# Omarchy release smoke test

Run this checklist on a target Omarchy machine before tagging a release. The
automated suite cannot reproduce compositor, monitor, fullscreen, suspend, or
Flatpak behavior from macOS/CI.

## Setup

```bash
npm run check
npm run omarchy:validate
omarchy plugin add "$(pwd)" --enable
omarchy-shell com.mirashif.omarest status | jq
```

Create short one-minute focus boundaries for the applications under test and
use **Settings → Test break screen** before starting the longer mixed-use run.

## Required checks

- [ ] Firefox, Chromium, Steam, Discord, and one Flatpak report stable classes.
- [ ] The anchored popup opens beside the OmaRest bar widget on every bar edge.
- [ ] The popup lists every visible Omarchy launcher application with search and icons.
- [ ] Focus time advances only while the matching application is in front.
- [ ] A brief app switch preserves the continuous session.
- [ ] Staying away for the configured reset duration restores the allowance.
- [ ] Lock and screensaver time are not charged.
- [ ] Suspend/resume does not add the suspended interval.
- [ ] A warning appears once at the configured threshold.
- [ ] The gate covers every connected monitor.
- [ ] The gate remains above fullscreen video and a fullscreen game.
- [ ] Completing a break closes every gate surface and records completion.
- [ ] Holding the override closes every surface and records an override.
- [ ] `Ctrl+Shift+Esc` always dismisses the gate immediately.
- [ ] Reduced motion stops breathing/orbit animation.
- [ ] Hot-reloading the shell restores rules and does not double-count time.
- [ ] Malformed state files recover to safe defaults without blocking input.
- [ ] Configuration above 256 KiB and state above 1 MiB are rejected without stalling the shell.
- [ ] Oversized `hyprctl -j activewindow` output is rejected without replacing the Wayland result.
- [ ] Missing `hyprctl` leaves Wayland tracking operational and reports status.
- [ ] **Delete local data** clears focus boundaries and daily activity.

## Thirty-minute attribution run

Alternate among at least two monitored apps, an unmonitored app, lock, and a
short suspend. Compare `status`, the popup countdowns, and a stopwatch. Total
attributed foreground time should remain within one timer tick per transition,
with no attribution during lock or suspend.