fix(desktop): hold-to-quit no longer gets stuck

[saphid]

What Changed

Completed hold-to-quit gestures now finish after Cmd+Q repeats stop, even when macOS omits the useful key-up event. At the 1.2-second threshold, T3 makes its focused window transparent so the interaction feels complete while the process continues consuming repeats. Native fullscreen exits before concealment. The process quits after release is safely detected.

The quiet timer follows the observed repeat cadence so slow keyboard settings do not send Cmd+Q into the next app. Once the window is concealed, other keys cannot cancel the committed quit and strand an invisible process.

This changes only the desktop window and shortcut handler. It does not change web, mobile, providers, or wire contracts.

Why

The original handler could wait forever after a completed hold when macOS suppressed the release event. Waiting for repeat silence fixed that hang, but left the window visible after the 1.2-second hold threshold and gave no clear indication that the gesture had completed.

Quitting the process at the threshold would reintroduce the shortcut spillover fixed in #7369. Concealing the still-focused window gives immediate visual completion while keeping T3 alive long enough to consume the remaining repeats.

Verification

• vp test run src/window/QuitHold.test.ts src/window/DesktopWindow.test.ts in apps/desktop: 51 passed, 0 failed.
• pnpm --filter @t3tools/desktop typecheck: passed.
• Targeted lint and format checks for the four changed desktop files: passed.
• Added regressions for missing release events, slow repeat cadences, concealment at the hold threshold, committed gestures surviving extra input, and fullscreen exit before concealment.
• Manually verified in an isolated T3 Code (Dev) instance on macOS in both normal-window and native-fullscreen modes. In each mode the window disappeared at the threshold, the process quit after release, and the next app stayed open.
• Two Fable 5 review rounds checked the handler state machine and Electron focus behavior. The final implementation includes the resulting committed-gesture and fullscreen safeguards.

UI Changes

The existing hold hint is unchanged. The T3 window now disappears when the hold reaches its threshold instead of remaining visible until the release grace period ends.

No recording was captured because the verification used native keyboard input in the maintainer's isolated desktop session.

Checklist

• This PR is small and focused
• I explained what changed and why
• Static screenshots are not applicable because no layout or visual design changed
• I included a video for the interaction timing change

Implementation used GPT-5.6 Sol in T3 Code. Independent review used Fable 5 through Claude Code.

Note

Fix hold-to-quit getting stuck in makeQuitShortcutHandler

• A completed hold now conceals the window via concealPendingQuitWindow in DesktopWindow.ts, which exits fullscreen and sets opacity to 0 before waiting for key release.
• The handler in QuitHold.ts tracks observed auto-repeat cadence and waits for the larger of the fixed grace period or twice the repeat interval before timing out, so continued repeats postpone quitting instead of stalling.
• Committed holds stay committed through unrelated keydowns and fresh qualifying presses; only keyup or the quiet-period timeout completes the quit.
• Risk: QuitShortcutOptions now requires a conceal callback — all callers of makeQuitShortcutHandler must supply it or compilation fails.

^{Macroscope summarized f72b306.}

[macroscopeapp]

Approvability

Verdict: Approved at f72b306

Macroscope's review found this PR approvable — This is a focused desktop bug fix confined to the existing hold-to-quit handler and its window integration. It adds repeat-aware release handling and temporary concealment, with targeted coverage for normal, interrupted, missing-release, slow-repeat, and fullscreen cases.

^{You can add or adjust custom eligibility rules. Learn more.}

Dismissing prior approval to re-evaluate f72b306

[saphid]

What Changed

Completed hold-to-quit gestures now finish after Cmd+Q repeats stop, even when macOS omits the useful key-up event. At the 1.2-second threshold, T3 makes its focused window transparent so the interaction feels complete while the process continues consuming repeats. Native fullscreen exits before concealment. The process quits after release is safely detected.

The quiet timer follows the observed repeat cadence so slow keyboard settings do not send Cmd+Q into the next app. Once the window is concealed, other keys cannot cancel the committed quit and strand an invisible process.

This changes only the desktop window and shortcut handler. It does not change web, mobile, providers, or wire contracts.

Why

The original handler could wait forever after a completed hold when macOS suppressed the release event. Waiting for repeat silence fixed that hang, but left the window visible after the 1.2-second hold threshold and gave no clear indication that the gesture had completed.

Quitting the process at the threshold would reintroduce the shortcut spillover fixed in #7369. Concealing the still-focused window gives immediate visual completion while keeping T3 alive long enough to consume the remaining repeats.

Verification

• vp test run src/window/QuitHold.test.ts src/window/DesktopWindow.test.ts in apps/desktop: 51 passed, 0 failed.
• pnpm --filter @t3tools/desktop typecheck: passed.
• Targeted lint and format checks for the four changed desktop files: passed.
• Added regressions for missing release events, slow repeat cadences, concealment at the hold threshold, committed gestures surviving extra input, and fullscreen exit before concealment.
• Manually verified in an isolated T3 Code (Dev) instance on macOS in both normal-window and native-fullscreen modes. In each mode the window disappeared at the threshold, the process quit after release, and the next app stayed open.
• Two Fable 5 review rounds checked the handler state machine and Electron focus behavior. The final implementation includes the resulting committed-gesture and fullscreen safeguards.

UI Changes

The existing hold hint is unchanged. The T3 window now disappears when the hold reaches its threshold instead of remaining visible until the release grace period ends.

No recording was captured because the verification used native keyboard input in the maintainer's isolated desktop session.

Checklist

• This PR is small and focused
• I explained what changed and why
• Static screenshots are not applicable because no layout or visual design changed
• I included a video for the interaction timing change

Implementation used GPT-5.6 Sol in T3 Code. Independent review used Fable 5 through Claude Code.

Note

Fix hold-to-quit getting stuck in makeQuitShortcutHandler

• A completed hold now conceals the window via concealPendingQuitWindow in DesktopWindow.ts, which exits fullscreen and sets opacity to 0 before waiting for key release.
• The handler in QuitHold.ts tracks observed auto-repeat cadence and waits for the larger of the fixed grace period or twice the repeat interval before timing out, so continued repeats postpone quitting instead of stalling.
• Committed holds stay committed through unrelated keydowns and fresh qualifying presses; only keyup or the quiet-period timeout completes the quit.
• Risk: QuitShortcutOptions now requires a conceal callback — all callers of makeQuitShortcutHandler must supply it or compilation fails.

^{Macroscope summarized f72b306.}

[macroscopeapp]

Approvability

Verdict: Approved at f72b306

Macroscope's review found this PR approvable — This is a focused desktop bug fix confined to the existing hold-to-quit handler and its window integration. It adds repeat-aware release handling and temporary concealment, with targeted coverage for normal, interrupted, missing-release, slow-repeat, and fullscreen cases.

^{You can add or adjust custom eligibility rules. Learn more.}

Dismissing prior approval to re-evaluate f72b306

[saphid]

What Changed

Completed hold-to-quit gestures now finish after Cmd+Q repeats stop, even when macOS omits the useful key-up event. At the 1.2-second threshold, T3 makes its focused window transparent so the interaction feels complete while the process continues consuming repeats. Native fullscreen exits before concealment. The process quits after release is safely detected.

The quiet timer follows the observed repeat cadence so slow keyboard settings do not send Cmd+Q into the next app. Once the window is concealed, other keys cannot cancel the committed quit and strand an invisible process.

This changes only the desktop window and shortcut handler. It does not change web, mobile, providers, or wire contracts.

Why

The original handler could wait forever after a completed hold when macOS suppressed the release event. Waiting for repeat silence fixed that hang, but left the window visible after the 1.2-second hold threshold and gave no clear indication that the gesture had completed.

Quitting the process at the threshold would reintroduce the shortcut spillover fixed in #7369. Concealing the still-focused window gives immediate visual completion while keeping T3 alive long enough to consume the remaining repeats.

Verification

• vp test run src/window/QuitHold.test.ts src/window/DesktopWindow.test.ts in apps/desktop: 51 passed, 0 failed.
• pnpm --filter @t3tools/desktop typecheck: passed.
• Targeted lint and format checks for the four changed desktop files: passed.
• Added regressions for missing release events, slow repeat cadences, concealment at the hold threshold, committed gestures surviving extra input, and fullscreen exit before concealment.
• Manually verified in an isolated T3 Code (Dev) instance on macOS in both normal-window and native-fullscreen modes. In each mode the window disappeared at the threshold, the process quit after release, and the next app stayed open.
• Two Fable 5 review rounds checked the handler state machine and Electron focus behavior. The final implementation includes the resulting committed-gesture and fullscreen safeguards.

UI Changes

The existing hold hint is unchanged. The T3 window now disappears when the hold reaches its threshold instead of remaining visible until the release grace period ends.

No recording was captured because the verification used native keyboard input in the maintainer's isolated desktop session.

Checklist

• This PR is small and focused
• I explained what changed and why
• Static screenshots are not applicable because no layout or visual design changed
• I included a video for the interaction timing change

Implementation used GPT-5.6 Sol in T3 Code. Independent review used Fable 5 through Claude Code.

Note

Fix hold-to-quit getting stuck in makeQuitShortcutHandler

• A completed hold now conceals the window via concealPendingQuitWindow in DesktopWindow.ts, which exits fullscreen and sets opacity to 0 before waiting for key release.
• The handler in QuitHold.ts tracks observed auto-repeat cadence and waits for the larger of the fixed grace period or twice the repeat interval before timing out, so continued repeats postpone quitting instead of stalling.
• Committed holds stay committed through unrelated keydowns and fresh qualifying presses; only keyup or the quiet-period timeout completes the quit.
• Risk: QuitShortcutOptions now requires a conceal callback — all callers of makeQuitShortcutHandler must supply it or compilation fails.

^{Macroscope summarized f72b306.}

[macroscopeapp]

Approvability

Verdict: Approved at f72b306

Macroscope's review found this PR approvable — This is a focused desktop bug fix confined to the existing hold-to-quit handler and its window integration. It adds repeat-aware release handling and temporary concealment, with targeted coverage for normal, interrupted, missing-release, slow-repeat, and fullscreen cases.

^{You can add or adjust custom eligibility rules. Learn more.}

Dismissing prior approval to re-evaluate f72b306

[saphid]

What Changed

Completed hold-to-quit gestures now finish after Cmd+Q repeats stop, even when macOS omits the useful key-up event. At the 1.2-second threshold, T3 makes its focused window transparent so the interaction feels complete while the process continues consuming repeats. Native fullscreen exits before concealment. The process quits after release is safely detected.

The quiet timer follows the observed repeat cadence so slow keyboard settings do not send Cmd+Q into the next app. Once the window is concealed, other keys cannot cancel the committed quit and strand an invisible process.

This changes only the desktop window and shortcut handler. It does not change web, mobile, providers, or wire contracts.

Why

The original handler could wait forever after a completed hold when macOS suppressed the release event. Waiting for repeat silence fixed that hang, but left the window visible after the 1.2-second hold threshold and gave no clear indication that the gesture had completed.

Quitting the process at the threshold would reintroduce the shortcut spillover fixed in #7369. Concealing the still-focused window gives immediate visual completion while keeping T3 alive long enough to consume the remaining repeats.

Verification

• vp test run src/window/QuitHold.test.ts src/window/DesktopWindow.test.ts in apps/desktop: 51 passed, 0 failed.
• pnpm --filter @t3tools/desktop typecheck: passed.
• Targeted lint and format checks for the four changed desktop files: passed.
• Added regressions for missing release events, slow repeat cadences, concealment at the hold threshold, committed gestures surviving extra input, and fullscreen exit before concealment.
• Manually verified in an isolated T3 Code (Dev) instance on macOS in both normal-window and native-fullscreen modes. In each mode the window disappeared at the threshold, the process quit after release, and the next app stayed open.
• Two Fable 5 review rounds checked the handler state machine and Electron focus behavior. The final implementation includes the resulting committed-gesture and fullscreen safeguards.

UI Changes

The existing hold hint is unchanged. The T3 window now disappears when the hold reaches its threshold instead of remaining visible until the release grace period ends.

No recording was captured because the verification used native keyboard input in the maintainer's isolated desktop session.

Checklist

• This PR is small and focused
• I explained what changed and why
• Static screenshots are not applicable because no layout or visual design changed
• I included a video for the interaction timing change

Implementation used GPT-5.6 Sol in T3 Code. Independent review used Fable 5 through Claude Code.

Note

Fix hold-to-quit getting stuck in makeQuitShortcutHandler

• A completed hold now conceals the window via concealPendingQuitWindow in DesktopWindow.ts, which exits fullscreen and sets opacity to 0 before waiting for key release.
• The handler in QuitHold.ts tracks observed auto-repeat cadence and waits for the larger of the fixed grace period or twice the repeat interval before timing out, so continued repeats postpone quitting instead of stalling.
• Committed holds stay committed through unrelated keydowns and fresh qualifying presses; only keyup or the quiet-period timeout completes the quit.
• Risk: QuitShortcutOptions now requires a conceal callback — all callers of makeQuitShortcutHandler must supply it or compilation fails.

^{Macroscope summarized f72b306.}

[macroscopeapp]

Approvability

Verdict: Approved at f72b306

Macroscope's review found this PR approvable — This is a focused desktop bug fix confined to the existing hold-to-quit handler and its window integration. It adds repeat-aware release handling and temporary concealment, with targeted coverage for normal, interrupted, missing-release, slow-repeat, and fullscreen cases.

^{You can add or adjust custom eligibility rules. Learn more.}

Dismissing prior approval to re-evaluate f72b306

[saphid]

What Changed

Completed hold-to-quit gestures now finish after Cmd+Q repeats stop, even when macOS omits the useful key-up event. At the 1.2-second threshold, T3 makes its focused window transparent so the interaction feels complete while the process continues consuming repeats. Native fullscreen exits before concealment. The process quits after release is safely detected.

The quiet timer follows the observed repeat cadence so slow keyboard settings do not send Cmd+Q into the next app. Once the window is concealed, other keys cannot cancel the committed quit and strand an invisible process.

This changes only the desktop window and shortcut handler. It does not change web, mobile, providers, or wire contracts.

Why

The original handler could wait forever after a completed hold when macOS suppressed the release event. Waiting for repeat silence fixed that hang, but left the window visible after the 1.2-second hold threshold and gave no clear indication that the gesture had completed.

Quitting the process at the threshold would reintroduce the shortcut spillover fixed in #7369. Concealing the still-focused window gives immediate visual completion while keeping T3 alive long enough to consume the remaining repeats.

Verification

• vp test run src/window/QuitHold.test.ts src/window/DesktopWindow.test.ts in apps/desktop: 51 passed, 0 failed.
• pnpm --filter @t3tools/desktop typecheck: passed.
• Targeted lint and format checks for the four changed desktop files: passed.
• Added regressions for missing release events, slow repeat cadences, concealment at the hold threshold, committed gestures surviving extra input, and fullscreen exit before concealment.
• Manually verified in an isolated T3 Code (Dev) instance on macOS in both normal-window and native-fullscreen modes. In each mode the window disappeared at the threshold, the process quit after release, and the next app stayed open.
• Two Fable 5 review rounds checked the handler state machine and Electron focus behavior. The final implementation includes the resulting committed-gesture and fullscreen safeguards.

UI Changes

The existing hold hint is unchanged. The T3 window now disappears when the hold reaches its threshold instead of remaining visible until the release grace period ends.

No recording was captured because the verification used native keyboard input in the maintainer's isolated desktop session.

Checklist

• This PR is small and focused
• I explained what changed and why
• Static screenshots are not applicable because no layout or visual design changed
• I included a video for the interaction timing change

Implementation used GPT-5.6 Sol in T3 Code. Independent review used Fable 5 through Claude Code.

Note

Fix hold-to-quit getting stuck in makeQuitShortcutHandler

• A completed hold now conceals the window via concealPendingQuitWindow in DesktopWindow.ts, which exits fullscreen and sets opacity to 0 before waiting for key release.
• The handler in QuitHold.ts tracks observed auto-repeat cadence and waits for the larger of the fixed grace period or twice the repeat interval before timing out, so continued repeats postpone quitting instead of stalling.
• Committed holds stay committed through unrelated keydowns and fresh qualifying presses; only keyup or the quiet-period timeout completes the quit.
• Risk: QuitShortcutOptions now requires a conceal callback — all callers of makeQuitShortcutHandler must supply it or compilation fails.

^{Macroscope summarized f72b306.}

[macroscopeapp]

Approvability

Verdict: Approved at f72b306

Macroscope's review found this PR approvable — This is a focused desktop bug fix confined to the existing hold-to-quit handler and its window integration. It adds repeat-aware release handling and temporary concealment, with targeted coverage for normal, interrupted, missing-release, slow-repeat, and fullscreen cases.

^{You can add or adjust custom eligibility rules. Learn more.}

Dismissing prior approval to re-evaluate f72b306

[saphid]

What Changed

Completed hold-to-quit gestures now finish after Cmd+Q repeats stop, even when macOS omits the useful key-up event. At the 1.2-second threshold, T3 makes its focused window transparent so the interaction feels complete while the process continues consuming repeats. Native fullscreen exits before concealment. The process quits after release is safely detected.

The quiet timer follows the observed repeat cadence so slow keyboard settings do not send Cmd+Q into the next app. Once the window is concealed, other keys cannot cancel the committed quit and strand an invisible process.

This changes only the desktop window and shortcut handler. It does not change web, mobile, providers, or wire contracts.

Why

The original handler could wait forever after a completed hold when macOS suppressed the release event. Waiting for repeat silence fixed that hang, but left the window visible after the 1.2-second hold threshold and gave no clear indication that the gesture had completed.

Quitting the process at the threshold would reintroduce the shortcut spillover fixed in #7369. Concealing the still-focused window gives immediate visual completion while keeping T3 alive long enough to consume the remaining repeats.

Verification

• vp test run src/window/QuitHold.test.ts src/window/DesktopWindow.test.ts in apps/desktop: 51 passed, 0 failed.
• pnpm --filter @t3tools/desktop typecheck: passed.
• Targeted lint and format checks for the four changed desktop files: passed.
• Added regressions for missing release events, slow repeat cadences, concealment at the hold threshold, committed gestures surviving extra input, and fullscreen exit before concealment.
• Manually verified in an isolated T3 Code (Dev) instance on macOS in both normal-window and native-fullscreen modes. In each mode the window disappeared at the threshold, the process quit after release, and the next app stayed open.
• Two Fable 5 review rounds checked the handler state machine and Electron focus behavior. The final implementation includes the resulting committed-gesture and fullscreen safeguards.

UI Changes

The existing hold hint is unchanged. The T3 window now disappears when the hold reaches its threshold instead of remaining visible until the release grace period ends.

No recording was captured because the verification used native keyboard input in the maintainer's isolated desktop session.

Checklist

• This PR is small and focused
• I explained what changed and why
• Static screenshots are not applicable because no layout or visual design changed
• I included a video for the interaction timing change

Implementation used GPT-5.6 Sol in T3 Code. Independent review used Fable 5 through Claude Code.

Note

Fix hold-to-quit getting stuck in makeQuitShortcutHandler

• A completed hold now conceals the window via concealPendingQuitWindow in DesktopWindow.ts, which exits fullscreen and sets opacity to 0 before waiting for key release.
• The handler in QuitHold.ts tracks observed auto-repeat cadence and waits for the larger of the fixed grace period or twice the repeat interval before timing out, so continued repeats postpone quitting instead of stalling.
• Committed holds stay committed through unrelated keydowns and fresh qualifying presses; only keyup or the quiet-period timeout completes the quit.
• Risk: QuitShortcutOptions now requires a conceal callback — all callers of makeQuitShortcutHandler must supply it or compilation fails.

^{Macroscope summarized f72b306.}

[macroscopeapp]

Approvability

Verdict: Approved at f72b306

Macroscope's review found this PR approvable — This is a focused desktop bug fix confined to the existing hold-to-quit handler and its window integration. It adds repeat-aware release handling and temporary concealment, with targeted coverage for normal, interrupted, missing-release, slow-repeat, and fullscreen cases.

^{You can add or adjust custom eligibility rules. Learn more.}

Dismissing prior approval to re-evaluate f72b306

[saphid]

What Changed

Completed hold-to-quit gestures now finish after Cmd+Q repeats stop, even when macOS omits the useful key-up event. At the 1.2-second threshold, T3 makes its focused window transparent so the interaction feels complete while the process continues consuming repeats. Native fullscreen exits before concealment. The process quits after release is safely detected.

The quiet timer follows the observed repeat cadence so slow keyboard settings do not send Cmd+Q into the next app. Once the window is concealed, other keys cannot cancel the committed quit and strand an invisible process.

This changes only the desktop window and shortcut handler. It does not change web, mobile, providers, or wire contracts.

Why

The original handler could wait forever after a completed hold when macOS suppressed the release event. Waiting for repeat silence fixed that hang, but left the window visible after the 1.2-second hold threshold and gave no clear indication that the gesture had completed.

Quitting the process at the threshold would reintroduce the shortcut spillover fixed in #7369. Concealing the still-focused window gives immediate visual completion while keeping T3 alive long enough to consume the remaining repeats.

Verification

• vp test run src/window/QuitHold.test.ts src/window/DesktopWindow.test.ts in apps/desktop: 51 passed, 0 failed.
• pnpm --filter @t3tools/desktop typecheck: passed.
• Targeted lint and format checks for the four changed desktop files: passed.
• Added regressions for missing release events, slow repeat cadences, concealment at the hold threshold, committed gestures surviving extra input, and fullscreen exit before concealment.
• Manually verified in an isolated T3 Code (Dev) instance on macOS in both normal-window and native-fullscreen modes. In each mode the window disappeared at the threshold, the process quit after release, and the next app stayed open.
• Two Fable 5 review rounds checked the handler state machine and Electron focus behavior. The final implementation includes the resulting committed-gesture and fullscreen safeguards.

UI Changes

The existing hold hint is unchanged. The T3 window now disappears when the hold reaches its threshold instead of remaining visible until the release grace period ends.

No recording was captured because the verification used native keyboard input in the maintainer's isolated desktop session.

Checklist

• This PR is small and focused
• I explained what changed and why
• Static screenshots are not applicable because no layout or visual design changed
• I included a video for the interaction timing change

Implementation used GPT-5.6 Sol in T3 Code. Independent review used Fable 5 through Claude Code.

Note

Fix hold-to-quit getting stuck in makeQuitShortcutHandler

• A completed hold now conceals the window via concealPendingQuitWindow in DesktopWindow.ts, which exits fullscreen and sets opacity to 0 before waiting for key release.
• The handler in QuitHold.ts tracks observed auto-repeat cadence and waits for the larger of the fixed grace period or twice the repeat interval before timing out, so continued repeats postpone quitting instead of stalling.
• Committed holds stay committed through unrelated keydowns and fresh qualifying presses; only keyup or the quiet-period timeout completes the quit.
• Risk: QuitShortcutOptions now requires a conceal callback — all callers of makeQuitShortcutHandler must supply it or compilation fails.

^{Macroscope summarized f72b306.}

[macroscopeapp]

Approvability

Verdict: Approved at f72b306

Macroscope's review found this PR approvable — This is a focused desktop bug fix confined to the existing hold-to-quit handler and its window integration. It adds repeat-aware release handling and temporary concealment, with targeted coverage for normal, interrupted, missing-release, slow-repeat, and fullscreen cases.

^{You can add or adjust custom eligibility rules. Learn more.}

Dismissing prior approval to re-evaluate f72b306

[saphid]

What Changed

Completed hold-to-quit gestures now finish after Cmd+Q repeats stop, even when macOS omits the useful key-up event. At the 1.2-second threshold, T3 makes its focused window transparent so the interaction feels complete while the process continues consuming repeats. Native fullscreen exits before concealment. The process quits after release is safely detected.

The quiet timer follows the observed repeat cadence so slow keyboard settings do not send Cmd+Q into the next app. Once the window is concealed, other keys cannot cancel the committed quit and strand an invisible process.

This changes only the desktop window and shortcut handler. It does not change web, mobile, providers, or wire contracts.

Why

The original handler could wait forever after a completed hold when macOS suppressed the release event. Waiting for repeat silence fixed that hang, but left the window visible after the 1.2-second hold threshold and gave no clear indication that the gesture had completed.

Quitting the process at the threshold would reintroduce the shortcut spillover fixed in #7369. Concealing the still-focused window gives immediate visual completion while keeping T3 alive long enough to consume the remaining repeats.

Verification

• vp test run src/window/QuitHold.test.ts src/window/DesktopWindow.test.ts in apps/desktop: 51 passed, 0 failed.
• pnpm --filter @t3tools/desktop typecheck: passed.
• Targeted lint and format checks for the four changed desktop files: passed.
• Added regressions for missing release events, slow repeat cadences, concealment at the hold threshold, committed gestures surviving extra input, and fullscreen exit before concealment.
• Manually verified in an isolated T3 Code (Dev) instance on macOS in both normal-window and native-fullscreen modes. In each mode the window disappeared at the threshold, the process quit after release, and the next app stayed open.
• Two Fable 5 review rounds checked the handler state machine and Electron focus behavior. The final implementation includes the resulting committed-gesture and fullscreen safeguards.

UI Changes

The existing hold hint is unchanged. The T3 window now disappears when the hold reaches its threshold instead of remaining visible until the release grace period ends.

No recording was captured because the verification used native keyboard input in the maintainer's isolated desktop session.

Checklist

• This PR is small and focused
• I explained what changed and why
• Static screenshots are not applicable because no layout or visual design changed
• I included a video for the interaction timing change

Implementation used GPT-5.6 Sol in T3 Code. Independent review used Fable 5 through Claude Code.

Note

Fix hold-to-quit getting stuck in makeQuitShortcutHandler

• A completed hold now conceals the window via concealPendingQuitWindow in DesktopWindow.ts, which exits fullscreen and sets opacity to 0 before waiting for key release.
• The handler in QuitHold.ts tracks observed auto-repeat cadence and waits for the larger of the fixed grace period or twice the repeat interval before timing out, so continued repeats postpone quitting instead of stalling.
• Committed holds stay committed through unrelated keydowns and fresh qualifying presses; only keyup or the quiet-period timeout completes the quit.
• Risk: QuitShortcutOptions now requires a conceal callback — all callers of makeQuitShortcutHandler must supply it or compilation fails.

^{Macroscope summarized f72b306.}

[macroscopeapp]

Approvability

Verdict: Approved at f72b306

Macroscope's review found this PR approvable — This is a focused desktop bug fix confined to the existing hold-to-quit handler and its window integration. It adds repeat-aware release handling and temporary concealment, with targeted coverage for normal, interrupted, missing-release, slow-repeat, and fullscreen cases.

^{You can add or adjust custom eligibility rules. Learn more.}

Dismissing prior approval to re-evaluate f72b306