Thanks for tackling this, @jensenpat — the "stuck slider" behavior is a real papercut for operators, and the lease-based approach is a solid design choice over the permanent suppression.

I read through the full diff and compared against main. The implementation is clean and well-scoped (only MainWindow.cpp / .h, no scope creep). A few observations:

Looks good

• Lease pattern is sound. The beginSliderShortcutLease → renewSliderShortcutLease → releaseSliderShortcutLease lifecycle is clear. Using QPointer for m_sliderShortcutLease is the right call — safely handles slider destruction mid-lease.
• Mouse-drag safety. isSliderDown() check in both renewSliderShortcutLease (stops the timer) and releaseSliderShortcutLease (defers the clear) correctly prevents yanking focus during an active drag. Nice detail.
• PageUp/Down and Home/End handling. The old code only intercepted arrow keys. Adding these makes the slider feel complete before the lease expires.
• No convention violations. Uses AppSettings pattern (not QSettings), QPointer for RAII-safe weak reference, constexpr for the timeout constant. Files changed match stated scope.

Questions / potential issues

1. Double disable on focus change. When a slider is clicked, the eventFilter fires beginSliderShortcutLease (via MouseButtonPress), which calls setShortcutsEnabled(false). Then the focusChanged signal fires for the same slider and calls beginSliderShortcutLease again, which calls setShortcutsEnabled(false) a second time. This is idempotent and harmless, but the timer restart in renewSliderShortcutLease means the 2-second clock resets from the focus event rather than the click — which is actually fine since they're nearly simultaneous. Just flagging that the double-entry is intentional/benign.

2. s_sliderShortcutLeaseActive static vs. member. The lease state is tracked in both a file-static s_sliderShortcutLeaseActive (for shortcutInputCaptured()) and the member m_sliderShortcutLease pointer. This works because there's only ever one MainWindow, but it's worth a brief comment noting the static is needed because shortcutInputCaptured() is a free function called from the shortcutGuard callback. (Not blocking — just a readability note.)

3. Keyboard-only focus. If a user Tabs into a slider (no mouse click), the focusChanged path calls beginSliderShortcutLease, which starts the timer. After 2 seconds of idle, releaseSliderShortcutLease(true) calls slider->clearFocus(). This means Tab-focused sliders lose focus after 2s even if the user hasn't interacted with them yet. In practice this may be fine for this app (operators rarely Tab to sliders), but worth considering whether the clearFocus should only happen if there was an actual key nudge. A simple guard like m_sliderHadKeyNudge would do it.

4. Thread safety note. All paths run on the GUI thread (event filter, focus signal, QTimer), so the static flag is safe. No issue here, just confirming.

Overall this is a well-considered improvement. The core approach is correct and the edge cases are handled. Item 3 above is the only thing I'd suggest addressing before merge — the Tab-focus scenario could surprise users who use keyboard navigation. The rest is ready.

Reviewed by AetherClaude