags what's worth reviewing, then lets the user act:
- Apps to uninstall — surface stale/large apps (last-used + size) for review.
- App updates — apps with updates available (Sparkle / App Store / Electron) + Homebrew outdated.
- Startup items to review — new/again-enabled login items & launch agents worth a look.
- Big disk users — top space hogs (drive from `mo analyze`) for review.
- Cleanable junk — the Clean dry-run total.
- Maintenance — Optimize tasks.
Then a one-tap "run the safe set" that executes the chosen items, with:
- Persisted status + progress across pane switches and relaunch (not a transient sheet).
- Expanding section cards → done summary.
- Conservative defaults + per-item opt-out + a visible pre-run plan.
- Honest auth handling (today each elevated step prompts separately; pooling needs the backlogged privileged helper).
Notes
- This is the "Smart Care"-style hub from `plans/ui-followups-2026-06-14.md` (PR-E), reworked from sheet → persistent pane with auto-flagged reviews.
- Reuse existing engines/flows (Clean/Optimize/Updates/Startup/Analyze) — no new process paths.
A first cut of a one-tap Tune-Up run-all shipped on `feat/ui-followups-2026-06-14` and was reverted before 0.7.2 — it was too thin. Tracking the real design here.
Problems with the first cut
Desired design
A persistent Tune-Up pane that, on entry, auto-runs the read-only/preview scans and flags what's worth reviewing, then lets the user act:
Then a one-tap "run the safe set" that executes the chosen items, with:
Notes