- Fix early return in RefreshWslDistros to still update with default-only list
- Store all PythonScriptActions (not just IsShown=true) so hide logic works
- Pass cancellation token to Task.Delay in debounce handler
- Use FileNotFoundException(message, fileName) constructor properly
- Display script SHA-256 hash in trust dialog for user verification
- Validate Python import names before embedding in shell commands
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Fix PythonScriptNotFound to format {0} placeholder and use PasteActionException
- Dispose old CancellationTokenSource before creating new one in debounce handler
- Remove .Wait() on UI-scheduled task (fire-and-forget is sufficient)
- Add WaitForExit after Kill() in RefreshWslDistros to prevent ReadToEnd blocking
- Localize '(System default)' WSL distro display name via resource string
- Kill child Python/WSL process on user cancellation to prevent orphaned processes
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Fix TryParseTag to support presence-based @advancedpaste:disabled tag
- Change 'if not input_value' to 'if input_value is None' (empty string is valid)
- Wrap HTML output with HtmlFormatHelper.CreateHtmlFormat for CF_HTML compliance
- Add try-catch for FileNotFoundException during ComputeHash
- Change else-if to sequential ifs in DataPackageFromViewAsync (preserve all formats)
- Remove unused System.* PackageReferences from UITest project
- Fix '..' typo to '...' in search placeholder text
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Fix ScriptsFolder to fall back to default folder when setting is blank
- Align Settings tag parsing with runtime: use @advancedpaste:disabled
- Remove misleading 'written back to headers' description string
- Add modifier release/restore to SendInput Ctrl+C path (matches Ctrl+V)
- Sanitize pip package names to prevent shell metacharacter injection
- Fix RefreshWslDistros to WaitForExit before ReadToEnd (prevents hang)
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Remove KernelFunctionDescription from PythonScript to prevent broken AI kernel registration
- Normalize file/files format names in _runner.py so files-input scripts match correctly
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Restore FuzzTests to use UTF-8 decoding and GetAwaiter().GetResult()
- Restore defensive try/catch in JsonHelper.ToJsonFromXmlOrCsvAsync
- Add null guard for ReadMetadata in PasteFormatExecutor
- Restore is_enabled_by_default() override in dllmain.cpp
- Localize hard-coded strings in AdvancedPastePage.xaml via x:Uid
- Fix _runner.py docstring to include audio/video input types
- Fix typos in UITestAdvancedPaste.md
- Filter Settings script list to match runtime discovery behavior
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Enable users to write custom clipboard transformation scripts in Python
that integrate directly into the Advanced Paste menu. Scripts are auto-
discovered from a configurable folder and can run on native Windows
Python or inside WSL.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #48840
- [x] Closes: #32437
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
The following additions and changes were made in response to the
feedback given in #48840 and also following an audit of the IPA set.
#### IPA
| Character | Action
|--|--|
| `ɱ` | Added to **M** |
| `ʍ` | Added to **W** |
| `ɚ` | Added to **E** |
| ` ͡ ` and ` ͜ ` | Added to **PERIOD**
| `ɡ` | Added to **G**
| `ɫ` | Added to **I**
| `ʱ` | Added to **H**
| `◌̝`, `◌̥`, `◌̚`, `ˈ` and `ˌ` | added to **PERIOD**
| `ʎ` | Added to **Y**
| `ʔ` | Added to **COMMA** and **SLASH**
| `æ` | Added to **A** and **E**
| `œ` | Added to **O** and **E**
| `ʘ` | Added to **B** and **O**
| `β`, `ɓ` | Added to **B**
| `χ` | Added to **C** and **X**
| `ç`, `ǂ` | Added to **C**
| `ð`, `ɗ`, `ɖ`, `ǀ` | Added to **D**
| `ɠ`, `ʛ` | Added to **G**
| `ħ`, `ɥ`, `ɧ` | Added to **H**
| `ʄ` | Added to **J**
| `ɫ`, `ǁ` | Added to **L**
| `ø` | Added to **O**
I removed the caron vowel characters `ǎ`, `ǒ` and `ǔ`, as they should
not have been in the IPA set. These characters are available in the
Pinyin set.
A small number of keys had entries reordered where common mappings would
have been towards the end.
This IPA update includes:
- Click Consonants (`ʘ`, `ǀ`, `ǃ`, `ǂ`, `ǁ`), which are used in the
phonetic transcription of Southern and Eastern African languages, most
notably the Khoisan language groups and several Bantu languages (like
Zulu and Xhosa).
- Implosives and Ejectives (`ɓ`, `ɗ`, `ʼ`, etc.), which are essential
for transcribing languages across the globe, including Indigenous
languages of the Americas (e.g., Navajo and Mayan), the Caucasus (e.g.,
Georgian), Southeast Asia (e.g., Vietnamese), and widely across the
African continent.
#### SPECIAL set
| Character | Action | Notes
|--|--|--|
| `‽` and `⸘` | Added to **SLASH** |
| `⟨`, `⟩`, `⟪` and `⟫` | Replaced full-width CJK brackets with the
Western versions.
| `‰` and `‱` | Added to **P**
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
Manual testing, with specific tests to confirm that the new combining
marks display OK in the UI.
PowerToys has deeply nested source paths that exceed the legacy
260-character MAX_PATH limit. Contributors who haven't enabled Windows
long path support hit cryptic 'path too long' / 'could not find file'
errors during their first build.
Add an EnsureLongPathsEnabled MSBuild target in Directory.Build.targets
that reads
HKLM\SYSTEM\CurrentControlSet\Control\FileSystem\LongPathsEnabled and
fails fast with an actionable error (PTLONGPATH) pointing at
tools\build\setup-dev-environment.ps1. Covers both Visual Studio and the
command-line build scripts, skips design-time builds, and can be
bypassed with /p:SkipLongPathsCheck=true.
**What happens if Long file path isn't enabled.**
<img width="824" height="916" alt="image"
src="https://github.com/user-attachments/assets/30731f65-4011-48c0-94b9-e521b4c7d266"
/>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Opening the More button caused Narrator to read a long cascade of
overlapping announcements — popup window, filter TextBox placeholder,
ListView item name, position ("1 of 4"), keyboard shortcut text — all
from a single keypress with no further input.
This PR replaces that cascade with a single, clean announcement:
**"Menu, {0} commands. {1}, {2} of {0}"**
(num of commands, first item in list, num of order in list, num of
commands)
**"Menu, 3 commands. Calculator, 1 of 3."**
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #48899
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
| File | Change |
|------|--------|
| `ContextMenu.xaml` | Set `AccessibilityView="Raw"` on ListView and
TextBox; added `NarratorAnnouncer` TextBlock for raising UIA
notifications |
| `ContextMenu.xaml.cs` | Added `AnnounceOpened()`,
`AnnounceSelectedItem()`, and `_isOpening` guard to prevent programmatic
selection changes from triggering UIA events during the flyout
transition |
| `CommandBar.xaml.cs` | Call `AnnounceOpened()` on flyout open |
| `DockControl.xaml.cs` | Same for dock context menu |
| `Resources.resw` | Added `ScreenReader_Announcement_ContextMenuOpened`
localized format string |
## How it works
- ListView and TextBox are permanently `AccessibilityView="Raw"` —
invisible
to Narrator, preventing all system-driven UIA announcements
- A zero-size `NarratorAnnouncer` TextBlock
(`AccessibilityView="Content"`,
`LiveSetting="Assertive"`) serves as the sole notification source
- On open: a deferred `RaiseNotificationEvent` fires one consolidated
announcement
- On arrow navigation: `AnnounceSelectedItem()` fires item name and
position
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
https://github.com/user-attachments/assets/4660741f-6b91-4a32-9a56-83c64343b67a
- [x] Build succeeds (0 errors)
- [x] 124 ViewModel unit tests pass
- [x] Manual Narrator testing: open, arrow navigate, close, reopen
- [x] Keyboard filtering (type to filter) still works
- [x] Enter to invoke selected command still works
- [x] Escape to close still works
## Summary of the Pull Request
When the OS theme differs from the PowerToys theme (e.g. OS in Light,
PowerToys set to Dark), the **What's New / release-notes (Scoobe)** page
renders heading titles and hyperlinks with brushes pinned to the wrong
theme, making them unreadable. The system caption buttons
(min/max/close) are also not tinted to match the window theme.
The release-notes page uses the CommunityToolkit `MarkdownTextBlock`,
which captures its heading and link brushes from
`Application.Current.Resources` when its theme config is created. Those
resolve against the **OS (application) theme** rather than the window's
selected element theme, so headings/links break whenever the two themes
differ.
## PR Checklist
- [x] Closes: #43970
- [x] Closes: #48832
- [x] **Communication:** Local workaround for a known upstream control
bug
- [ ] **Tests:** Manually validated (UI/theming change)
- [x] **Localization:** No new end-user-facing strings
- [ ] **Dev docs:** N/A
## Detailed Description of the Pull Request / Additional comments
These are deliberately **local workarounds** until the upstream control
resolves brushes against the element theme via
[CommunityToolkit/Labs-Windows#785](https://github.com/CommunityToolkit/Labs-Windows/pull/785).
Comments + `TODO`s in the code point at that PR so the workaround can be
removed once it ships.
**`ScoobeReleaseNotesPage`**
- Pin the `MarkdownTextBlock.RequestedTheme` to the selected app theme
and reassign the `H1`–`H6` heading brushes and the link brush (resolved
for that theme) before the markdown is rendered. The themed brushes are
read from the control's own `Foreground` (`TextFillColorPrimaryBrush`)
and a hidden `LinkBrushProvider` carrier element
(`AccentTextFillColorPrimaryBrush`).
- Re-run the workaround and force a re-render on runtime theme changes
(subscribe to the page's `ActualThemeChanged`), so titles/links stay
readable when the user switches Light/Dark while the window is open.
**`TitleBarHelper` (new) + `ScoobeWindow`**
- Add a small shared
`TitleBarHelper.ApplySystemThemeToCaptionButtons(window, theme)` (port
of the WinUI Gallery helper + PowerToys conventions) and drive the
Scoobe window's caption-button colors from the content's actual theme,
updating on `ActualThemeChanged`. `ScoobeWindow` uses the built-in WinUI
`TitleBar`, which — unlike the custom PowerToys `TitleBar` control used
by the other Settings windows — does not tint the system caption buttons
to the app theme.
## Validation Steps Performed
- OS Light + PowerToys Dark: opened What's New → headings, hyperlinks,
body text and caption buttons all render readable/dark-themed
(previously black/unreadable titles). Confirmed working.
- Switched OS Light → Dark while the Scoobe window was open → markdown
content and caption buttons update live.
- OS Dark: content pane, tables and titles render with the dark theme
(addresses the "light pane in a dark window" report).
<img width="1673" height="929" alt="Screenshot 2026-06-26 130638"
src="https://github.com/user-attachments/assets/1db7fb37-f5ee-485b-863e-fc1ba0d13f6f"
/>
_OS is in Light Mode, while the app settings are set to Dark mode_
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Adds a Settings option that controls how much the PowerDisplay flyout
sliders (brightness, contrast, volume) change per mouse-wheel notch. The
value is chosen from a **preset dropdown** (`1, 2, 5, 10, 15, 20, 25`)
and defaults to **5**, preserving today's behavior. Previously the
per-notch step was hardcoded as
`helpers:SliderExtensions.MouseWheelChange="5"` in four places in the
flyout.
<img width="1282" height="620" alt="image"
src="https://github.com/user-attachments/assets/3b299a47-eb7b-4b53-b3dc-0540fbb25bfc"
/>
## PR Checklist
- [x] Closes: #48805
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass — _added
`MouseWheelIncrementSettingsTests` to the existing
`PowerDisplay.Lib.UnitTests`; passing via `vstest.console.exe`._
- [x] **Localization:** All end-user-facing strings can be localized —
_new `PowerDisplay_MouseWheelIncrement.Header`/`.Description` in
`en-us/Resources.resw`, surfaced via `x:Uid`._
- [ ] **Dev docs:** Added/updated — _N/A: small settings addition, no
behavioral/architecture docs affected._
- [ ] **New binaries:** Added on the required places — _N/A: no new
binaries or projects (the unit test was added to an existing test
project)._
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** _N/A._
## Detailed Description of the Pull Request / Additional comments
A single `int MouseWheelIncrement` is added to `PowerDisplayProperties`
and persisted to `settings.json` under `mouse_wheel_increment`; it is
edited in the Settings UI and read by the flyout app. The value applies
uniformly to all four flyout sliders.
**Data model** —
`src/settings-ui/Settings.UI.Library/PowerDisplayProperties.cs`
- `MouseWheelIncrement` (`int`,
`[JsonPropertyName("mouse_wheel_increment")]`, default `5` set in the
constructor, mirroring `MonitorRefreshDelay`). Old `settings.json`
without the key deserializes to `5` (no migration). The whole type is
already registered with the source-generated JSON contexts, so the new
property serializes on both the Settings and flyout sides with no
context change.
**Settings UI (write side)** — `src/settings-ui/Settings.UI/...`
- `PowerDisplayViewModel`: `MouseWheelIncrement` (get/set via
`SetSettingsProperty`, calls `SignalSettingsUpdated()` so an open flyout
updates live) and `MouseWheelIncrementOptions` = `{ 1, 2, 5, 10, 15, 20,
25 }`.
- `PowerDisplayPage.xaml`: a `ComboBox` `SettingsCard` in the
flyout-settings expander, immediately after "Monitor refresh delay",
matching that card's markup.
- `Strings/en-us/Resources.resw`: `Mouse wheel increment` + description.
**Flyout (read side)** — `src/modules/powerdisplay/PowerDisplay/...`
- `MainViewModel`: `[ObservableProperty] int MouseWheelIncrement`
(default 5), loaded from settings in `LoadUIDisplaySettings()` (runs at
startup and on the settings-updated IPC event, so the all-displays
slider updates live).
- `MonitorViewModel`: a read-only proxy `MouseWheelIncrement =>
_mainViewModel?.MouseWheelIncrement ?? 5` plus
`RefreshMouseWheelIncrement()`, called from `ApplySettingsFromUI`'s
per-monitor loop so the per-monitor sliders update live.
- `MainWindow.xaml`: the four sliders'
`SliderExtensions.MouseWheelChange` now bind to the setting — the
all-displays slider to `ViewModel.MouseWheelIncrement`, the three
per-monitor sliders (brightness/contrast/volume, inside the
`MonitorViewModel` `DataTemplate`) to `MouseWheelIncrement`.
## Validation Steps Performed
**Automated**
- Unit tests
(`PowerDisplay.Lib.UnitTests/MouseWheelIncrementSettingsTests.cs`), run
via `vstest.console.exe` — passing:
- default value is `5`;
- legacy `settings.json` missing the key deserializes to `5` (no
migration);
- round-trip preserves a non-default value;
- serialization emits the `mouse_wheel_increment` snake_case key.
- `PowerToys.Settings` and the PowerDisplay flyout app both compile
clean (the XAML compiler validates the new `x:Bind` bindings).
- Confirmed the property flows through the source-generated JSON
contexts (whole-type `[JsonSerializable(typeof(PowerDisplaySettings))]`)
on both the Settings and flyout sides.
**Pending (manual, on-device — reason this PR is a draft)**
- [ ] Settings: the dropdown shows `5` on a fresh profile; changing it
writes the new `mouse_wheel_increment` value to `settings.json`.
- [ ] Flyout: scrolling each slider (all-displays brightness,
per-monitor brightness/contrast/volume) steps by the selected value,
including live update while the flyout is open.
---------
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
## Summary
Addresses #48924. Several MSIX context-menu items did not ship the full
standard logo set (e.g. File Locksmith was missing the 44x44 logo), so
Windows fell back to a default icon in some surfaces.
This updates the MSIX assets for **File Locksmith**, **Image Resizer**
and **PowerRename** to provide the full logo set: 44x44 logo,
small/large/150x150 tiles, and store logo.
## PR Checklist
- [x] Closes#48924
- [ ] Tested manually
## Validation
Icon-asset only change; no code modified.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Adds a reference to the WinGet Manifest Keyboard Shortcuts schema spec
in `.github/copilot-instructions.md` so AI agents know where to find the
correct field definitions, file naming conventions, and the `+` prefix
rule when creating or editing Shortcut Guide V2 manifest files.
## Summary of the Pull Request
Adds a new `## Shortcut Guide V2 Manifests` section to
`.github/copilot-instructions.md` linking to [`doc/specs/WinGet Manifest
Keyboard Shortcuts
schema.md`](../doc/specs/WinGet%20Manifest%20Keyboard%20Shortcuts%20schema.md).
This ensures agents authoring new manifest files follow the correct
schema and naming scheme (e.g., `<PackageId>.<locale>.yml`, `+` prefix
for apps without a WinGet package).
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
`.github/copilot-instructions.md` gains a dedicated section:
```markdown
## Shortcut Guide V2 Manifests
When creating or editing Shortcut Guide keyboard shortcut manifest files, follow the schema and naming conventions in the spec:
- [WinGet Manifest Keyboard Shortcuts schema](<../doc/specs/WinGet Manifest Keyboard Shortcuts schema.md>) – manifest file format, field definitions, file naming, and the `+` prefix convention for apps without a WinGet package
```
No production code changes.
## Validation Steps Performed
- Verified the relative link resolves to the correct spec file in the
repository.
- Confirmed the section is correctly placed before "Detailed
Documentation".
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
## Summary
Extension Gallery detail pages crash with `LayoutCycleException` /
`AG_E_LAYOUT_CYCLE` when scrolling the screenshot strip horizontally on
extensions that have many screenshots.
## Root Cause
The screenshot `ItemsView` uses a horizontal `StackLayout` and is placed
in a Grid row with `Height="Auto"`, inside a vertical `ScrollViewer`.
The `ItemsView` itself contains an internal `ScrollView` for horizontal
scrolling.
When screenshots load asynchronously (`BitmapImage` decode completes),
this triggers a layout feedback loop:
1. Image decodes → `ItemContainer` re-measures
2. Auto-height Grid row re-measures, offering `ItemsView` new available
height
3. `ItemsView`'s internal `ScrollView` recalculates extents
4. Parent Grid row invalidates → outer `ScrollViewer` re-layouts →
re-measures children → back to step 1
## Fix
Two XAML-only changes to `ExtensionGalleryItemPage.xaml`:
1. **Fixed Grid row height**: Changed the screenshot strip row from
`Height="Auto"` to `Height="232"` (200px image + 16px top/bottom
padding). The parent no longer queries `ItemsView` for desired height,
breaking the cycle.
2. **Fixed item width**: Added `Width="356"` to the screenshot `Border`
template (16:9 ratio: 200 × 16/9 ≈ 356). Provides a stable size before
async image decode, preventing re-measure triggers.
Fixes#47901
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Fixes#47719
VS Code stores recently opened workspaces as URIs. A workspace on a
Windows network share can be stored as
`file://server/share/workspace.code-workspace`.
The VS Code Workspaces plugin interpreted the server part of that URI as
a VS Code remote authority. Because it was not a recognized remote
authority, valid UNC workspaces were discarded.
This change recognizes file URIs with a server authority as local UNC
paths and converts them to the Windows UNC format: `\\server\share\...`.
## PR Checklist
* [x] Closes#47719
* [x] **Communication:** This implementation follows the suggested
direction in the issue discussion.
* [ ] **Tests:** No automated regression test added.
* [x] **Localization:** No user-facing strings were changed.
* [x] **New binaries:** No binaries were added.
* [x] **Documentation updated:** No documentation changes are required.
## Validation Steps Performed
* Built the full PowerToys solution locally in `Release | x64` with 0
errors.
* Started the locally built PowerToys instance.
* Verified that the VS Code Workspaces plugin finds local workspaces.
* Verified that UNC workspaces using both a hostname and an IP address
appear in PowerToys Run.
* Opened a UNC workspace successfully from PowerToys Run.
## Summary
Refactors the reusable transparent-overlay infrastructure in
`src/common/Common.UI.Controls/` into a clean separation between a pure
host window and a self-animating acrylic surface.
### What changed
- **`TransparentWindow`** is now animation-agnostic. It raises `Showing`
/ `Hiding` events; `Hiding` exposes a deferral so the HWND stays visible
until the surface's out-animation finishes.
- **`TransientSurface`** (renamed from `TransparentCard`) is a
self-animating "pseudo-window" content control. It owns all chrome —
`ThemeShadow`, always-active desktop acrylic, 1px border, rounded
corners — and its own show/hide slide animations.
- `SlideFrom` (`None`/`Left`/`Top`/`Right`/`Bottom`) selects the slide
edge. `None` is the default and plays **no animation at all** (instant
show/hide).
- `AcrylicKind` (new) is exposed and bound to the backdrop via
`TemplateBinding`, defaulting to **thin acrylic**. Consumers can
override to `Default`/`Base`.
- **`AlwaysActiveDesktopAcrylicBackdrop`** gains a matching `Kind`
dependency property.
- **CmdPal `ToastWindow`** is migrated to the new pattern as the proving
consumer (`Surface.SubscribeTo(this)`).
### Coordination model
A module declares a `<TransientSurface>` as the window's content and
calls `SubscribeTo(window)` once. The window raises `Showing`/`Hiding`;
the surface animates itself in/out and uses the `Hiding` deferral to
keep the window alive until the out-animation completes.
## Testing
- `Common.UI.Controls` builds clean (x64 Debug, exit 0).
- `Microsoft.CmdPal.UI` builds clean (x64 Debug, exit 0).
- ToastWindow keeps its slide-up animation (`SlideFrom="Bottom"`).
https://github.com/user-attachments/assets/a06b0f1a-740a-4fcd-bba8-6f7a64ed261b
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Adds a follow-up metadata fix to the existing Mouse Highlighter ripple
v2.1 work by allowing the term `xhair` in repo spell-check
configuration.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
- Added `xhair` to `.github/actions/spell-check/expect.txt`.
- This addresses spelling feedback on MouseHighlighter ripple/crosshair
code without changing runtime behavior.
- No functional changes to Mouse Highlighter logic were made in this
follow-up commit.
## Validation Steps Performed
- Verified the only content change is the new `xhair` entry in
spell-check expected words.
- Ran secret scanning on changed file
(`.github/actions/spell-check/expect.txt`) with no findings.
- Ran parallel validation:
- Code Review: no issues.
- CodeQL: skipped as trivial metadata-only change.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Niels Laute <niels.laute@live.nl>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Fallback results were showing in Command Palette search results even
when the user had disabled them in settings.
When a fallback command is disabled (e.g., VS Code for Command Palette
with `IsEnabled = false`), it is excluded from
`configuredGlobalFallbackIds` (which only contains enabled + global
fallbacks). However, during search filtering, all fallbacks NOT in
`configuredGlobalFallbackIds` were unconditionally added to
`commonFallbacks`, which gets scored and displayed in results. This
means disabled fallbacks still appeared.
To fix, I added an `IsEnabled` check when building the `commonFallbacks`
list in `MainListPage.cs`. Disabled fallback commands are now properly
excluded from search results.
Fixes#48504
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Selecting an extension in the CmdPal Extension Gallery crashed the app
when that extension had **no `homepage`** defined.
## Root cause
In `ExtensionGalleryItemPage.xaml`, the "View repository"
`HyperlinkButton` bound its `NavigateUri` (a `System.Uri`) directly to
the raw string property `ViewModel.Homepage`:
```xml
NavigateUri="{x:Bind ViewModel.Homepage, Mode=OneWay}"
```
`x:Bind` evaluates **all** bindings on an element regardless of
`Visibility`, and to assign a `string` to a `Uri` target it generates a
`new Uri(value)` conversion. When `homepage` is undefined, `Homepage` is
`null`, so the binding executes `new Uri(null)` →
`ArgumentNullException` → the page crashes on load. The
`Visibility="{... HasHomepage}"` collapse did not help because the
`NavigateUri` binding still runs.
Every other `NavigateUri` x:Bind in the codebase (`Link`, `LinkUri`) is
already bound to a `Uri?`, so the homepage hyperlink was the lone
offender.
## Fix
- **`ExtensionGalleryItemViewModel.cs`** — Added a validated `Uri?
HomepageUri` property backed by the existing `_homepageHttpUri` (already
`null` for missing or non-web homepages).
- **`ExtensionGalleryItemPage.xaml`** — Bound `NavigateUri` to
`ViewModel.HomepageUri` instead of the raw string. `null` is valid for
`NavigateUri`, so no conversion occurs. The tooltip still shows the
`Homepage` string.
- **Tests** — Added coverage asserting `HomepageUri` is set for a web
homepage and `null` when missing.
## Verification
- `Microsoft.CmdPal.UI.ViewModels`, `Microsoft.CmdPal.UI` (XAML
compile), and the unit test project all built cleanly (x64/Release, exit
code 0).
- All 24 `ExtensionGalleryItemViewModelTests` pass, including the two
new cases.
- Manually verified in VS that opening an extension without a homepage
no longer crashes.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
https://github.com/user-attachments/assets/b268bafb-6bee-4862-9fbf-7a0e06675e36
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Fixes a memory leak in the Performance Monitor dock extension where
`GetItems()` created **new** `ListItem` instances for `_networkUpItem`
and `_networkDownItem` on every call.
## Problem
When the dock subscribes to `ItemsChanged` and calls `GetItems()` to
refresh, the band page path allocates 2 new `ListItem` objects each time
— the old ones are replaced in the fields but never collected (they
remain referenced by the `DockItemViewModel` wrappers until the next
refresh cycle). Under normal operation this leaks ~2 objects/second
indefinitely.
## Fix
Move `_networkUpItem`/`_networkDownItem` creation into the constructor
(matching the pattern used by CPU, Memory, GPU, and Battery items).
`GetItems()` now returns stable references. The `Updated` event handler
already updates their `.Title` properties, which propagates to the UI
via `PropChanged` → `CommandItemViewModel.Model_PropChanged`.
## Validation
- Build succeeds (`Microsoft.CmdPal.Ext.PerformanceMonitor.csproj`)
- Network up/down band items still receive title updates via the
existing `Updated` handler
- No `RaiseItemsChanged()` needed — `ListItem.Title` setter fires
`PropChanged`, which `DockItemViewModel` already observes
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
The auto-labeler workflow now skips pull requests that already have
labels applied before running the AI classification. This avoids
overwriting or duplicating labels that were manually set by contributors
or maintainers.
## Changes
- Added a check in `labelIssue()` that returns early for PRs with
existing labels, logging which labels are already present.
- Issues continue to be labeled regardless (only PRs get the skip
logic).
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Update notices for ZoomIt dependency
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
Fixes#48698 by preventing the Command Palette settings frame from
navigating to the same page again, which avoids adding a self-navigation
entry to the back stack.
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Fixes stuck modifier keys and silently-dropped remaps on Keyboard
Manager's **single key → text** path, and adds unit coverage (including
a mockable injection-failure seam).
## What this changes
1. **Insert a dummy key event before releasing held modifiers.**
Releasing a lone Win or Alt key-up otherwise triggers the Start Menu /
menu bar. The dummy key absorbs it so the release is inert. The dummy +
releases are only injected when a modifier is actually held.
2. **Accept `WM_SYSKEYDOWN` as well as `WM_KEYDOWN`.** While Alt is held
the system delivers `WM_SYSKEYDOWN`, so the previous `WM_KEYDOWN`-only
guard silently dropped the remap whenever Alt was down.
3. **Route `Helpers::SendTextInput` through `InputInterface`** instead
of calling Win32 `SendInput` directly. Besides making the path mockable,
this stops the existing unit tests from injecting real keystrokes into
the OS during a test run. Text is still flushed per character to
preserve the existing batching workaround.
4. **Never re-press released modifiers.** Once a modifier key-up is
injected, `GetAsyncKeyState` reports it as up, so re-pressing risks
leaving it stuck if the user let go during injection. Leaving it
released is always safe.
## Testing
- New `MockedInput` failure seam (`SetSendVirtualInputShouldFail`).
- `RemappedKey_ShouldPassOriginalKeyThrough_WhenInjectionFails` —
verifies the original key is passed through when injection fails (the
core stuck-key behavior, previously untestable because the mock always
succeeded).
-
`HandleSingleKeyToTextRemapEvent_ShouldFireAndReleaseAlt_WhenAltIsHeld`
— covers fix#2 by asserting the remap still fires (and releases the
held Alt) when the key arrives as `WM_SYSKEYDOWN`.
- Full Keyboard Manager engine suite: **98/98 passing**, Release x64,
against current `main`.
This is one of a small set of related "stuck key" hardening fixes; each
stands alone.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This is a bear of a PR. Watch out.
This PR adds support for compact mode to the command palette. In compact
mode, the results of the command palette window are collapsed by
default. And the only thing that is visible is the search bar. When the
user types, the window is expanded only just enough to show the
available results[^1]
https://github.com/user-attachments/assets/fd11bbc9-1173-426f-8f44-b513baf2ac5f
This is made possible by a fairly annoyingly substantial refactoring to
how our windowing is done. Animating the size or bounds of an HWND on
Windows is not fun. With pretty much any XAML application, you're going
to get at least one frame of blackness when you resize your window. The
trick then is to create an experience where it looks like your
application is resizing, but the HWND never actually resizes.
So the main bulk of this PR is actually just refactoring our window
handling. Our `MainWindow` class now becomes a dummy holder of _some
content_, and most of the main content is moved into the
`CmdPalMainControl` class. `MainWindow`'s job then is to handle a
transparent window that hosts some XAML content inside of it, and
pretend that content is the real bounds of the window. We need to fake
our NCHITTEST results, so that the edges of the XAML content act like
they're the edge of the actual HWND. We need to hide our actual window
frame and shadow, but then also re-create them in XAML around our
content.
Previously we've done work like this using a single full screen
transparent window with XAML content inside of it. However that has the
downside of not allowing the XAML content to be movable across different
monitors. By faking out the NCHITTEST results, we allow users to resize
and move the window using the normal user32 move size loop.
Our HWND is also cropped (with SetWindowRgn to the bounds of the shadow
around our XAML content. We need to include the shadow in the hit
testable region, because if we don't, then the shadow will be visibly
cropped on the edges.
In compact mode, instead of centering our window in the middle of the
monitor, the user can set a relative height where the search box opens
on that monitor. This defaults to about 60% up from the bottom of the
monitor, so that there's room for the results to expand downwards and
feel centered within the screen. This position is a setting, so users
can customize it to whatever they like.
I've also added a developer only debug build only internal setting,
which allows you to see the actual frame of our HWND. This makes it
easier to visually debug where the bounds of the window are and
understand a little bit more about the layout of our application. This
setting and functionality is disabled in release builds.
<img width="1334" height="1164" alt="cmdpal-compact-diagram"
src="https://github.com/user-attachments/assets/cb1c273d-37cc-4cb7-8680-e1878aa20c9c"
/>
Closes#38423
[^1]: with some caveats: pages with details expand fully always.
Command Palette currently shows the "Uninstall" context menu option for
system apps like Windows Settings, Registry Editor, and Task Manager.
Users can accidentally trigger uninstall on packages that should not be
removable. This PR hides the uninstall option for system apps.
It's probably not a perfect solution, but seemed to catch all I tried to
test.
Fixes#46826
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Closes#48677.
## Root cause
`GPUStats.GetData()` summed the cooked `Utilization Percentage` value of
every `engtype=3D` GPU-Engine performance-counter instance for a given
physical adapter. Each instance is a `(process, engine)` pair. Summing
across processes for a single engine is correct (gives total engine
utilization), but summing across **multiple 3D engines** on the same
adapter compounds them and produces values >100% (the 104%/118%
screenshots in the issue). The displayed value `sum / 100` therefore
exceeded 1.0, and the chart fed the raw inflated `sum` directly.
## Fix
Mirroring the bounded-counter approach used for CPU in #46381:
1. Bucket cooked values by `(physId, engineId)` rather than just
`physId`.
2. Reduce per-adapter to the **maximum** engine utilization (what Task
Manager surfaces as the overall GPU number).
3. Clamp the per-adapter value to `[0, 100]` before storing it as a 0–1
fraction in `Data.Usage` and pushing to the chart.
4. Defensive checks: skip `NaN` / `Infinity` / negative cooked values,
and skip instances whose engine id can't be parsed.
Co-authored-by: root <root@io.bbq>
## Summary
Bookmarks pinned to the CmdPal dock disappear after a restart.
## Root Cause
`BookmarksCommandProvider` did not override `GetCommandItem(string id)`.
When `InitializeCommands` in `CommandProviderWrapper` processes pinned
dock band settings, the fallback path calls
`provider.GetCommandItem(commandId)`, which hit the base-class no-op
returning `null`. Any bookmark band that `GetDockBands()` failed to
return (due to a race, transient file error, or exception) was
irreversibly lost from the dock.
## Fix
Added `GetCommandItem` override to `BookmarksCommandProvider`. Given a
`"Bookmarks.Launch.{guid}"` ID, it:
1. Parses the GUID suffix
2. Looks up the `BookmarkData` in `_bookmarksManager` (under lock)
3. Constructs a `BookmarkListItem` (with `asBand: true`) and wraps it in
a `WrappedDockItem`
Uses `BookmarkTitle` (maps to `_bookmark.Name`, set synchronously in the
constructor) rather than `Title` (set asynchronously after
classification completes), matching the pattern used by
`AllAppsCommandProvider`, `SystemCommandExtensionProvider`, and others.
Fixes#47576
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Win+Q and Win+S were both labeled "Open search" in Shortcut Guide. On
Copilot+ PCs, Win+Q launches Click to Do — not Search — so the duplicate
label was misleading.
## Changes
- `+WindowsNT.Shell.en-US.yml`: Renamed Win+Q entry from `Open search` →
`Open Click to Do` with description `On Copilot+ PCs`; Win+S retains
`Open search`
## PR Checklist
- [ ] Closes: #xxx
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
The Windows key shortcuts manifest had two entries with identical names
(`Open search`) mapped to Win+Q and Win+S respectively. On Copilot+ PCs,
Win+Q opens Click to Do, not Search. The Win+Q entry is corrected to
`Open Click to Do` with a `On Copilot+ PCs` description to scope its
applicability.
## Validation Steps Performed
- Visually reviewed the YAML diff to confirm only the Win+Q entry name
and description changed; Win+S entry is unchanged.
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
## Summary
Fixes#46693
- The `static bool isAltRightKeyInvoked` flag in
`HandleShortcutRemapEvent` is set when AltGr (RAlt+LCtrl) is pressed,
but only reset inside Case 1's else branch — which requires a shortcut
to be actively invoked (`isShortcutInvoked == true`)
- If the user presses and releases AltGr when **no shortcut is active**,
the flag stays `true` permanently across all future calls
- Once stale, the flag blocks modifier key restoration and state resets
at 13 locations throughout the function (lines 646, 670, 791, 800, 866,
880, 889, 907, 927, 978, 996), and causes LCtrl key-up events to `break`
out of the handler loop (line 620-622) instead of being properly
processed
- This makes LCtrl permanently stuck for any shortcut remap that uses
LCtrl as a modifier
## Fix
Reset `isAltRightKeyInvoked` when `VK_RMENU` (Right Alt) is released,
regardless of whether a shortcut is currently invoked. This is added as
an `else if` right after the existing flag-set condition.
## Test plan
- [ ] Configure a shortcut remap using LCtrl (e.g. LCtrl+Y → Backspace)
- [ ] Press and release AltGr without triggering any shortcut
- [ ] Use the LCtrl shortcut — verify Ctrl does not become stuck
- [ ] Verify AltGr-based shortcuts still work correctly
- [ ] Verify normal AltGr character input (e.g. AltGr+Q for @ on German
layout) is unaffected
---------
Co-authored-by: fluffyspace <fluffyspace@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Follow-up to #48409. That PR rewrote
`ShellViewModel.Frame_NavigationFailed` to set `e.Handled = true` and
log instead of re-throwing, but the unit tests it added only covered the
`GetPageDisplayName` formatting helper. The actual contract that matters
- "this handler must never throw, regardless of which fields on
`NavigationFailedEventArgs` happen to be null" - was not directly
testable because `NavigationFailedEventArgs` is a sealed WinRT type that
cannot be constructed from MSTest.
## Change
Tiny refactor: split the failure-handling logic out of
`Frame_NavigationFailed` into a pure static
`HandleNavigationFailure(Type sourcePageType, Exception exception)`. The
WinUI-shaped handler now just sets `Handled = true` and delegates.
This makes the "must not throw" invariant testable in isolation - no
WinUI Frame, no Microsoft.UI.Xaml.Navigation types needed.
## Tests
Added four new cases under `ShellViewModelTests`, exercising all four
`(SourcePageType, Exception)` null permutations:
| `sourcePageType` | `exception` |
| - | - |
| null | null |
| typeof(...) | null |
| null | new Exception(...) |
| typeof(...) | new Exception(...) |
Each test simply calls the helper and relies on MSTest's default "if it
throws, the test fails" behavior. Any future change that re-introduces
an unguarded dereference of `e.SourcePageType` or `e.Exception` will
turn the corresponding test red.
## Validation
All six tests pass:
```
Passed GetPageDisplayName_ReturnsFullName_ForKnownType
Passed GetPageDisplayName_ReturnsPlaceholder_ForNullType
Passed HandleNavigationFailure_DoesNotThrow_ForNullInputs
Passed HandleNavigationFailure_DoesNotThrow_ForNullException
Passed HandleNavigationFailure_DoesNotThrow_ForNullPageType
Passed HandleNavigationFailure_DoesNotThrow_ForBothInputsPresent
Total tests: 6, Passed: 6
```
`PowerToys.Settings.csproj` and `Settings.UI.UnitTests.csproj` both
build clean (Release|x64).
## Note
This PR is stacked on top of #48409 (same branch lineage). If #48409 is
merged first this PR will rebase cleanly to a small diff on
`ShellViewModel.cs` + the additional test cases.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Adds a new "Show AI paste section" setting to Advanced Paste that allows
users to hide the AI paste input box from the Advanced Paste window.
This addresses user requests for a cleaner UI when the AI paste feature
is not needed or desired.
## PR Checklist
- [x] Closes: #32967
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
This PR implements a new boolean setting `ShowAIPaste` that controls the
visibility of the AI paste input box (PromptBox) in the Advanced Paste
window.
### Changes Made
**Settings UI Library:**
- `src/settings-ui/Settings.UI.Library/AdvancedPasteProperties.cs` -
Added `ShowAIPaste` property with JSON serialization
**Settings UI:**
- `src/settings-ui/Settings.UI/ViewModels/AdvancedPasteViewModel.cs` -
Added view model property with change notification
-
`src/settings-ui/Settings.UI/SettingsXAML/Views/AdvancedPastePage.xaml`
- Added checkbox setting in the UI Behavior section
- `src/settings-ui/Settings.UI/Strings/en-us/Resources.resw` - Added
localized strings for header and description
**Advanced Paste Module:**
- `src/modules/AdvancedPaste/AdvancedPaste/Helpers/IUserSettings.cs` -
Added interface property
- `src/modules/AdvancedPaste/AdvancedPaste/Helpers/UserSettings.cs` -
Implemented setting loading from properties
-
`src/modules/AdvancedPaste/AdvancedPaste/ViewModels/OptionsViewModel.cs`
- Added `ShowAIPasteSection` property combining user setting with GPO
policy
-
`src/modules/AdvancedPaste/AdvancedPaste/AdvancedPasteXAML/Pages/MainPage.xaml`
- Bound PromptBox visibility to new property
**Tests:**
-
`src/modules/AdvancedPaste/AdvancedPaste.UnitTests/Mocks/IntegrationTestUserSettings.cs`
- Added mock property
-
`src/modules/AdvancedPaste/UITest-AdvancedPaste/TestFiles/settings.json`
- Updated test settings file
### Behavior
- Default value is `true` (AI paste section visible) to maintain
backward compatibility
- The setting respects GPO policy - if AI is disabled by GPO, the
section will be hidden regardless of user preference
- Setting is persisted in the Advanced Paste settings JSON file
## Validation Steps Performed
- [x] Verified setting appears in Settings UI under Advanced Paste →
UI Behavior
- [x] Verified toggling the setting hides/shows the AI paste input box
in the Advanced Paste window
- [x] Verified setting persists after closing and reopening Settings
- [x] Verified default value is `true` (visible) for new installations
- [x] Verified existing unit tests pass with updated mock
Fixes#32967
---------
Co-authored-by: Niels Laute <niels.laute@live.nl>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
This PR addresses issue #20618 by adding the inverted exclamation mark
(**¡**) to the **!** key (`VK_1`) and the inverted question mark (**¿**)
to the **?** key (`VK_SLASH_`) for both Spanish (`SP`) and Catalan
(`CA`) character mappings in Quick Accent.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #20618
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
Adds the inverted punctuation marks `¡` (inverted exclamation mark) and
`¿` (inverted question mark) to the character mappings of Spanish (`SP`)
and Catalan (`CA`) languages in `CharacterMappings.cs`.
- `LetterKey.VK_1` (`!`) now maps to `¡`
- `LetterKey.VK_SLASH_` (`?`) now maps to `¿`
This enables quick accent popups when pressing these respective keys,
aligning with the expected behavior for these languages.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
1. Verified character mapping changes build successfully.
2. Verified that the PowerToys CI pipeline checks completed
successfully.
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Add webp and jpg support for screenshots.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
## Summary
Harden Peek's `AppWindow.Closing` path so a stale cached preview-handler
factory can't fail-fast the Peek process. Also clean up the matching
path in RegistryPreview.
## Background
Spotted while reading through Peek's `MainWindow` teardown sequence and
the `ShellPreviewHandlerPreviewer` cache for an unrelated review of how
Peek manages out-of-process preview-handler lifetimes.
The Peek `MainWindow` subscribes to `AppWindow.Closing`. The handler
doesn't actually close the window — it sets `args.Cancel = true` and
calls `Uninitialize()`, which in turn calls
`ShellPreviewHandlerPreviewer.ReleaseHandlerFactories()`.
`ReleaseHandlerFactories()` looked like this:
```csharp
public static void ReleaseHandlerFactories()
{
foreach (var factory in HandlerFactories.Values)
{
try { Marshal.FinalReleaseComObject(factory); } catch { }
}
}
```
Two problems:
1. The static `HandlerFactories` dictionary is never cleared. After
`FinalReleaseComObject`, the entries still point at separated RCWs. A
subsequent activation that races with this cleanup (or a second close in
the same process) can pick up the dead RCW from the cache.
2. The cached factory had `LockServer(true)` called on it when it was
first cached, but the matching `LockServer(false)` was never paired.
Any managed exception that escapes a WinRT event callback is projected
back to CFlat as a failed HRESULT and the CsWinRT dispatcher fail-fasts
the process. So a single `InvalidComObjectException` (HRESULT
0x80131527) thrown out of `Uninitialize()` is enough to terminate Peek.
## Changes
* **`ShellPreviewHandlerPreviewer.ReleaseHandlerFactories`** — snapshot
then clear the dictionary up front so that a subsequent call (or a
concurrent `LoadPreviewAsync`) can't pick up a stale RCW. Call
`LockServer(false)` before `FinalReleaseComObject` to mirror the
cache-time `LockServer(true)`. Both COM calls remain individually
wrapped because the RCW may already be unreachable during process
teardown.
* **`Peek.UI/MainWindow.xaml.cs` — `AppWindow_Closing`** — wrap the body
in try/catch + `Logger.LogError`. Any future exception in
`Uninitialize()` (or its callees) will now log instead of fail-fasting
the process.
* **`RegistryPreview/MainWindow.Events.cs` — `AppWindow_Closing`** —
same defensive try/catch, plus null-guard `jsonWindowPlacement` before
`SetNamedValue`. The placement dictionary can legitimately be null on
first run or after a corrupt placement file; previously that would NRE →
fail-fast.
## Risk
Low. The `ReleaseHandlerFactories` change matches the documented
`LockServer`/`FinalReleaseComObject` pairing and only widens the
lifetime window of the cache by `Clear()`-ing earlier; nothing in Peek
calls this method outside of teardown. The two try/catch wrappers
strictly add defense — the success path is unchanged.
## Validation
Spot-built locally; this repo's `dotnet restore` runtime-pack issue
(unrelated to this PR — same NU1102 pattern that's affecting other open
PRs) prevents a full `Build.cmd` here. The C++ side of Peek is
untouched.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---
ADO:
https://microsoft.visualstudio.com/DefaultCollection/OS/_workitems/edit/58765809/
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Boliang Zhang <122517415+LegendaryBlair@users.noreply.github.com>
## Summary
Fixes a class of "stuck drag" bugs in FancyZones where closing or
destroying a window **while it is being dragged** leaves FancyZones in a
half-dragging state — zone overlays stay on screen and subsequent
keystrokes (notably number keys) are swallowed or misrouted.
## What this changes
- **Subscribe to and dispatch `EVENT_OBJECT_DESTROY`.** `FancyZonesApp`
never subscribed to the destroy event, and the consumer's
`WM_PRIV_WINDOWDESTROYED` branch could therefore never fire. The event
is now registered and routed through `HandleWinHookEvent`.
- **Abort the drag (without snapping) when the dragged window is
destroyed.** On `WM_PRIV_WINDOWDESTROYED`, if the destroyed HWND is the
one being dragged, call the new `WindowMouseSnap::Abort()` (tears down
overlays/highlights/transparency) instead of `MoveSizeEnd()`, which
would try to snap the now-dead HWND and corrupt zone state. Dragging
state is then disabled.
- **Always clear dragging state in `MoveSizeEnd()`**, even when the
snapper was already null, so the state can't get stranded.
- **Require Win+Ctrl+Alt to switch layouts while dragging.** Previously
any digit switched layouts while `dragging` was true; if drag state was
stuck this "stole" number keys from the focused app. This is the
root-cause fix for the number-key-stealing symptom.
- **Only swallow the bare Shift key during a drag**, not `Shift+<other>`
combos, so real keystrokes are no longer eaten by an in-progress drag.
## Testing
- Builds Release x64 (FancyZones) clean against current `main`.
- Manually verified drag → close window mid-drag no longer leaves
overlays up or steals number keys. (FancyZones has no unit-test harness
for this path.)
This is one of a small set of related "stuck key / stuck state"
hardening fixes; each stands alone.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Muyuan Li <116717757+MuyuanMS@users.noreply.github.com>
## Summary
Hardens the runner's centralized keyboard hook against stuck and "ghost"
key activations — cases where a hotkey action fires after the key was
already released, or a pending timer fires after the hook was torn down.
## What this changes
- **`vkCodePressed` is now `std::atomic<DWORD>`.** It is read/written
from the low-level hook callback and from the timer/teardown paths; the
plain `DWORD` was a data race.
- **Revalidate the key is still physically held before firing a held-key
timer.** The pressed-key timer callback now checks
`GetAsyncKeyState(virtualKey) & 0x8000` before invoking the action,
preventing ghost activations after the user has let go.
- **Tag the injected dummy `0xFF` key-up** with
`CENTRALIZED_KEYBOARD_HOOK_DONT_TRIGGER_FLAG` so the hook does not
reprocess its own synthetic event.
- **`Stop()` kills all pending pressed-key timers and resets
`vkCodePressed` before unhooking**, so a timer can't fire a callback
into a half-removed hook.
## Testing
- Builds Release x64 (runner / `PowerToys.exe`) clean against current
`main`.
This is one of a small set of related "stuck key" hardening fixes; each
stands alone.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
The ZoomIt module description contained `"snip screenshots"` — both
words map to the same Chinese term (截图), causing the translation system
to emit the garbled `"截图截图截图"` in the Chinese UI.
## Changes
- **`src/settings-ui/Settings.UI/Strings/en-us/Resources.resw`**:
Changed `"to snip screenshots to the clipboard or to a file"` → `"to
copy screenshots to the clipboard or to a file"` in both
`ZoomIt.ModuleDescription` and `Oobe_ZoomIt.Description`
- `copy` has an unambiguous, distinct Chinese translation (`复制`) with no
overlap with `截图`
- Consistent with the existing `ZoomIt_SnipGroup.Description` which
already uses `"Copy a selected area of the screen to the clipboard or to
a file"`
## PR Checklist
- [ ] Closes: #xxx
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
Before (zh-CN rendered):
> 还可以使用 Zoomlt 将屏幕**截图截图截图**剪贴板或文件中
After (expected zh-CN):
> 还可以使用 ZoomIt 将截图**复制到**剪贴板或文件中
Root cause: `snip` (Windows snipping action) and `screenshots` are both
translated to `截图` by the localization pipeline, producing triple
repetition. Replacing `snip` with `copy` — already used in the adjacent
Snip group description — resolves the collision without changing the
functional meaning.
## Validation Steps Performed
- Verified both string keys (`ZoomIt.ModuleDescription`,
`Oobe_ZoomIt.Description`) updated in `en-us/Resources.resw`
- Confirmed no other occurrences of `"snip screenshots"` remain in the
file
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: MuyuanMS <116717757+MuyuanMS@users.noreply.github.com>
## Summary
Removes outdated/unused documentation and asset files:
- **`doc/unofficialInstallMethods.md`** — removed.
- **`doc/planning/`** — removed entire folder (`awake.md`,
`FancyZonesBacklog.md`, `PowerToysBacklog.md`,
`ShortcutGuideBacklog.md`).
- **`doc/images/icons/Video Conference Mute.png`** — removed (unused
module icon, not referenced anywhere in the repo).
Verified there are no remaining references to any of the removed files
across tracked sources (checked literal and `%20`-encoded paths).
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
`PowerToys.Update` reads `downloadedInstallerFilename` from the
persisted `UpdateState.json` and combines it with the `Updates` folder
to locate the installer to run. If that cached state is stale,
corrupted, or otherwise unexpected, the stored value could contain path
separators or an absolute path, which would make the updater look for
the installer outside the `Updates` folder.
This PR validates that the cached value is a plain filename and that the
resolved path stays inside the `Updates` folder before using it;
otherwise the update is treated as unavailable. The normal update flow
(a bare asset filename produced by the download step) is unaffected.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already.
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** No end-user-facing strings added
- [x] **Dev docs:** N/A
- [x] **New binaries:** None
## Detailed Description of the Pull Request / Additional comments
- The filename check is factored into an inline helper
`updating::IsSafeDownloadedInstallerFilename` in
`src/common/updating/updateLifecycle.h` (next to the other inline update
helpers) so it can be unit-tested without a project reference.
- `ObtainInstaller` in `src/Update/PowerToys.Update.cpp` now calls that
helper and additionally confirms, via `weakly_canonical`, that the
resolved installer path's parent is the `Updates` directory.
- No behavior change for normal installer filenames, so the regular
update flow does not regress.
## Validation Steps Performed
- Added `IsSafeDownloadedInstallerFilenameTests` to `Updating.UnitTests`
covering normal filenames, empty values, parent-directory components,
nested path components, and absolute/drive-relative/UNC paths.
- Built `Updating.UnitTests` (Debug x64): 0 warnings, 0 errors.
- Ran the full `Updating.UnitTests` suite: 36/36 passed (30 existing + 6
new).
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Fixes#25353 — the FancyZones Settings page bounces / jumps when the
user scrolls down to the bottom (the **Excluded apps** expander).
Several closed dupes report the same behavior: #31500, #34259, #30164,
#45173, #32099, #32361.
## Root cause
The Excluded-apps `TextBox` (lines 283-305 of `FancyZonesPage.xaml`)
forced its own inner `ScrollViewer` to be permanently enabled with a
visible scrollbar, nested inside the page-level `ScrollViewer` provided
by `SettingsPageControl`:
```xml
ScrollViewer.IsVerticalRailEnabled="True"
ScrollViewer.VerticalScrollBarVisibility="Visible"
ScrollViewer.VerticalScrollMode="Enabled"
```
The expander is `IsExpanded="True"` and sits at the very bottom of the
page, so when you wheel-scroll down the page the pointer ends up over
the TextBox. The TextBox''s inner scroller intercepts the wheel events;
when it has nothing left to scroll the event bubbles back to the outer
page scroller, which tries to over-scroll past content end. The result
is the bounce/jump visible in the videos on the issue.
The unbounded `MinHeight="160"` (with no `MaxHeight`) made it worse,
because the box can grow as users type, triggering layout re-measures
during scrolling.
## Fix
Minimal change to the same TextBox:
- Drop `ScrollViewer.IsVerticalRailEnabled` and
`ScrollViewer.VerticalScrollMode` — `AcceptsReturn="True"` already wires
up the inner scroller correctly.
- Change `ScrollViewer.VerticalScrollBarVisibility` from `Visible` to
`Auto`, so the inner scroller only competes with the outer one when
there is actual overflow to scroll.
- Add `MaxHeight="320"` so the box cannot keep growing as the user
types.
## Validation
- Builds clean (`tools\build\build.cmd` from
`src\settings-ui\Settings.UI`, arm64 Debug, exit 0, 0 errors).
- The same nested-scroller pattern exists on 9 other Settings pages
(AlwaysOnTop, GrabAndMove, MouseUtils, MouseWithoutBorders, PowerAccent,
PowerLauncher, ShortcutGuide). Keeping this PR atomic to the page called
out in #25353; happy to follow up with a sibling PR for those if
reviewers want them in the same pass.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
`PidToIconConverter.Convert` called `Icon.ExtractAssociatedIcon(path)`
without a guard. When a running process's image path is non-empty but
the file no longer exists on disk, the call throws
`FileNotFoundException`. The converter runs per-row while the process
`ListView` virtualizes, so the exception reached
`App_UnhandledException` (which only logs and never sets `e.Handled =
true`) and WinUI 3 fast-failed the whole app (`0xc000027b`).
This is routinely triggered by self-updating software that deletes its
old versioned directory while the old process keeps running — e.g.
Windows Defender (`SenseAPZ`, `MsMpEng`, `MpDefenderCoreService` under
versioned `...\Platform\<ver>\` / `...\DataCollection\<ver>\` paths).
Right-clicking a drive root enumerates every process and reliably
includes such a stale-path process, so the crash is easy to hit in
normal use.
The fix wraps the icon extraction in a try/catch and falls through to
the existing placeholder `BitmapImage`, logging a warning — mirroring
the exception handling already present in `MainViewModel.WatchProcess`.
## PR Checklist
- [x] Closes: #48693
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass — no existing test harness
covers this WinUI converter path; validated manually (steps below)
- [x] **Localization:** All end-user-facing strings can be localized —
n/a; the only new string is a developer log warning, not end-user UI
- [x] **Dev docs:** Added/updated — n/a
- [x] **New binaries:** Added on the required places — n/a; no new
binaries
## Detailed Description of the Pull Request / Additional comments
`src/modules/FileLocksmith/FileLocksmithUI/Converters/PidToIconConverter.cs`:
```csharp
if (!string.IsNullOrEmpty(y))
{
try
{
icon = Icon.ExtractAssociatedIcon(y);
}
catch (Exception ex)
{
// The process image path can be non-empty but no longer exist on disk
// (e.g. self-updating software that deletes its old versioned directory while
// the old process is still running). ExtractAssociatedIcon then throws and,
// because this converter runs per-row during ListView virtualization, the
// exception would otherwise reach App_UnhandledException and fast-fail the app.
// Fall through to the placeholder icon instead of crashing.
Logger.LogWarning($"Couldn't extract the icon for '{y}'. {ex}");
}
}
```
(+ `using ManagedCommon;` — already a project reference; `Logger` is
used identically in `App.xaml.cs`.)
Scope note: this keeps to the targeted converter guard. I deliberately
did **not** also blanket-set `e.Handled = true` in
`App_UnhandledException`, since that would mask unrelated genuine
crashes; the converter guard fully addresses this crash.
## Validation Steps Performed
1. Reproduced on the installed 0.100 build: right-click `C:\` → **Unlock
with File Locksmith** → scroll the list → crash (`0xc000027b`, faulting
module `Microsoft.UI.Xaml.dll`; the FileLocksmith log shows an unhandled
`FileNotFoundException` for `...\SenseAPZ.exe` in
`MainPage...ProcessBindings` / `PidToIconConverter`). Reproduced twice.
2. Built `FileLocksmithUI` (x64/Release) with the fix — 0 warnings / 0
errors.
3. Ran the produced exe on `C:\` and scrolled ~40× over the same list
(which includes the stale `SenseAPZ` row):
- No crash; the process stayed alive and responsive.
- The log shows the new warning firing for the exact `...\SenseAPZ.exe`
path, confirming the catch branch executed on the previously-fatal row.
- 0 unhandled exceptions in the session.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
When you're pinning something to the dock with the dialog, I have _no_
idea what display is 1,2,3,4. That's meaningless to me.
This attempts to resolve that by adding a teachingtip to each dock while
pinning. When we open that pin to dock dialog, we'll display the
teachingtips, and dismiss them when the dialog is closed.
That way folks have at least some indication of where each display is
when pinning.
## Summary
The `TopLevelCommandManager` handles **alot.** And by "alot" I mean too
much. In preparation for future types of extensions, this PR attempts to
pull out the methods for loading/managing extensions to individual
`IExtensionService`s.
These `ExtensionService`s will be injected the `TopLevelCommandManager`
via dependency injection. It will iterate through them asking them to
load their extensions. During that process, they will surface
`ICommandProvider`s back to the `TopLevelCommandManager` for display in
Command Palette.
Currently, there are two types of `IExtensionService`s:
`BuiltInExtensionService` and `WinRTExtensionService`.
- The `BuiltInExtensionService` receives all `ICommandProvider`s
registered with DI.
- The `WinRTExtensionService` manages out-of-process WinRT AppExtension
providers.
## Additional Changes
- **Circular DI dependency resolved**: `BuiltInsCommandProvider` no
longer depends on `IRootPageService`.
Previously, it required the `IRootPageService` solely for the dock
command to open Command Palette (using the RootPage as its command.)
Now, it uses a new `GoHomeDockCommand` which returns
`CommandResult.GoHome()` with the existing `onBeforeShowConfirmation`
callback to show the palette at the dock button position.
Also, fixes#48643
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Shortcut Guide overlay crashes and closes when navigating between
sidebar sections (e.g. clicking PowerToys, then clicking the Windows
icon to come back). Repro and crash logs in #48448. Crash logs in #48441
show the same propagation path plus a follow-up access violation in
coreclr, indicating exceptions that escape local catches.
This PR fixes the immediate navigation race and adds broader crash
hardening so future exceptions on the UI/background threads are logged
instead of tearing down the overlay.
### Navigation-race fix (commit 1)
Root cause: `WindowSelector_SelectionChanged` calls
`App.TaskBarWindow.Activate()` and then `SetWindowPosition()`
synchronously. `Activate()` runs a reentrant `Window_Activated` →
`BringToFront` → `TaskbarWindow.Activated` chain that can leave
`App.TaskBarWindow.AppWindow` momentarily null, so `SetWindowPosition`
throws `NullReferenceException`.
Because the initial `SelectedItem = MenuItems[0]` is set from
`SetNavItems`, the NRE bubbles up into `InitializeNavItemsAsync`'s catch
block — which sets `_closeType = "InitializationFailed"` and closes the
window. That matches both user-visible symptoms: the overlay "flashes
and disappears" (#48441) on open and "closes when clicking the Windows
icon" (#48448).
Edits in
`src/modules/ShortcutGuide/ShortcutGuide.Ui/ShortcutGuideXAML/MainWindow.xaml.cs`:
- **`SetWindowPosition`**: null-guard `App.TaskBarWindow?.AppWindow` and
skip the taskbar-overlap height adjustment when it is not currently
observable. Wrap the body in `try`/`catch` with `Logger.LogError` so any
future positioning hiccup keeps the previous layout instead of taking
down the overlay.
- **`WindowSelector_SelectionChanged`**: null-guard
`App.TaskBarWindow?.Hide()` / `Activate()` and wrap the body in
`try`/`catch`. This breaks the propagation path that lets a navigation
exception reach `InitializeNavItemsAsync`'s "fatal init failure" branch.
### Crash hardening (commit 2)
Additional defensive changes to cover other unguarded paths surfaced
while reviewing #48441:
- **`App.xaml.cs`**: register `App.UnhandledException`,
`AppDomain.CurrentDomain.UnhandledException`, and
`TaskScheduler.UnobservedTaskException` so a stray exception (e.g. an IO
failure during a fire-and-forget UI handler, or a background `Task`
fault) gets logged instead of tearing the process down with an access
violation in coreclr. Mirrors what Peek, AdvancedPaste, and CmdPal
already do.
- **`App.OnLaunched`**: wrap the launch sequence in `try`/`catch` and
exit cleanly with an error log on failure (`LoadData` / `MainWindow` /
`TaskbarWindow` ctors and `Activate` are all reachable failure points).
- **`App.LoadData`**: broaden the `Pinned.json` deserialize catch to
also handle `IOException` / `UnauthorizedAccessException`, and guard the
round-trip `SaveSettings` call as best-effort with a warning log.
- **`PinnedShortcutsHelper.Save`**: catch `IOException` /
`UnauthorizedAccessException` / `JsonException` and log; `Pin` / `Unpin`
runs from a synchronous UI handler so an unguarded `File.WriteAllText`
would tear down the overlay on any disk hiccup.
- **`TaskbarWindow.UpdateTasklistButtons`**: move the `AppWindow.Move`
calls inside the existing `try` block, null-guard
`App.MainWindow?.AppWindow` up front, and wrap the whole body so the
method (which runs from the ctor and from `Activated`) cannot tear the
overlay down when `MainWindow` is in a transient state.
## PR Checklist
- [x] Closes: #48448
- [x] Likely also fixes: #48441
- [x] **Communication:** Defensive fix to known crash paths; no API
change.
- [ ] **Tests:** No automated tests added — the failures are reentrancy
/ timing races that are hard to deterministically trigger in CI.
Validated with a synthetic repro (see below).
- [x] **Localization:** N/A — only logger strings.
- [x] **Dev docs:** N/A
- [x] **New binaries:** N/A
## Detailed Description of the Pull Request / Additional comments
The fix is intentionally defensive (rather than restructuring the
reentrant activation flow) because the legacy taskbar UIA enumeration
(`TasklistPositions.GetButtons`) on Windows 10 is what most reliably
widens the race window and is out of scope to redesign for a hotfix.
## Validation Steps Performed
- Build: `tools\build\build.cmd` from
`src\modules\ShortcutGuide\ShortcutGuide.Ui` — exit 0, errors log empty
for both commits.
- Synthetic repro: temporarily injected `throw new
NullReferenceException()` at the top of `SetWindowPosition` to exercise
the exact propagation path the reporter's log shows (`SetWindowPosition`
→ `SelectionChanged` → `set_SelectedItem` →
`InitializeNavItemsAsync.catch` → `Close("InitializationFailed")`).
- Before fix: overlay flashes and closes, log shows `Failed to
initialize navigation items.` with the NRE.
- After fix: overlay stays open, page navigates, log shows `Failed to
set Shortcut Guide window position; keeping previous layout.` from the
new `catch`.
- Did not reproduce the live race on a Win11 dev box; the reporter's
repro is Windows 10 19045 / Microsoft Store install where the legacy
taskbar UIA timing makes the reentrant chain more likely to expose the
null.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
The per-user **"Autorun for <user>"** logon scheduled task
(created in `src/runner/auto_start_helper.cpp`) was registered with a
security descriptor that granted **full access to Everyone**
(`D:(A;;FA;;;WD)`). This PR scopes the task's permissions to only the
accounts that actually need them.
## Changes
- `src/runner/auto_start_helper.cpp`
- Added a small `get_auto_start_task_sddl()` helper that builds the
task's security descriptor granting full control to **Local System**,
the **Administrators** group, and **the user the task belongs to** (SID
read from the current process token).
- Use it when calling `RegisterTaskDefinition` instead of the previous
Everyone descriptor.
- Added `#include <sddl.h>` for `ConvertSidToStringSidW`.
## Behavior / compatibility
- No change to the run-at-startup or run-elevated flows. The decision
logic in `general_settings.cpp` is untouched — only the descriptor
string passed to `RegisterTaskDefinition` changed.
- The task owner can still fully manage (enable / disable / delete) its
own task in both elevated and non-elevated runner modes, because a
user's token SID is identical across elevation.
- The Task Scheduler service (SYSTEM) retains full control, so the task
still runs at logon as before.
## Validation
- Built `src/runner/runner.vcxproj` (x64 / Debug): **0 errors, 0
warnings**.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Profiles in the Environment Variables utility override **User**
variables only, as described in
[`doc/devdocs/modules/environmentvariables.md`](https://github.com/microsoft/PowerToys/blob/main/doc/devdocs/modules/environmentvariables.md).
This change makes the implementation apply that documented behavior
consistently: every registry write performed while applying, unapplying,
or editing a profile now goes through dedicated user-scope helpers
instead of the generic `ParentType`-based helpers. It also removes some
dead code along the way.
## PR Checklist
- [ ] Closes: N/A (code consistency / cleanup)
- [x] **Tests:** No automated test project exists for this module;
validated via build + manual steps (see below)
- [x] **Localization:** No end-user-facing strings changed
- [x] **Dev docs:** Existing docs already describe the user-scoped
profile behavior; no change needed
- [x] **New binaries:** None
## Detailed Description of the Pull Request / Additional comments
- Added `SetProfileVariableWithoutNotify` /
`UnsetProfileVariableWithoutNotify` to `EnvironmentVariablesHelper`.
They centralize writing profile-driven variables to the current-user
environment, reusing the existing
`SetEnvironmentVariableFromRegistryWithoutNotify` primitive.
- Updated `ProfileVariablesSet.Apply` / `UnapplyVariable` and the
profile branch of `Variable.Update` to use these helpers.
- Removed an unused local (`var applyToSystem =
variable.ApplyToSystem;`) in `ProfileVariablesSet.Apply`.
- Removed the now-unused `SetVariableWithoutNotify` /
`UnsetVariableWithoutNotify` helpers. The notify variants (`SetVariable`
/ `UnsetVariable`) are unchanged and still back the default User/System
set editing.
- Net change: 3 files, +18 / −33.
## Validation Steps Performed
- Built `EnvironmentVariablesUILib` (Debug, x64): **succeeded, 0
warnings / 0 errors**.
- Manual: created, enabled, disabled, and edited profiles; confirmed
profile variables apply, back up the overridden values, and restore them
on disable exactly as before.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Mouse Without Borders encrypts the socket streams it uses to talk
between machines. Previously the PBKDF2 salt and the AES-CBC IV were
both derived from a single fixed constant, so every connection reused
identical key-derivation material.
This change generates a fresh random salt and IV for each connection and
exchanges them in-band as a small cleartext header ahead of the cipher
text. The AES key is now derived from the per-connection salt.
## Changes
- `Encryption.cs`: generate a random per-connection salt + IV,
write/read them as a cleartext header in
`GetEncryptedStream`/`GetDecryptedStream`, and derive the key from that
salt.
- Removed the now-unused fixed-IV helper (`GenLegalIV`), the `InitialIV`
constant, and the derived-key cache (a per-connection salt makes caching
pointless).
- The header exchange reuses the same tolerant socket-close handling as
the existing random first-block handshake, so the exception behavior
seen by callers (`SocketStuff`, `Clipboard`) is unchanged. Call sites
are untouched.
## Compatibility
This changes the on-the-wire format, so all paired machines must run
this version. Mouse Without Borders already requires the same version on
every machine (it surfaces *"make sure you run the same version in all
machines"* on a handshake mismatch). Existing keys and settings are
preserved - no re-pairing is needed once every machine is updated.
## Validation
- Built `MouseWithoutBorders` App and UnitTests (Release | x64) - clean
(exit 0).
- Added `EncryptionTests` covering encrypt -> decrypt round-tripping and
per-connection uniqueness (salt/IV/header differ on every connection).
All 3 tests pass.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Hardens the local `github-artifacts` MCP helper
(`tools/mcp/github-artifacts/server.js`) so the `GITHUB_TOKEN` bearer
token is only ever sent to allowlisted GitHub API hosts. Previously the
same token was attached when downloading image and ZIP URLs extracted
from issue/PR Markdown, which could forward the token to any host
referenced in untrusted issue content.
This is repository MCP dev-tooling, not the shipped PowerToys runtime.
## What changed
- **New `auth.js`** centralizes token handling:
- `AUTH_ALLOWED_HOSTS = { api.github.com }`
- `shouldSendGitHubToken(url)` — true only for HTTPS + allowlisted host
- `headersForUrl(url, token, extra)` — attaches `Authorization` only
when the host is allowlisted
- **`server.js`** routes all three request sites (`fetchJson`,
`downloadBytes`, `downloadZipBytes`) through `headersForUrl`. Image/ZIP
URLs from issue content are now fetched **without** `Authorization`.
Removed the ZIP dual-attempt auth retry that previously sent the token
to untrusted hosts.
- **New `test-token-allowlist.js`** regression test (wired into `npm
test` / `npm run test:unit`).
## Verification
- Unit test: 13/13 pass — token attached for `api.github.com`, withheld
from arbitrary hosts, loopback, look-alike hostnames
(`api.github.com.attacker.com`), non-HTTPS, and
`*.githubusercontent.com`.
- Cross-origin redirects: confirmed Node's `fetch` (undici) strips
`Authorization` when a redirect leaves the origin.
- Functionality preserved (real public issues, real token): issue #25595
images (5/5 downloaded) and issue #39476 `PowerToysReport_*.zip`
(downloaded + extracted) both succeed without the token — GitHub serves
this content via public/signed URLs.
## Note
For private-repo attachments that would require an `Authorization`
header on a non-API host, the token is intentionally not sent (the
recommended tradeoff). PowerToys is public, so the helper's normal use
is unaffected.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
This PR adds `Microsoft.Plugin.Folder.UnitTests` to SLNX as a project
instead of a plain file.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
Remove `UPDATE_STAGE2_RESTART_PT` and `UPDATE_STAGE2_DONT_START_PT`
constants from `UpdateUtils.h`. Never referenced anywhere in the
codebase.
Fixes#46968. Found during multi-agent code review of #46889.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
On dual-GPU laptops, Power Display stopped detecting the built-in panel
(and adjusting its brightness) when the **discrete GPU** drives the
display — it showed "can't detect the display". This fixes that by
classifying displays by **capability** (does WMI brightness work on it?)
instead of by the nominal `OutputTechnology` value, which the discrete
GPU misreports for the internal panel.
## PR Checklist
- [x] Closes: #48587
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized (no
new user-facing strings added)
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
## Detailed Description of the Pull Request / Additional comments
### Root cause
On a hybrid / MUX laptop, when the **discrete GPU** drives the built-in
eDP panel, `QueryDisplayConfig` reports the panel's
`DISPLAYCONFIG_VIDEO_OUTPUT_TECHNOLOGY` as `DISPLAYPORT_EXTERNAL` (`10`)
instead of the `INTERNAL` flag (`0x80000000`) it reports under the
integrated GPU. It is the *same physical panel* (same EDID) — only the
reported connector type changes with the active GPU.
PR #47740 introduced a strict classifier: `OutputTechnology` →
internal/external, then **internal → WMI-only, external → DDC/CI-only,
with no fallback**. So under the discrete GPU the built-in panel was
classified *external* and sent to DDC/CI only — but a laptop eDP panel
does not speak DDC/CI, so it was dropped and Power Display reported it
couldn't detect any monitor. (`WmiMonitorBrightness` still exposes that
panel regardless of which GPU drives it, so the panel was actually
controllable — it just never got routed to WMI.)
### Fix: classify by capability, not by nominal output technology
- **`MonitorManager`** now runs **WMI discovery first** over the full
`QueryDisplayConfig` inventory. Every display `WmiMonitorBrightness`
exposes is treated as internal (WMI-controlled); whatever WMI does
**not** claim is routed to DDC/CI. The `OutputTechnology`-based
classifier is gone.
- **`WmiController`** matches the system-wide `WmiMonitorBrightness`
results against the full inventory by `Monitor.Id`. The persisted
`Monitor.Id` is still taken from the matched `DevicePath`
(byte-identical to the DDC route and to prior releases), so saved
brightness/per-monitor settings survive upgrades.
- New **`MonitorIdentity.FromInstanceName`** reduces a WMI
`InstanceName` to the same canonical `Monitor.Id` as `FromDevicePath`;
the separate `PnpHardwareKey` helper is removed.
- **Deleted** `DisplayClassifier` and `MonitorDisplayInfo.IsInternal`
(net ~150 fewer lines).
### Accepted trade-off
A monitor that exposes **both** `WmiMonitorBrightness` **and** DDC/CI is
now controlled via WMI only and won't get DDC-only features (contrast /
volume / input source / color temperature / power). This is uncommon
(typical laptop panels are WMI-only; typical external monitors are
DDC-only) and is a deliberate decision: it removes the entire class of
`OutputTechnology` misclassification bugs while keeping the performance
win of not DDC-probing internal panels.
## Validation Steps Performed
- Built the Power Display app (`PowerDisplay.csproj`) and
`PowerDisplay.Lib.UnitTests` (x64 / Debug) with MSBuild — both succeed,
including after merging latest `main` (Windows App SDK 2.2.0).
- Ran the unit test suite: **128/128 pass**, including new
`FromInstanceName` tests — the `FromInstanceName == FromDevicePath`
equivalence invariant and a concrete #48587 regression case (the BOE
panel reported as `OutputTechnology=10`).
- Traced the fix against the reporter's diagnostic logs: the panel that
previously went `OutputTechnology=10 → External → DDC → dropped` is now
claimed by WMI and controllable.
- Reviewed the diff for regressions (Monitor.Id persistence, monitor
blacklist, mirror mode, dual-internal-panel devices, external-only
desktops).
---------
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Increase sleep duration while waiting for IPC setup in the interop unit
test.
A number of PRs have had failures recently even though they do not touch
interop, likely due to this sleep being too short. An existing comment
points to the addition of the delay being because of this, and
resource-constrained build agents/environments could need a longer
duration.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
See comments on #47211 and #48106.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
N/A
## Summary of the Pull Request
Converts the literal-digit shortcut keys in the bundled Shortcut Guide
manifests to the `<N>` special-key convention, so they render as the
correct number.
Per the manifest spec, a bare number in `Keys:` is a virtual-key code. A
literal digit key authored as a bare number is therefore misread (VK `9`
is Tab, VK `1` is the left mouse button, VK `0` is undefined) and
renders incorrectly. The fix authors these as the `<N>` token (for
example `"<9>"`), which `KeyVisual` strips to display the digit.
This is a data-only change: **91 literal-digit keys across 14
manifests** become `<N>` tokens. No code or doc changes; the renderer
and converter already support `<N>`, and the convention is documented in
the spec.
Follow-up to #48461, which introduced and documented the `<N>`
convention (per @noraa-junker's request for a separate PR to fix the
remaining manifests). Together with #48461 this resolves the rendering
reported in #48460.
Files touched (all under
`src/modules/ShortcutGuide/ShortcutGuide.Ui/Assets/ShortcutGuide/Manifests/`):
Adobe.Illustrator (18), Adobe.Photoshop (17), SlackTechnologies.Slack
(13), Adobe.InDesign (11), JetBrains.IntelliJIDEA.Community (11),
BlenderFoundation.Blender (3), Figma.Figma (3), Google.Chrome (3),
Microsoft.Edge (3), Microsoft.VisualStudioCode (3), Mozilla.Firefox (3),
+WindowsNT.Notepad (1), Adobe.AfterEffects (1), GIMP.GIMP (1).
## PR Checklist
- [ ] **Closes:** N/A (follow-up to #48461; contributes to #48460)
- [ ] **Communication:** discussed in #48461; @noraa-junker requested
this separate PR.
- [ ] **Tests:** N/A for data; the converter and `<N>` convention are
unit-tested in #48461. Validated here by deserializing every manifest
with YamlDotNet (see below).
- [x] **Localization:** unchanged; these are per-language manifest
files.
- [ ] **Dev docs:** N/A (the `<N>` convention is documented in the spec
via #48461).
- [ ] **New binaries:** N/A.
- [ ] **Documentation updated:** N/A.
## Detailed Description of the Pull Request / Additional comments
Each change wraps a single bare digit in angle brackets, for example:
```yaml
Keys:
- - 9
+ - "<9>"
```
Quoted tokens (`"<9>"`) are used to match the dominant special-token
style already in the manifests (`"<Enter>"`, `"<Down>"`, etc.).
Note (out of scope): `Adobe.Photoshop.en-US.yml` has a shortcut with an
empty `Name: ""` (around line 799). That is a pre-existing data issue
unrelated to digit rendering; the digit is still converted, and the
empty name is left as-is.
## Validation Steps Performed
- Confirmed the diff touches only the 14 manifests, 91 insertions and 91
deletions, with each changed line being a digit wrapped as `"<N>"` (no
whitespace, indentation, or encoding churn).
- Confirmed zero bare-digit `Keys` entries remain and exactly 91 new
`"<N>"` tokens exist.
- Deserialized all 32 manifests with YamlDotNet (the same library the
app uses at runtime): 0 parse errors.
- Rendering behavior for `<N>` is already verified in #48461 (the
renderer strips the brackets to show the digit).
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
## Summary of the Pull Request
Adds a Shortcut Guide manifest for **Postman** and fixes a rendering bug
where single-digit keys in manifests displayed incorrectly.
- **Fix numbered-key rendering** —
`src/modules/ShortcutGuide/ShortcutGuide.Ui/Converters/ShortcutDescriptionToKeysConverter.cs`:
a single digit (`0`–`9`) in a manifest's `Keys` was treated as a Windows
virtual-key code instead of the literal digit. Since VK `1` is the left
mouse button, VK `9` is Tab, and VK `0` is undefined, shortcuts such as
`Ctrl+0` (reset zoom) and `Ctrl+9` (last tab) rendered as
blank/incorrect glyphs. Single digits are now rendered as the literal
character.
- **Add Postman shortcuts** —
`src/modules/ShortcutGuide/ShortcutGuide.Ui/Assets/ShortcutGuide/Manifests/Postman.Postman.en-US.yml`:
new manifest for `Postman.exe` covering Tabs, Sidebar, Request,
Interface, Window and modals, and Console. Auto-included via the
existing `Manifests/*.yml` glob in `ShortcutGuide.Ui.csproj`.
- **Show tab-number ranges** — Edge, Chrome, Firefox, and Postman
manifests: the "switch to a specific tab" entry used the literal key
`1`, which (after the fix above) read as `Ctrl + 1`. It now uses a `1 -
8` range so the keycap conveys "any tab number 1 through 8". The
separate "last tab" (`9`) and "reset zoom" (`0`) entries remain literal
single keys.
- **Add unit tests** — new `ShortcutGuide.UnitTests` (MSTest) project
covering `ShortcutDescriptionToKeysConverter.GetKeysList`, including the
single-digit regression.
## PR Checklist
- [x] Closes: #48460
- [ ] **Communication:** I've discussed this with core contributors
already. <!-- Filed #48460; the v0.100 announcement invites app-shortcut
contributions via PR. -->
- [x] **Tests:** Added/updated and all pass <!-- New
ShortcutGuide.UnitTests (MSTest); 8 tests pass locally via
vstest.console. -->
- [x] **Localization:** All end-user-facing strings can be localized
<!-- Shortcut names live in per-language manifest files (`*.en-US.yml`);
other locales fall back to en-US, consistent with existing manifests.
-->
- [ ] **Dev docs:** Added/updated <!-- N/A: no behavior requiring
dev-doc changes. -->
- [ ] **New binaries:** Added on the required places <!-- N/A: the new
manifest is a data asset under an already-shipped, globbed folder. The
new test project is auto-discovered by the existing `**\*UnitTest*.dll`
VSTest glob, so no CI pipeline change is required. -->
- [ ] **Documentation updated:** <!-- N/A -->
## Detailed Description of the Pull Request / Additional comments
The Shortcut Guide displays per-app shortcuts from YAML manifests,
matched to the foreground window via `WindowFilter`. Keys are converted
to keycaps by `ShortcutDescriptionToKeysConverter`. Numeric key strings
were unconditionally parsed as virtual-key codes, so literal-digit
shortcuts rendered wrong. The fix adds a `>= 0 and <= 9` case that emits
the digit character as-is; non-digit numeric codes (arrows, etc.) are
unchanged.
The new Postman manifest exercises this with `Ctrl+0` / `Ctrl+9`. The
browser/Postman "specific tab" entries were updated from the literal `1`
to the `1 - 8` range string, rendered verbatim by `KeyVisual` (the same
path used by the existing `Number (1-9)` key in the Windows Explorer
manifest).
A new `ShortcutGuide.UnitTests` (MSTest) project covers the converter:
single digits render literally (regression test), modifier ordering,
non-numeric passthrough (e.g. `1 - 8`), and arrow-key VK mapping.
## Validation Steps Performed
Built and ran locally (x64 Debug):
- Built `ShortcutGuideModuleInterface`, `ShortcutGuide.Ui`, and
`ShortcutGuide.IndexYmlGenerator`; launched the Debug `PowerToys.exe`.
- Triggered Shortcut Guide (`Win+Shift+/`) with **Postman** focused: the
Postman section renders with all categories, and `Ctrl+1` / `Ctrl+9` /
`Ctrl+0` display correctly (previously blank/incorrect).
- Verified the "specific tab" entry renders as `Ctrl + 1 - 8` in
**Edge**, **Chrome**, **Firefox**, and **Postman**.
- Built `ShortcutGuide.UnitTests` and ran via `vstest.console.exe`:
**8/8 tests pass**.
<img width="845" height="1432" alt="PowerToys Shortcut Guide Running
Postman"
src="https://github.com/user-attachments/assets/6359617e-3e2c-48b0-8005-b3684594ec94"
/>
Co-Authored-By: Claude Opus 4.8
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #37801
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
This uses `SetWindowAffinity()` with the `WDA_EXCLUDEFROMCAPTURE`
constant on the main Color Picker window to ensure the ZoomWindow bitmap
creation excludes the corner of the picker UI. The ability to capture
the picker window is restored immediately after, so it's still visible
in Snipping Tool, Remote Desktop and other tools.
The fix uses a new helper with simple `Include()` / `Exclude()` methods
as an adapter to the native code. This may potentially be included in
Common if other utilities needed it.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
1. Repro'd the original issue on the current Color Picker release:
<img width="262" height="263" alt="image"
src="https://github.com/user-attachments/assets/bc12b2f8-b67f-4b56-a803-57ab2fe0fa17"
/>
2. Confirmed that the fix worked and normal window affinity was restored
(otherwise I would not have been able to snip this):
<img width="262" height="262" alt="image"
src="https://github.com/user-attachments/assets/bc336ad0-4114-49fa-8440-b78466de363f"
/>
3. Repeated zooming in and out over a normal session.
Closes#45025
## [QuickAccent] Add Pitjantjatjara / Yankunytjatjara Language Support
### Summary
This pull request adds support for Pitjantjatjara/Yankunytjatjara
to the Quick Accent feature. Pitjantjatjara/Yankunytjatjara is an
Australian Aboriginal language
spoken in the Western and South Australian deserts.
### What's New
- Users can now access Pitjantjatjara/Yankunytjatjara retroflex
consonant characters (ḻ, ṉ, ṟ, ṯ)
through Quick Accent
- Language is available in the Quick Accent settings dropdown menu
- Full localization support for UI display
### Why This Matters
Pitjantjatjara/Yankunytjatjara uses unique retroflex consonants that are
essential for:
- **Correct pronunciation:** Retroflex sounds are phonetically distinct
from regular consonants
- **Written accuracy:** Proper character representation in educational
materials
- **Cultural preservation:** Supporting Aboriginal language
documentation and teaching
- **Accessibility:** Users without specialized keyboards can now type
these characters easily
### Character Mappings
The implementation adds 4 retroflex consonants:
- **L → ḻ** (Retroflex lateral approximant)
- **N → ṉ** (Retroflex nasal)
- **R → ṟ** (Retroflex approximant)
- **T → ṯ** (Retroflex stop)
### Files Changed
1. **Language.cs** - Added `PJT` enum value
2. **CharacterMappings.cs** - Added language entry with character
mappings and display order
3. **Resources.resw** - Added localized UI string
4. **expect.txt** - Added language name to spell-check whitelist
### Testing
- ✅ Language appears in Quick Accent settings dropdown
- ✅ Retroflex characters appear when holding L/N/R/T + activation key
- ✅ No build errors or warnings
## Summary
Fixes VS Code Workspaces recent entries discovery after VS Code 1.118
moved `history.recentlyOpenedPathsList` to the shared application
storage database.
The plugin now probes both the legacy `User/globalStorage/state.vscdb`
database and the new shared storage database for VS Code Stable,
Insiders, Exploration, VSCodium, VSCodium Insiders, and portable
installs.
It also deduplicates results that may appear in both locations.
Fixes#47445
## Validation
- Reviewed the change against the issue's documented VS Code 1.118
storage paths.
- Attempted local focused build from
`src/modules/launcher/Plugins/Community.PowerToys.Run.Plugin.VSCodeWorkspaces`
with `tools/build/build.cmd -Platform x64 -Configuration Debug`.
- Build was blocked by local environment/tooling issues unrelated to
this C# change: missing/invalid VC tooling/Windows SDK.
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [X] Closes: #48443
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
The PR modifies the `CharacterMappings.cs` file by adding the Philippine
peso symbol to the symbols for the "P" key under the Currency group.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
No validation steps were performed as I assume this would not create any
breaking changes.
## Summary
Refines the **Grab and Move** drag/resize overlay so it matches the
polish of **Always on Top (AoT)**, and lowers the AoT default border
thickness. Created at the request of @crutkas.
Two related border-refinement changes, kept in one PR because the Grab
and Move "double layer" is designed around AoT's border.
https://github.com/user-attachments/assets/0b605f92-60bd-44a0-a540-70e6d425146a
### 1. Always on Top - default border thickness 15 -> 4
The default highlight border was `15px`, which is visually heavy.
Dropped to `4px` for a tighter, Fluent-style frame.
- `src/modules/alwaysontop/AlwaysOnTop/Settings.h` (C++ default)
- `src/settings-ui/Settings.UI.Library/AlwaysOnTopProperties.cs`
(`DefaultFrameThickness`)
- Existing users keep their configured value; only fresh installs /
"reset" pick up `4`. Slider range (1-30) is unchanged.
### 2. Grab and Move - tight, warning-gold overlay (fill + border)
Previously the overlay was a full translucent **white wash** sized to
`GetWindowRect`, which includes the invisible resize-border / shadow
margins (~7px) - so it sat *off* the visible window. It now hugs the
visible frame, mirroring AoT:
- **Keeps the translucent white wash** over the visible window (the
familiar "grabbed" feedback) and adds a tight **warning-gold border on
top**. Both hug the visible frame and are rounded to match the window
corners.
- **Tight geometry:** anchored to `DWMWA_EXTENDED_FRAME_BOUNDS` (inset
by the invisible-border margins) instead of `GetWindowRect`.
- **Corner detection:** matches the window's corner radius via
`DWMWA_WINDOW_CORNER_PREFERENCE` (same mapping AoT uses); border
thickness and radius scale with the target window DPI.
- **Distinct accent:** Fluent **warning gold `#FFB900`** - the literal
equivalent of WinUI
[`SystemFillColorCaution`](https://learn.microsoft.com/en-us/windows/apps/design/style/color)
(used as a `ThemeResource` for warnings across the Settings UI; a Win32
layered window can't resolve a `ThemeResource`, so a literal is
required). Keeps Grab and Move visually distinct from AoT's accent-blue.
- **Double layer, for free:** the Grab and Move border is drawn just
**inside** the visible edge, while AoT draws its border just **outside**
the visible edge. The two naturally stack into a clean double layer, so
Grab and Move stays a constant **4px** with no AoT detection / window
enumeration.
Rendering keeps the existing GDI + `UpdateLayeredWindow` per-pixel-alpha
path and adds **GDI+** (a Windows system library - no new third-party
dependency) for the antialiased, rounded fill and border. Frame metrics
are computed **once per drag/resize** (never in the mouse-move hot
path). The optional geometry label is unchanged.
## Before / After
| | Before | After |
|---|---|---|
| Grab and Move overlay | Full white wash, offset from the window edge |
Same wash, now tight to the visible frame + gold border, corner-matched
|
| AoT default border | 15px | 4px |
| AoT + Grab and Move together | white wash over AoT border | GM gold
inside the edge + AoT accent outside it = double layer |
## Validation
- Builds clean (exit 0, 0 warnings/errors) for **x64 Debug**:
`GrabAndMove`, `AlwaysOnTop`, and `Settings.UI.Library` (Code Analysis /
C26451 clean).
- Smoke-tested live by running the standalone module exes: tight gold
border + wash on Alt-drag / Alt-right-drag, AoT 4px border, and the
inside/outside double layer on a pinned window.
- WARNING: still **draft** pending broader visual validation (border
tightness across DPIs, the exact gold, rounded vs square corners, AoT
z-order during fast drags - AoT renders from a separate process and
follows on a ~100ms timer). Screenshots to be added.
## Follow-up (not in this PR)
AoT and Grab and Move remain **separate** overlay systems (AoT:
persistent per-window Direct2D border; Grab and Move: transient
GDI/`UpdateLayeredWindow` overlay). They can't share one runtime window,
but the frame-geometry + corner-detection + DPI helpers are worth
extracting into `src/common` (seeded by AoT's
`WindowCornersUtil`/`ScalingUtils`). Tracked separately to keep this PR
atomic (`src/common` is an ABI-careful area).
## Notes
- No IPC/JSON schema changes; no new settings.
- No new third-party dependencies (GDI+ is a system library).
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This pull request adds a custom `docmd` plugin to improve how
documentation links to source files are handled, ensuring that
repo-root-relative links work both when editing locally and on the
published site. The main changes include introducing the new plugin,
updating configuration to use it, and documenting its behavior.
**Plugin integration and configuration:**
* Added a new local plugin `github-source-links` in `docmd-plugins/`,
which rewrites repo-root-relative links (e.g., `/src/.../Foo.cpp`) in
Markdown files to absolute GitHub blob URLs during the documentation
build process, ensuring links remain functional on the published site.
[[1]](diffhunk://#diff-c2c746e6974a6cfdd229031c2977f2bb0dca37c6d5f598ed45dc0d9f0b74c7caR1-R52)
[[2]](diffhunk://#diff-a97f3ce59c97313aacd716b9874b445d162c45ee2e6ef9fd2db59fe17235e1cfR1-R8)
* Updated `docmd.config.json` to register the new plugin under the
`plugins` key, enabling it for documentation builds.
* Updated `package.json` to include the plugin as a dependency,
referencing the local plugin directory.
**Documentation updates:**
* Updated `README.md` to document the new `docmd-plugins/` folder,
explain the purpose of repo-root-relative links, and describe how the
plugin rewrites these links for the published site.
This pull request introduces a new, automated workflow for building and
publishing the developer documentation website using
[docmd](https://docmd.io/). The static site is now generated from
`doc/devdocs`, built in the `doc/devdocs-website` folder, and deployed
to GitHub Pages via a GitHub Actions workflow. The build output is not
committed to the repository but is instead published as an artifact.
Supporting configuration files, documentation, and `.gitignore` entries
are also added to streamline local development and CI/CD.
**Automated build and deployment:**
* Added `.github/workflows/regenerate-devdocs-website.yml` to build the
static site with docmd and deploy it to GitHub Pages automatically on
changes to `doc/devdocs` or `doc/devdocs-website`, or via manual
trigger.
**Project setup and configuration:**
* Added `doc/devdocs-website/package.json` to define the Node.js
project, pin the docmd version, and provide scripts for local
development and builds.
* Added `doc/devdocs-website/docmd.config.json` to configure docmd (site
title, source, output directory, base path).
* Added `doc/devdocs-website/.npmrc` to disable lockfile generation,
ensuring fresh dependency installs each build.
**Documentation and housekeeping:**
* Added `doc/devdocs-website/README.md` with instructions for editing,
building, and publishing the docs website.
* Added `doc/devdocs-website/.gitignore` to exclude the generated
`site/` output from version control.
This pull request updates the `auto-labeler.yml` GitHub Actions workflow
to improve reliability and maintain compatibility. The most important
changes are an upgrade to the `actions/github-script` version and
improved error handling when applying labels, ensuring the workflow does
not fail due to restricted permissions.
**Dependency upgrade:**
* Upgraded `actions/github-script` from version 7 to version 9 in the
`Apply area labels with AI` step, ensuring continued support and access
to the latest features and security updates.
**Error handling improvements:**
* Added a `try/catch` block around the label application logic to
gracefully handle cases where the workflow lacks permission to write
labels (e.g., due to restricted integration tokens), logging a message
and skipping the operation instead of failing the entire workflow.
This is a totally minor nitpick.
I don't have the dock enabled on all my displays. But the current "pin
to dock" dialog lets me pin it to a display I don't have the dock on.
When that happens, it effectively results in _nothing_ happening. Not
great.
This PR mitigates that situation, but only listing the enabled docks
when pinning.
This PR fixes Issue #48680 where the CmdPal dock performance meter shows
'???' after restart.
**Root cause**: The PerformanceWidgetsPage initially returns items with
a placeholder title ('???') because ContentData hasn't loaded yet. The
DataManager timer starts when the dock subscribes to ItemsChanged, and
the Updated event fires 1 second later with real data. However, the
Updated event handlers were updating ListItem.Title directly without
calling RaiseItemsChanged() on the page, so the dock was never notified
that the items had changed and continued to display the stale '???'
title.
**Fix**: Added a RaiseItemsChanged() call to each of the 5 Updated event
handlers (CPU, Memory, Network, GPU, Battery) after the item titles are
updated. This causes the dock to re-call GetItems() and refresh the
displayed titles.
Fixes#48680
This pull request makes minor adjustments to the
`.github/workflows/auto-labeler.yml` GitHub Actions workflow, focusing
on permissions and event triggers.
Workflow configuration updates:
* Added `pull-requests: write` permission to the workflow to ensure it
has the necessary access for managing pull requests.
* Removed the `edited` event from the list of triggers for
`pull_request_target`, so the workflow will no longer run when a pull
request is edited.
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Changed the icon from the sleep icon to the hibernate icon, which fixes
the issue in #48535
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #48535
## Detailed Description of the Pull Request / Additional comments
The icon was previously the sleep icon, and I have changed it to the
hibernate icon.
## Validation Steps Performed
I have visually checked that the icon is now updated to the correct
icon.
Regressed in one of the last two releases.
The problem was that `history.ToImmutableList()` ran on the projected
`IVector<string>`. `ImmutableList.CreateRange` checks for
`IReadOnlyCollection<string>`, and resolving that interface on a WinRT
object requires a helper type that AOT can't generate.
So we have to just _not do that_.
Closes#48445
## Summary of the Pull Request
Gives the CmdPal toast notification a glow-up: it now slides in/out with
a nice fade, has acrylic + a soft shadow, and stops fighting with
`SizeToContent`.
Along the way, the toast guts got refactored into a couple of reusable
bits that live in `PowerToys.Common.UI.Controls` so other PowerToys
utilities can grab them for their own transient overlays:
- **`TransparentWindow`** — a `WindowEx`-derived host that strips the
native frame, hides from taskbar/Alt-Tab, uses `TransparentTintBackdrop`
for transparency, and runs show/hide implicit animations on its content.
Supply your own animations via `ShowAnimations` / `HideAnimations`, or
take the defaults (fade + slide).
- **`TransparentCard`** — a templated `ContentControl` with acrylic
(`AlwaysActiveDesktopAcrylicBackdrop`), rounded corners, border, and
shadow. Drop whatever XAML you want inside.
- **`AlwaysActiveDesktopAcrylicBackdrop`** — small `SystemBackdrop`
wrapper so the acrylic doesn''t go grey when the window isn''t focused
(transient overlays are never focused).
CmdPal''s `ToastWindow` is now basically a 16-line wrapper: derives from
`TransparentWindow`, drops a bound `TextBlock` inside, and handles its
own 2.5s auto-hide timer + bottom-center positioning.
https://github.com/user-attachments/assets/3a62080c-22f0-480c-ac4d-028bcc32f07d
## PR Checklist
- [x] Closes: #40886
- [x] **Communication:** I''ve discussed this with core contributors
already.
- [ ] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
## Detailed Description of the Pull Request / Additional comments
A couple of design notes worth calling out:
- The window is sized slightly bigger than the visible card (~24px
breathing room on each side) so the shadow and slide animation have room
to render without clipping. That buffer area is transparent but NOT
click-through — kept it small on purpose. We explored `SetWindowRgn` and
`EnableWindow` tricks to make it click-through too, but neither plays
nicely with WinUI 3''s DesktopWindowXamlSource. Small transparent frame
is the pragmatic compromise.
- Animations use the Toolkit''s implicit `ShowAnimations` /
`HideAnimations` so there''s zero animation code in
`ToastWindow.xaml.cs`.
- `TransparentCard.xaml` is registered in `Themes/Generic.xaml` —
required for templated controls in a library project;
`<GenerateLibraryLayout>` alone doesn''t auto-merge per-control xaml.
## Validation Steps Performed
- Built clean (arm64 Debug).
- Triggered a CmdPal toast manually: fades + slides in, hangs for 2.5s,
fades + slides out.
- Acrylic stays active when the window isn''t focused (toasts are never
focused).
- Shadow renders fully without clipping.
---------
Co-authored-by: niels9001 <niels9001@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Fixes some typos and grammar mistakes inside docs, strings, and
comments.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
None
---------
Co-authored-by: Niels Laute <niels.laute@live.nl>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
This pull request updates the auto-labeling workflow to support both
issues and pull requests, improves clarity in logs and prompts, and
renames the workflow file for broader applicability. The changes enhance
automation for triaging by enabling area label assignment to new or
updated pull requests in addition to issues.
**Expanded triage coverage and workflow improvements:**
* The workflow now triggers on both issue and pull request events
(`opened`, `reopened`, `edited`, `synchronize`), allowing automatic area
labeling for pull requests as well as issues.
(`.github/workflows/auto-labeler.yml`,
[.github/workflows/auto-labeler.ymlL1-R7](diffhunk://#diff-f874b1d773361dc46f2496bc3ce97ee3441e91257188b27a2bd83693c3b8a82eL1-R7))
* The concurrency group logic has been updated to handle pull request
events separately from issues, ensuring that rapid updates to a PR or
issue are managed correctly. (`.github/workflows/auto-labeler.yml`,
[.github/workflows/auto-labeler.ymlR24-R26](diffhunk://#diff-f874b1d773361dc46f2496bc3ce97ee3441e91257188b27a2bd83693c3b8a82eR24-R26))
**User experience and clarity enhancements:**
* Console logs and prompt messages now refer generically to "item" or
distinguish between "Issue" and "Pull request" as appropriate, improving
clarity in workflow output and AI prompts.
(`.github/workflows/auto-labeler.yml`,
[[1]](diffhunk://#diff-f874b1d773361dc46f2496bc3ce97ee3441e91257188b27a2bd83693c3b8a82eL41-R44)
[[2]](diffhunk://#diff-f874b1d773361dc46f2496bc3ce97ee3441e91257188b27a2bd83693c3b8a82eL58-R79)
[[3]](diffhunk://#diff-f874b1d773361dc46f2496bc3ce97ee3441e91257188b27a2bd83693c3b8a82eL139-R146)
[[4]](diffhunk://#diff-f874b1d773361dc46f2496bc3ce97ee3441e91257188b27a2bd83693c3b8a82eL212-R217)
* The system prompt for the AI labeling assistant has been updated to
clarify that both issues and pull requests should be classified and
labeled. (`.github/workflows/auto-labeler.yml`,
[.github/workflows/auto-labeler.ymlL127-R133](diffhunk://#diff-f874b1d773361dc46f2496bc3ce97ee3441e91257188b27a2bd83693c3b8a82eL127-R133))
**File naming:**
* The workflow file has been renamed from
`.github/workflows/auto-label-issues.yml` to
`.github/workflows/auto-labeler.yml` to reflect its broader scope.
(`.github/workflows/auto-labeler.yml`,
[.github/workflows/auto-labeler.ymlL1-R7](diffhunk://#diff-f874b1d773361dc46f2496bc3ce97ee3441e91257188b27a2bd83693c3b8a82eL1-R7))
Adds a `SettingsManager` scaffold to the CmdPal extension template so
new extensions have a ready-to-use settings pattern out of the box.
## Summary of the Pull Request
Most extensions need configuration. The template previously had no
settings infrastructure, leaving developers to discover the pattern from
built-in extensions. This adds a wired-up `SettingsManager` following
the same conventions used across all built-in extensions.
## PR Checklist
- [ ] Closes: #xxx
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
**`Helpers/SettingsManager.cs`** (new)
- Extends `JsonSettingsManager` (already bundled in
`Microsoft.CommandPalette.Extensions`)
- Uses `Utilities.BaseSettingsPath()` for correct local-state path
(packaged and unpackaged)
- Includes commented-out `ToggleSetting` example illustrating the full
add/register/expose cycle
- Auto-saves via `Settings.SettingsChanged` event
**`TemplateCmdPalExtensionCommandsProvider.cs`** (updated)
- Instantiates `SettingsManager` as a field
- Sets `Settings = _settingsManager.Settings` to advertise settings to
CmdPal
- Exposes `_settingsManager.Settings.SettingsPage` in `MoreCommands` for
in-app settings access
```csharp
// Helpers/SettingsManager.cs
internal sealed partial class SettingsManager : JsonSettingsManager
{
private static readonly string _namespace = "templatecmdpalextension";
private static string Namespaced(string propertyName) => $"{_namespace}.{propertyName}";
// TODO: Add your settings here. For example:
// private readonly ToggleSetting _myToggle = new(
// Namespaced(nameof(MyToggle)), "My toggle setting", "Description", false);
// public bool MyToggle => _myToggle.Value;
public SettingsManager()
{
FilePath = SettingsJsonPath();
// Settings.Add(_myToggle);
LoadSettings();
Settings.SettingsChanged += (_, _) => SaveSettings();
}
}
// TemplateCmdPalExtensionCommandsProvider.cs
private readonly SettingsManager _settingsManager = new();
public TemplateCmdPalExtensionCommandsProvider()
{
// ...
_commands = [
new CommandItem(new TemplateCmdPalExtensionPage())
{
Title = DisplayName,
MoreCommands = [new CommandContextItem(_settingsManager.Settings.SettingsPage)],
},
];
Settings = _settingsManager.Settings;
}
```
The template wizard replaces `"TemplateCmdPalExtension"` throughout all
files, so `BaseSettingsPath` and the namespace string will update
automatically when scaffolding a new extension.
## Validation Steps Performed
- Verified `SettingsManager` pattern matches existing built-in
extensions (`ClipboardHistory`, `System`, `Registry`)
- No new package references required — `JsonSettingsManager`,
`ToggleSetting`, and `Utilities` are all included in the existing
`Microsoft.CommandPalette.Extensions` NuGet package
<!-- START COPILOT ORIGINAL PROMPT -->
<details>
<summary>Original prompt</summary>
>
> ----
>
> *This section details on the original issue you should resolve*
>
> <issue_title>CmdPal: Include basic SettingsManager in
Template</issue_title>
> <issue_description>### Description of the new feature / enhancement
>
> Include basic `SettingsManager` in template.
>
> ### Scenario when this would be used?
>
> I believe most extensions will require some form of configuration, so
adding basic `SettingsManager` would be helpful.
>
> ### Supporting information
>
> _No response_</issue_description>
>
> ## Comments on the Issue (you are @copilot in this section)
>
> <comments>
> </comments>
>
</details>
<!-- START COPILOT CODING AGENT SUFFIX -->
- Fixesmicrosoft/PowerToys#39274
<!-- START COPILOT CODING AGENT TIPS -->
---
🔒 GitHub Advanced Security automatically protects Copilot coding agent
pull requests. You can protect all pull requests by enabling Advanced
Security for your repositories. [Learn more about Advanced
Security.](https://gh.io/cca-advanced-security)
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: niels9001 <9866362+niels9001@users.noreply.github.com>
Co-authored-by: Muyuan Li (from Dev Box) <muyuanli@microsoft.com>
## Summary
Fixes incorrect Chinese translation where both Sleep and Hibernate
commands showed '休眠' (hibernation) instead of using distinct terms: '睡眠'
for Sleep and '休眠' for Hibernate.
## Changes
Added \<comment>\ elements with \ZH:\ guidance to 6 resource entries in
\Microsoft.CmdPal.Ext.System\\Properties\\Resources.resx\:
| Resource Key | Current Comment | Added ZH Guidance |
|---|---|---|
| \Microsoft_plugin_command_name_hibernate\ | *(none)* | translate as
'休眠' |
| \Microsoft_plugin_sys_hibernate\ | Existing English comment | use 休眠
not 睡眠 |
| \Microsoft_plugin_sys_hibernate_confirmation\ | Existing English
comment | use 休眠 not 睡眠 |
| \Microsoft_plugin_command_name_sleep\ | *(none)* | translate as '睡眠',
do NOT use '休眠' |
| \Microsoft_plugin_sys_sleep\ | Existing English comment | use 睡眠 not
休眠 |
| \Microsoft_plugin_sys_sleep_confirmation\ | Existing English comment |
use 睡眠 not 休眠 |
## Why comments only?
Chinese translations are managed by the internal CDPX localization
pipeline. Adding \<comment>\ elements to the English \.resx\ file is the
standard way to guide translators for the next localization pass. See
similar fix in PR #48649 for Japanese translation guidance.
## Summary
Adds translation guidance comments to English .resx resource strings
that have incorrect Japanese translations in the CDPX localization
pipeline.
## Problem
- **Download/Downloading** — translated as 受け取り (implies physical
package receipt) instead of the correct IT term 受信
- **Battery** (laptop/tablet context) — translated as 電池 (dry cell)
instead of バッテリー (built-in battery)
## Fix
Added comment elements to 9 resource entries across 3 files with the
prefix JA: to guide the localization team:
| File | Strings | Comment |
|------|---------|---------|
| Microsoft.CmdPal.Ext.WinGet\Properties\Resources.resx |
winget_downloading, winget_download_progress | JA: translate as 受信 not
受け取り |
| Microsoft.CmdPal.UI.ViewModels\Properties\Resources.resx |
winget_operation_status_downloading,
winget_operation_status_downloading_percent,
gallery_item_winget_action_downloading_with_progress,
gallery_item_winget_action_downloading | JA: translate as 受信 not 受け取り |
| Microsoft.CmdPal.Ext.WindowsSettings\Properties\Resources.resx |
BatterySaver, BatterySaverSettings, BatteryUse | JA: translate as バッテリー
not 電池 |
## Background
As documented in doc/devdocs/development/localization.md, localized
.resx files are generated at build time by the CDPX pipeline from .lcl
files managed by the internal localization team. The comment elements in
English .resx files are surfaced to translators and are the correct
mechanism for providing translation guidance. The actual .lcl file fixes
must be applied by the internal team.
Closes#48598
## Summary
Adds the two missing top-level exception handlers in the QuickAccess
(Preview) flyout host so that an unhandled XAML exception during launch
or page navigation no longer FailFasts `PowerToys.QuickAccess.exe`.
Spotted while reading through `App.OnLaunched` and `ShellPage` for an
unrelated review of the flyout startup path — none of the existing
handlers exist yet, so any throw during `MainWindow` construction,
`ShellHost.Initialize`, or `ContentFrame.Navigate(typeof(LaunchPage) |
typeof(AppsListPage), …)` bubbles all the way out to the Windows App SDK
runtime and is stowed as a XAML failure. Compare with
`src\settings-ui\Settings.UI\SettingsXAML\App.xaml.cs`, which already
wires `UnhandledException += App_UnhandledException`.
## Changes
**`src\settings-ui\QuickAccess.UI\QuickAccessXAML\App.xaml.cs`**
- Hook `Application.UnhandledException` in the constructor. The handler
logs the exception via `ManagedCommon.Logger.LogError` (same logger
Settings uses) and sets `e.Handled = true`. QuickAccess is a transient
launcher flyout owned by the runner, so swallowing a stray XAML error
and keeping the host alive for the next summon is the correct trade-off
— the failure is still recorded for diagnostics.
- Wrap the body of `OnLaunched` in a try/catch. If `MainWindow` (which
sets up window chrome, listener threads, the IPC coordinator, and the
XAML shell) fails to construct, log the exception and call `Exit()`
cleanly rather than letting the throw escape into the Windows App SDK
launch path.
**`src\settings-ui\QuickAccess.UI\QuickAccessXAML\Flyout\ShellPage.xaml.cs`**
- Subscribe to `ContentFrame.NavigationFailed` after
`InitializeComponent`. A page constructor or XAML-load failure in
`LaunchPage` / `AppsListPage` would otherwise bubble out of the `Frame`
and crash the launcher. The handler logs the failure
(`SourcePageType.FullName` + the exception) and marks it handled so the
next summon retries navigation.
No production behaviour changes when things work — only the failure
paths are different. No public API surface changes.
## Why both handlers, not just one
- `Application.UnhandledException` does not fire for
`Frame.NavigationFailed`. The Frame raises its own event first and, if
no handler runs or `e.Handled` is left `false`, then it rethrows on the
dispatcher.
- Conversely, `Frame.NavigationFailed` only fires for navigation
failures — not for an exception thrown directly in `OnLaunched` before
any navigation happens.
The two events are complementary, so both need a handler to fully cover
the launch + navigation paths.
## Testing
- The local NuGet feed on my dev box currently can't restore
`Microsoft.NETCore.App.Runtime.win-x64 = 10.0.9` (the feed only has
`11.0.0-preview.1.26104.118`), which fails the project restore for every
WinUI project including this one. That's the same environment issue I
called out on #48414 — pipeline restore uses a different feed and is
fine.
- All three patterns added here are copy-paste analogues of code that
already exists in `Settings.UI` (`App.xaml.cs:96, 106-109`,
`ShellViewModel.cs:86, 136`), so namespace and signature drift risk is
minimal. The only behavioural difference is `e.Handled = true`, which is
the actual goal of this PR.
## Risk
- Low. Two new event handlers and one try/catch. No behaviour change on
the success path.
- Worst-case regression is that a real, repeatable XAML failure becomes
silent in the runner's eyes (no process crash) instead of loud — but
it's logged via `Logger.LogError` so the user can still find the trace
in `%LOCALAPPDATA%\Microsoft\PowerToys\Logs\`.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
---
ADO:
https://microsoft.visualstudio.com/DefaultCollection/OS/_workitems/edit/61258633/
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #46464
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Yu Leng (from Dev Box) <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Addresses the **root cause chain** behind the NullReferenceException
crash in Settings navigation. Crash dump analysis with WinDbg revealed
that `Frame.Navigate()` can throw native WinUI ABI exceptions that
propagate unhandled through multiple code paths.
### Changes (3 files)
**NavigationService.cs** — Defensive try-catch for all
`Frame.Navigate()` call sites:
- `Navigate()`: Wrap in try-catch, log via `Logger.LogError()`, return
`false` on failure
- `EnsurePageIsSelected()`: Add matching try-catch for consistency (also
calls `Frame.Navigate()` unprotected)
- Protects **all** navigation in the Settings app
**ShellViewModel.cs** — Fix the crash-causing `Frame_NavigationFailed`
handler:
- Replace `throw e.Exception` with `e.Handled = true` +
`Logger.LogError()`
- The original code threw `NullReferenceException` when `e.Exception`
was null (native WinUI errors don't always marshal to managed Exception
objects)
- Mark the event as handled to prevent framework error propagation
**ShellPage.xaml.cs** — Protect `async void SearchBox_QuerySubmitted`:
- Wrap entire method body in try-catch
- `async void` methods that throw after `await` produce unhandled
exceptions that crash the process
- Covers both search indexing (`Task.Run`) and navigation failures
### Crash Dump Evidence
Analysis of `PowerToys.Settings_2025_03_26_48636.dmp` with WinDbg:
- 4 stowed exception records found at `0x000001c60d3b9100`
- Exception chain: `SearchBox_QuerySubmitted` → `Frame.Navigate()` fails
at native ABI → `NavigationFailed` fires with null Exception → `throw
null` → `NullReferenceException` → `FailFastWithStowedExceptions`
- Root failure in `Microsoft.UI.Xaml.dll!DirectUI::Frame::Navigate`
### Review Notes
- `catch (Exception)` pattern matches 92.5% of Settings.UI codebase
convention
- `Logger.LogError()` with `using ManagedCommon` matches standard import
pattern
- No security concerns: logged page type names are not sensitive data
- Build verified locally (38/38 applicable tests pass; 111 COM-dependent
tests fail identically on main)
Co-authored-by: Tucker Burns <tuck.burns@gmail.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
When a modifier key (Ctrl/Alt/Shift) is remapped to a non-modifier key
using
Keyboard Manager, the injected key event is delivered to applications as
WM_SYSKEYDOWN instead of WM_KEYDOWN. This causes unexpected behavior —
for
example, remapping Left Alt to Backspace results in whole words being
deleted
instead of single characters, because applications interpret
WM_SYSKEYDOWN +
VK_BACK as Alt+Backspace.
The fix resets the modifier state with a suppress-flag key-up event
before
injecting the target key, consistent with the existing approach used for
the
Caps Lock remapping scenario.
## PR Checklist
- [ ] Closes: #47191
- [ ] Communication: I've discussed this with core contributors already.
If the work hasn't been agreed, this work might be rejected
- [x] Tests: Added/updated and all pass
- [ ] Localization: All end-user-facing strings can be localized
- [ ] Dev docs: Added/updated
- [ ] New binaries: Added on the required places
## Detailed Description of the Pull Request / Additional comments
The root cause is that SendInput is called inside the low-level keyboard
hook
callback before the original modifier event is suppressed. At that point
the
modifier state is still active, so the OS delivers the injected key as
WM_SYSKEYDOWN (system key with Alt context) rather than WM_KEYDOWN.
This is the same mechanism that was already fixed for the Ctrl/Alt/Shift
↔
Caps Lock case. This PR extends the fix to cover modifier → non-modifier
remaps.
## Validation Steps Performed
1. Remapped Left Alt → Backspace in Keyboard Manager
2. Opened a text editor, typed text, pressed the remapped key
3. Confirmed single characters are deleted instead of whole words
4. All 93 unit tests pass (KeyboardManager.Engine.UnitTests)
## Summary
Add `docs/superpowers/` to `.gitignore` so all locally generated
superpowers artifacts (specs, design docs, and plans) are not committed.
These files are produced by local AI tooling and are workspace-local
only; they should not land in the repository.
## Change
```gitignore
# Superpowers-generated docs (specs, design, plans) — local-only, not committed
docs/superpowers/
```
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
## Summary
Upgrades the centrally-managed Windows App SDK package versions to the
**2.2.0 stable** umbrella released on NuGet.
| Package | Before | After |
|---|---|---|
| `Microsoft.WindowsAppSDK` | 2.0.1 | **2.2.0** |
| `Microsoft.WindowsAppSDK.Foundation` | 2.0.20 | **2.1.0** |
| `Microsoft.WindowsAppSDK.AI` | 2.0.185 | **2.2.3** |
| `Microsoft.WindowsAppSDK.Runtime` | 2.0.1 | **2.2.0** |
Foundation/AI/Runtime versions match the dependency graph declared by
`Microsoft.WindowsAppSDK` `2.2.0`'s own nuspec (`Foundation=2.1.0`,
`AI=2.2.3`, `Runtime=[2.2.0]`), so transitive resolution is exact and no
version-conflict warnings are introduced.
Also bumps the CmdPal `ExtensionTemplate` sample's local
`Directory.Packages.props` so the template stays in sync with the main
repo.
## Files changed
- `Directory.Packages.props`
-
`src/modules/cmdpal/ExtensionTemplate/TemplateCmdPalExtension/Directory.Packages.props`
## Validation
Ran `tools/build/build-essentials.cmd` on a clean `origin/main`
worktree:
- `msbuild PowerToys.slnx /t:restore /p:RestorePackagesConfig=true` —
**Build succeeded, 0 warnings, 0 errors** (00:02:50)
- `src/runner/runner.vcxproj` (x64 Debug) — **Build succeeded, 0
warnings, 0 errors** (00:04:15)
- `src/settings-ui/Settings.UI/PowerToys.Settings.csproj` (x64 Debug) —
**Build succeeded, 0 warnings, 0 errors** (00:03:14)
Full module test suite has **not** been run yet — this PR only certifies
the build-essentials baseline. CI will exercise the wider build/test
matrix.
## Notes
- This is an atomic packaging-only change. No source code touched, no
behavior changes.
- If WinAppSDK 2.2.0 surfaces any runtime regression downstream, it can
be reverted as a single commit.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Power Display could put a monitor to sleep but never wake it back up.
Selecting **On** in the per-monitor power-state list was a hard-coded
no-op, so the DDC/CI wake command (VCP `0xD6` = `0x01`) was never sent.
This removes that guard so selecting **On** wakes the display, and
cleans up the dead code/comment left behind by the original
one-directional design.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #48428
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
`MonitorViewModel.HandlePowerStateSelectionChanged` early-returned when
the selected power state was **On** (`0x01`), so `SetPowerStateAsync`
was never called for On and the wake write never reached the monitor. As
a result Power Display's power control was one-directional: it could
send Standby/Suspend/Off but could never turn a monitor back on.
The guard dates back to the very first power-state commit and was paired
with a single-monitor assumption — *"the monitor must be on to see the
UI"*, so On was treated as the always-current state and skipped. A later
change made the selection reflect the monitor's real power state (so a
monitor in the list can legitimately be asleep), and multi-monitor
support means the flyout can be shown on monitor A while the user wants
to wake monitor B. Those changes invalidated the assumption, but the
action-side guard survived a subsequent refactor.
The lower layers already do the right thing:
`MonitorManager.SetPowerStateAsync` →
`DdcCiController.SetPowerStateAsync` → `SetVcpFeatureAsync(monitor,
0xD6, value)` passes the value through unchanged, so the fix is purely
removing the UI-layer guard. DDC/CI stays reachable while the panel is
in Standby/Suspend/Off(DPM), so writing `0x01` turns it back on (this is
the same mechanism Twinkle Tray uses). `Off (Hard)` / `0x05` may still
require a physical wake on some monitors, since that state can cut the
DDC command channel.
Cleanup included in this PR:
- Removed the now-unused `PowerStateItem.PowerStateOn` constant (its
only consumer was the deleted guard).
- Removed the dead `SetPowerState` `[RelayCommand]` (the generated
`SetPowerStateCommand` had zero references — the XAML wires
`SelectionChanged`, not a command).
- Updated the `SetPowerStateAsync` doc comment from the one-directional
framing to a neutral bidirectional description.
Net change: 2 files, +5 / −24.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
- **Build:** `MSBuild PowerDisplay.csproj -p:Platform=x64
-p:Configuration=Debug` (CoreCompile) — **0 errors / 0 warnings**.
- **Static check:** repo-wide grep confirms no remaining references to
`PowerStateOn` or `SetPowerStateCommand`; the power-state ListView binds
only `ItemsSource` + `SelectionChanged` (no `SelectedItem` binding), so
opening the flyout cannot spuriously re-fire a selection.
- **Manual (requires a DDC/CI monitor):** enable *Power state control*
for a monitor → open its flyout and select **Standby** or **Off (DPM)**
(screen blanks) → reopen the flyout and select **On** → the display
wakes.
No automated test was added: with the guard removed the handler is an
unconditional pass-through (identical in shape to
`HandleInputSourceSelectionChanged`), and it is an `async void` WinUI
event handler over real DDC/CI hardware, which is outside the
`PowerDisplay.Lib` unit-test seam.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Updated the DiskAnalyzer plugin link and author information to point to
the ValleySoft organization repository instead of my personal account.
## Summary of the Pull Request
Documentation-only change. Updates the DiskAnalyzer entry in
`thirdPartyRunPlugins.md`:
- Plugin URL: `thetsaw/PowerToys.Plugin` →
`valley-soft/powertoys-diskanalyzer`
- Author: `thetsaw` → `ValleySoft`
## PR Checklist
- [ ] Closes: N/A
- [x] **Communication:** This is a minor doc-only update to correct a
repo link — no prior discussion needed
- [x] **Tests:** N/A — documentation change only
- [x] **Localization:** N/A — no user-facing strings changed
- [x] **Dev docs:** Updated `doc/thirdPartyRunPlugins.md`
- [ ] **New binaries:** N/A
## Detailed Description of the Pull Request / Additional comments
The DiskAnalyzer PowerToys Run plugin was originally submitted under my
personal GitHub account (`thetsaw`). The project has since been moved to
the official **ValleySoft** organization at:
https://github.com/valley-soft/powertoys-diskanalyzer
## Validation Steps Performed
Verified all URLs are live and resolve correctly:
- https://github.com/valley-soft/powertoys-diskanalyzer✅
- https://github.com/valley-soft✅
## Summary
The `Check Spelling` workflow has been failing on PRs against `main`
(e.g. #48546) due to issues introduced by the recent ZoomIt webcam-blur
/ noise-cancellation change (#48266) plus two pre-existing duplicate
entries in `.github/actions/spell-check/excludes.txt`.
## What the bot reported
| Severity | Type | Location |
|---|---|---|
| ❌ | `forbidden-pattern` (Should be `a`) |
`src/modules/ZoomIt/ZoomIt/rnnoise/kiss_fft.h:79` — third-party kiss_fft
header contains `an fft` |
| ⚠️ | `large-file` (~30 MB) |
`src/modules/ZoomIt/ZoomIt/rnnoise/rnnoise_data_little.c` |
| ⚠️ | `binary-file` |
`src/modules/ZoomIt/ZoomIt/selfie_segmentation.onnx` |
| ⚠️ | `duplicate-pattern` ×2 | `excludes.txt` lines 115/116 duplicate
lines 108/109 (`FuzzyMatcher{Comparison,Diacritics}Tests.cs`) |
## Fix
`.github/actions/spell-check/excludes.txt`:
- **Drop 2 duplicate** `FuzzyMatcher*Tests.cs` lines.
- **Add 2 new exclusions** for the new third-party ZoomIt assets:
- `^src/modules/ZoomIt/ZoomIt/rnnoise/` — entire third-party
rnnoise/kiss_fft tree (covers both the `an fft` forbidden-pattern in
`kiss_fft.h` and the 30 MB `rnnoise_data_little.c` large-file).
- `^src/modules/ZoomIt/ZoomIt/selfie_segmentation\.onnx$` — the ML model
binary.
Net change: `-2` duplicates, `+2` new exclusions → file count unchanged
at 148 lines.
## Notes
- Third-party content under `rnnoise/` should not be spell-checked; this
matches how other vendored/third-party trees in the repo are handled
(e.g. `src/common/CalculatorEngineCommon/exprtk.hpp`,
`src/common/sysinternals/Eula/`).
- No source code changes; pure config.
- Unblocks #48546 and any other PR currently failing `Check Spelling` on
`main`.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Clicking the clock dock band in the CmdPal Dock now opens the Windows
notification center (Action Center). A separate bell-icon-only dock band
is also exposed for users who prefer a dedicated notification center
shortcut.
Closes#46327
## Detail
- **Clock band left-click**: replaced the previous `NoOpCommand` on
`NowDockBand` with `OpenUrlCommand("ms-actioncenter:")`, dismissing the
Dock on invoke. The `ms-actioncenter:` URI is the correct shell
mechanism - `SendInput` Win+N was tested but dropped because it requires
foreground focus, which the Dock holds at click time.
- **Notification center band**: new `NotificationCenterDockBand`
(`ListItem`) in `TimeDateCommandsProvider.cs`, with a bell icon
(`\uEA8F`, Segoe Fluent Icons) and the same `ms-actioncenter:` command.
Exposed as a second `WrappedDockItem` from `GetDockBands()` under the id
`com.microsoft.cmdpal.timedate.notificationCenterBand`. Users can pin it
from the Dock's edit mode.
- **New resource strings**:
`timedate_show_notification_center_command_name` and
`timedate_notification_center_band_title` added to `Resources.resx` /
`Resources.Designer.cs`.
- **VS 2026 C++ build fixes** (pre-existing failures on `HEAD`): added
`_SILENCE_EXPERIMENTAL_COROUTINE_DEPRECATION_WARNINGS` to
`CalculatorEngineCommon.vcxproj`.
## Screenshots
<img width="339" height="991" alt="image"
src="https://github.com/user-attachments/assets/e0ef8c9a-ec1f-40fa-9620-1e83e6aeeb8d"
/>
## How tested
- Built `Microsoft.CmdPal.UI.csproj` (Debug x64) - 0 errors.
- Launched dev `Microsoft.CmdPal.UI.exe`, clicked the clock band -
notification center opened correctly.
- Right-click context menu on the clock band still shows "Copy time" and
"Copy date" unchanged.
- Pinned the notification center band via edit mode - bell icon renders
icon-only, click opens notification center.
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
This change adds the [RNNoise](https://github.com/xiph/rnnoise) filter
for noise cancellation (audio) and the [Google
mediapipe](https://github.com/google-ai-edge/mediapipe/tree/master)
`selfie_segmentation_cpu` model for webcam background detection and
blurring.
It also fixes an issue introduced with
ba68b88ca1 causing the ZoomIt shortcuts to
fail to register in the standalone version.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
The Settings UI has been extended with a Noise cancellation option, a
Background selection for the webcam and a Brightness slider.
The functionality for these is added to ZoomIt itself. Also, restored
the Mono checkbox which was accidentally masked by
b93fd97e80.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Mario Hewardt <marioh@microsoft.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
## Summary of the Pull Request
Adds linked brightness control to PowerDisplay so multiple
brightness-capable monitors can be controlled from a single "All
Displays" slider.
This PR:
- Adds a linked brightness mode with one master brightness slider.
- Seeds the master slider from the linked display with the lowest
Windows DISPLAY number, falling back to monitor ID for determinism.
- Persists linked mode enabled/disabled state.
- Persists per-monitor exclusions by monitor ID.
- Keeps individual display cards available under an expandable section
while linked mode is enabled.
- Shows linked-state guidance in the link icon tooltip instead of a
separate info banner.
- Allows excluded displays to keep their own independent brightness
slider.
- Keeps profiles as per-monitor snapshots; applying a profile turns
linked brightness off before applying the profile values.
- Adds unit tests for linked-brightness selection/seed behavior and
settings compatibility.
## PR Checklist
- [x] Closes: #47319
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
**Screenshots**
| State | Light | Dark |
| --- | --- | --- |
| Linked mode off | <img width="519" height="817" alt="image"
src="https://github.com/user-attachments/assets/bdfae94b-b2e2-4ad3-a45c-7925bb9e5dcd"
/> | <img width="520" height="817" alt="image"
src="https://github.com/user-attachments/assets/69290a70-0375-480d-957c-c9e0af43d18e"
/> |
| Linked mode on | <img width="520" height="307" alt="image"
src="https://github.com/user-attachments/assets/a2b3572b-e51f-4bdc-9209-23ad2f96d27a"
/> | <img width="520" height="307" alt="image"
src="https://github.com/user-attachments/assets/8b14b665-b641-4256-a15b-eced82e62728"
/> |
| Linked mode on — individual displays expanded | <img width="520"
height="895" alt="image"
src="https://github.com/user-attachments/assets/0b40e60d-e78a-4814-baf6-00be7e283edd"
/> | <img width="520" height="895" alt="image"
src="https://github.com/user-attachments/assets/4f59bbfa-d6e5-4cb7-af84-cb484f922a7c"
/> |
The first version is intentionally scoped to brightness-only linked
control. Contrast, volume, color temperature, input source, and
LightSwitch-specific behavior remain independent.
Linked brightness is stored as global PowerDisplay settings:
- `linked_levels_active`
- `excluded_from_sync_monitor_ids`
Newly connected brightness-capable monitors are included by default,
because the exclusion list is the explicit exception. Hotplugging a
monitor does not immediately write brightness; linked hardware writes
happen only after the user changes the master slider.
Profiles remain per-monitor snapshots. This PR does not add
profile-level linked brightness configuration. If linked brightness is
active when a profile is applied, linked mode is turned off first, then
the saved per-monitor profile values are applied. That avoids leaving
the master linked slider active while hardware brightness has been
changed independently per monitor.
When linked mode is turned on, the master slider is seeded from the
linked brightness-capable display with the lowest Windows DISPLAY
number, falling back to monitor ID for determinism. Excluded displays
and displays without brightness support are ignored; if no linked target
remains, the master slider stays disabled. The seed only positions the
slider; it is never written to hardware, so the first user gesture is
the first broadcast.
## Validation Steps Performed
- Built `PowerDisplay.Lib.UnitTests` Debug x64:
```powershell
.\tools\build\build.ps1 -Platform x64 -Configuration Debug -Path src\modules\powerdisplay\PowerDisplay.Lib.UnitTests
```
- Ran `PowerDisplay.Lib.UnitTests` with `vstest.console.exe`
- Ran the XAML styling script:
```powershell
.\.pipelines\applyXamlStyling.ps1 -Main
```
- Result: the XAML styling script completed successfully and processed
`src/modules/powerdisplay/PowerDisplay/PowerDisplayXAML/MainWindow.xaml`.
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
This fixes a bug introduced by
`ba68b88ca1617e52647c6dde467c56f53ca2422a` in the hotkey processing
logic.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
A complex condition ended up being incorrectly broken into 2 conditions,
leading to a missed `else` execution. This led to the mishandling of
keyboard shortcuts especially during reassignment in the standalone mode
for ZoomIt.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
## Summary of the Pull Request
This PR fixes Shortcut Guide foreground app detection by resolving app
IDs from the window that was in the foreground before the Shortcut Guide
UI takes focus.
Based on review feedback, it also adds the missing XML `<param>`
documentation for `foregroundWindowHandle` in
`ManifestInterpreter.GetAllCurrentApplicationIds(...)` to satisfy
documentation/style requirements.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
### Functional change
- Capture and reuse the foreground window handle taken before Shortcut
Guide UI activation.
- Use that captured handle for current-application ID resolution instead
of querying foreground window later, improving app-specific shortcut
matching reliability.
### Follow-up feedback fix
- Added missing XML parameter documentation:
-
`src/modules/ShortcutGuide/ShortcutGuide.Ui/Helpers/ManifestInterpreter.cs`
- Added `<param name="foregroundWindowHandle">...</param>`
## Validation Steps Performed
- Ran `parallel_validation` (Code Review: no issues; CodeQL: skipped as
trivial doc-only follow-up).
- Attempted local `dotnet build` for `ShortcutGuide.Ui.csproj`; blocked
by transient external package feed/network failure while restoring
`Microsoft.Build.CopyOnWrite/1.0.282`.
---------
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
## Summary
Flips the default for the `Shortcut Guide` entry in `enabledModules`
from `true` to `false`. New PowerToys installs (no prior
`settings.json`) will start with Shortcut Guide disabled, matching how
other newer modules (PowerToys Run, MouseJump, AdvancedPaste,
MouseWithoutBorders, CropAndLock, QuickAccent, TextExtractor,
MousePointerCrosshairs, KeyboardManager) already ship off-by-default in
`EnabledModules.cs`.
## Why
Shortcut Guide has been on-by-default since the early days, but it is an
opt-in style utility (overlay launched on hotkey). It should not be
active for users who never asked for it on a fresh install.
## Changes
- `src/settings-ui/Settings.UI.Library/EnabledModules.cs` -- flip
`shortcutGuide` backing field from `= true` to bare declaration with the
file's `// defaulting to off` comment convention.
- `src/settings-ui/Settings.UI.UnitTests/ViewModelTests/General.cs` --
update the corresponding `AllModulesAreEnabledByDefault` assertion.
## Compatibility
Existing installs are unaffected: their `settings.json` already persists
the user's prior value (`true`), so anyone who has it on today keeps it
on. Only freshly created `EnabledModules` instances pick up the new
default.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Visual polish for the Shortcut Guide UI, follow-up to #48383 and #48384.
- Fixes the Settings footer icon being clipped in the rail at startup.
Root cause: the icon `RowDefinition` in `CustomNavigationViewStyle.xaml`
was `*`, so the rail's `MinHeight` constraint compressed the 20px icon
down. Changing it to `Auto` lets the row size to its content. This also
removes the need for the `FakeSettingsButton` workaround.
- Replaces the laptop `FontIcon` used for the `Windows` nav entry with a
4-square Windows logo rendered as a `PathIcon`.
- Replaces the bitmap PowerToys app icon with a theme-aware vector
`PathIcon` (rounded square frame + menu bar + three dots) so it matches
the rest of the icons.
- Path data for both icons lives as `x:String` resources in `App.xaml`.
- Tunes the rail item `Row 0` height so the icon vertically aligns with
the selection pill center.
<img width="681" height="1405" alt="image"
src="https://github.com/user-attachments/assets/10787087-e32f-4018-b004-3f824648b962"
/>
## PR Checklist
- [x] **Closes:** part of the Shortcut Guide 0.100 polish set
- [x] **Communication:** internal only
- [x] **Tests:** manual verification
- [x] **Localization:** no new strings
- [x] **Dev docs:** N/A
## Validation Steps Performed
- Built locally and verified the Settings footer icon is no longer
clipped on first show.
- Verified the Windows nav entry now uses the Windows logo PathIcon and
renders correctly in light and dark.
- Verified the PowerToys entry now renders as a vector and follows the
theme foreground.
- Verified the selection pill aligns vertically with the icon center.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Transitions the telemetry PR detection workflow from testing phase to
ready status. All components are now fully functional and automatically
triggered on every new PR.
- Skips checks on draft PRs to avoid noise
- Prevents multiple concurrent runs per PR
- Safely requests `@chatasweetie` as reviewer on telemetry changes
- update commit messages
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Carlos Zamora <carlos.zamora@microsoft.com>
We have no idea why this event isn't flowing in 0.99. We fixed the other
"new" event, `CmdPal_ExtensionInvoked`, when we merged #47121. But this
event didn't show up.
So, as a blind experiment, we're removing the potentially long payload.
Just to see if something happens.
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #47572
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
In the powerdisplay tool, Added a new button, that allows to copy the
current display configuration, to allow to share for troubleshooting, or
just sharing.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Fixes labels not associated with controls in Installed Apps pane.
Narrator only announces "space, checkbox, checked/unchecked" for
Extensions > Installed Apps page checkboxes instead of reading the label
too.
this is an a11y internal bug
The Adaptive Cards renderer (`AdaptiveCards.Rendering.WinUI3` v2.x)
renders `Input.Toggle` as a CheckBox whose `Content` is a TextBlock
containing just `" "`. Without an AutomationProperties.Name`, Narrator
reads the CheckBox.Content (a space character), making the settings
inaccessible to screen reader users.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Michael Jolley <mike@baldbeardedbuilder.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Fixes#48224
The Performance Monitor extension was still storing its settings in the
shared settings.json file. Since Command Palette built-in extensions now
use extension-specific sibling settings files, the extension's settings
could be overwritten when Command Palette personalization settings were
saved.
This change updates SettingsJsonPath() to store Performance Monitor
settings in an extension-specific settings file
(performanceMonitor.settings.json), ensuring the network speed unit
setting persists correctly.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #48224
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
The Performance Monitor extension had not been migrated to the newer
extension-specific settings file architecture. As a result, its settings
were stored in the shared settings.json file and could be lost when the
Command Palette host rewrote its configuration.
Following @zadjii-msft's guidance in the issue, this PR updates
SettingsManager.cs to store Performance Monitor settings in an
extension-specific settings file:
performanceMonitor.settings.json
instead of the shared:
settings.json
This aligns the extension with the current Command Palette settings
architecture and prevents the network speed unit setting from being
overwritten.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
* Changed Network Speed Unit to "Binary bytes per second" in Performance
Monitor settings.
* Verified the setting was successfully saved to the newly created
`performanceMonitor.settings.json` file.
* Modified core Command Palette personalization settings.
* Verified the Performance Monitor setting was preserved and not
overwritten.
* Restarted PowerToys and verified the setting persisted correctly.
**Note to reviewers:** I left the "Tests" box unchecked because this is
a single-line file path configuration change. I did not add an automated
test, but I have thoroughly verified the fix manually as described in
the Validation steps above.
## Summary of the Pull Request
Fixes#48053 — Remote Desktop extension now allows connecting to
arbitrary hostnames when navigated into the extension's list page.
Previously, `RemoteDesktopListPage` only displayed previously saved
connections. This converts it from `ListPage` to `DynamicListPage`,
adding search-driven behavior:
- Typing a valid hostname/IP that **doesn't** match an existing saved
connection shows a **"Connect to {hostname}"** item at the top of the
list
- Typing something that **exactly matches** a saved connection shows
only the normal results (no duplicate arbitrary-host item)
- Typing an invalid string (not a valid hostname/IP) shows no
arbitrary-host item
The hostname validation reuses the same logic as
`FallbackRemoteDesktopItem`: strip the port suffix (e.g. `myhost:3389` →
`myhost`), then check `Uri.CheckHostName` against `IPv4`, `IPv6`, `Dns`.
## PR Checklist
- [x] Closes: #48053
- [ ] **Communication:** I've discussed this with core contributors
already.
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
---------
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Rework the confirmation dialog shown when enabling Power Display (and
its potentially-destructive sub-features) so it is shorter, friendlier,
and consistent across all entry points. The five separate prefix-driven
variants are now a single `PowerDisplayWarningDialog` user control
selected via an enum, sharing the title, learn-more link, and
Enable/Cancel buttons. The previous hand-rolled red warning text is
replaced with a Fluent `InfoBar Severity=""Warning""`.
## PR Checklist
- [ ] Closes: #xxx
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
## Detailed Description of the Pull Request / Additional comments
### Before
- `DangerousFeatureWarningDialog` took a resource-key prefix string and
probed up to five optional keys per variant (`_WarningHeader`,
`_WarningConfirm`, `_WarningList_Item1/2`, etc.).
- Each of the five flows (EnableModule, ColorTemperature, PowerState,
InputSource, MaxCompatibility) had a slightly different title (some
questions, some statements, mixed `Warning:` prefixes) and a hand-rolled
red `TextBlock` warning header with a `⚠️` emoji and
`SystemFillColorCriticalBrush`.
- ~30 fragmented `PowerDisplay_*_Warning*` resw keys.
### After
- New `PowerDisplayWarningDialog` selected via `PowerDisplayWarningKind`
enum (`EnableModule`, `ColorTemperature`, `PowerState`, `InputSource`,
`MaxCompatibility`).
- Shared chrome lives in the control:
- Single title `Before you continue` for every variant.
- `InfoBar Severity=""Warning""` replaces the hand-rolled red header.
- Learn-more `HyperlinkButton` pointing at
`aka.ms/powerToysOverview_PowerDisplay_Note` (URL is a `private const`
so translators don't see it).
- Consistent Enable / Cancel buttons.
- Per-variant content collapses to one InfoBar message + one body
paragraph in resw (12 keys total, down from ~30). Bullets are inlined as
`• ` + newlines with `xml:space=""preserve""`.
- `PowerDisplayViewModel.ConfirmDangerousFeatureAsync` and
`TryCommitDangerousChangeAsync` now take the enum instead of a magic
string.
### Files
- **Added:** `ViewModels/PowerDisplayWarningKind.cs`,
`SettingsXAML/Views/PowerDisplayWarningDialog.xaml{,.cs}`
- **Removed:**
`SettingsXAML/Views/DangerousFeatureWarningDialog.xaml{,.cs}`
- **Updated:** `Strings/en-us/Resources.resw`,
`ViewModels/PowerDisplayViewModel.cs`,
`SettingsXAML/Views/PowerDisplayPage.xaml.cs`
## Validation Steps Performed
- Built `PowerToys.Settings.csproj` (Debug arm64) — clean.
- Manually exercised the EnableModule and MaxCompatibility flows;
verified the new title, InfoBar, body paragraph, learn-more link, and
Enable/Cancel button behavior.
- Verified `aka.ms/powerToysOverview_PowerDisplay_Note` opens in the
default browser.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Make the new WinUI 3 Keyboard Manager editor the default by flipping
`useNewEditor` from `false` to `true` everywhere a default is supplied.
## Behavior
- **New installs / new `settings.json`** → new editor enabled
- **Upgrades from a version before the `useNewEditor` key existed** →
new editor enabled (the key is missing, so the default kicks in)
- **Users who have explicitly toggled the setting** → unchanged (we only
change the default, not stored values)
- The "Go back to classic" button in the KBM settings page is untouched
## Changes
- `src/settings-ui/Settings.UI.Library/KeyboardManagerProperties.cs` —
`UseNewEditor` property initializer now `true`
- `src/modules/keyboardmanager/dll/dllmain.cpp` — `m_useNewEditor`
member init + `GetNamedBoolean` fallback now `true`; warn-log message
updated to match
-
`src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.PowerToys/Modules/KeyboardManagerModuleCommandProvider.cs`
— both fallback paths in `IsUseNewEditorEnabled` (file missing /
unreadable) now return `true`, so the "Open New Editor" CmdPal command
surfaces by default
## Validation
Built locally on ARM64 Debug (exit code 0):
- `src/modules/keyboardmanager/dll`
- `src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.PowerToys`
- `src/settings-ui/Settings.UI.Library`
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Reword the Shortcut Guide module description and OOBE description so
they describe the feature without referring to `your apps`. The module
description now reads as a single sentence covering Windows and the
active app; the OOBE description is shortened to one paragraph and
refers to `various applications` instead of enumerating the bundled
apps.
## PR Checklist
- [ ] Closes: #xxx
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
## Detailed Description of the Pull Request / Additional comments
Two strings in
`src/settings-ui/Settings.UI/Strings/en-us/Resources.resw` changed:
- `ShortcutGuide.ModuleDescription` — now: `Shows an on-screen overlay
of keyboard shortcuts for Windows and the active application.`
- `Oobe_ShortcutGuide.Description` — collapsed to one sentence
describing Windows + various applications, no longer enumerating bundled
apps or mentioning future additions.
No code or behavioral changes.
## Validation Steps Performed
- Built Settings.UI (Debug arm64) – clean.
- Verified the Shortcut Guide module card in Settings UI shows the new
description and the OOBE Shortcut Guide page shows the new paragraph.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
## Summary
Fixes#46055 — Standardizes built-in fallback title/subtitle format so
scoring is consistent across all action fallbacks.
## Problem
`MainListPage.cs` scores fallback items by fuzzy-matching the query
against both Title and Subtitle, but with different weights:
- `nameScore = FuzzyScore(query, Title)`
- `descriptionScore = (FuzzyScore(query, Subtitle) - 4) / 2`
Fallbacks that embedded the raw query in Title got artificially higher
scores than those using Subtitle. This made ranking unpredictable.
## Fix
All "action" fallbacks now follow a consistent pattern:
- **Title** = static action description (no query text)
- **Subtitle** = raw query string (unquoted)
"Result" fallbacks (that found a specific matched item) are left
unchanged — they correctly show the matched item name in Title.
## Full Fallback Audit (example query: `notepad`)
| Fallback | Title | Subtitle | Status |
|---|---|---|---|
| WebSearch: Search | "Search the web with Edge" | `Search for notepad`
| **Changed** — was `Search for "notepad"` in subtitle |
| WebSearch: Open URL | "Open in Microsoft Edge" | `Open notepad.com` |
**Changed** — was `Open "notepad.com"` in title |
| Shell: Run | "Run" | `notepad` | Unchanged — already correct |
| Calculator | "3" (result) | `1+2` (query) | Unchanged — intentional
exception |
| Indexer: single result | "notepad.exe" | `C:\Windows\notepad.exe` |
Unchanged — result fallback |
| Indexer: multiple results | "File search" | `Search for notepad in
files` | **Changed** — was `Search for "notepad" in files` in title |
| Windows Settings: single | "Notepad settings" | `Settings > Apps` |
Unchanged — result fallback |
| Windows Settings: multiple | "Search Windows settings..." | `Search
for notepad` | **Changed** — was `Search for "notepad" in Windows
settings` in title |
| Remote Desktop: exact match | "MyPC" (connection) | "Connect to MyPC"
| Unchanged — result fallback |
| Remote Desktop: arbitrary host | "Remote Desktop" | `Connect to
notepad-host` | **Changed** — was `Connect to notepad-host` in title |
| TimeDate | "Monday, May 23" (result) | "Current date" | Unchanged —
result fallback |
| System | "Shut down" (result) | "Shuts down the computer" | Unchanged
— result fallback |
| PowerToys | "Color Picker" (static) | "Pick a color..." | Unchanged —
result fallback |
## Changes
- `FallbackExecuteSearchItem.cs` — Subtitle uses raw query instead of
`"Search for \"{query}\""` format
- `FallbackOpenURLItem.cs` — Title shows browser name (was query),
Subtitle shows raw query (was browser)
- `FallbackOpenFileItem.cs` — Multi-result: Title is static display
name, Subtitle is raw query
- `FallbackWindowsSettingsItem.cs` — Multi-result: Title is static
description, Subtitle is raw query
- `FallbackRemoteDesktopItem.cs` — Arbitrary host: Title is static
"Remote Desktop", Subtitle is raw query
---------
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
When an extension updates its `Subtitle` property asynchronously after
initial render, the `TextVisibilityStates` visual state group
transitions from `TitleOnly` → `TextVisible`. This transition sets
`SubtitleText.Visibility = Visible`, overriding the `CompactStates`
setter that had hidden it.
## Fix
Added `control.UpdateCompactState()` to `OnTextPropertyChanged` in
`DockItemControl.xaml.cs`. This re-applies the compact state after any
text property change. When `IsCompact` is `false`, `UpdateCompactState`
is a no-op — no behavior change for the non-compact path.
Fixes#47980
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Fixes#47939
The Performance Monitor dock band displayed network stats as Receive →
Send, but Task Manager shows Send → Receive. This swaps the order to
match Task Manager.
## Changes
- `PerformanceWidgetsPage.cs`: Swap `_networkUpItem` (Send) before
`_networkDownItem` (Receive) in the band items array.
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
When a keyboard shortcut opens CmdPal to an extension while the palette
is already showing a dock-launched transient page, `GoHome(false)`
cannot restore the root page — the frame's back stack is empty because
the transient dock page was never pushed on top of root. The user ends
up with only the hotkey-target page in the frame with no way to navigate
back to the main list.
## Root Cause
In `ShellPage.SummonOnUiThread()`, the hotkey-to-page branch called
`GoHome(false)` before sending `ShowWindowMessage`. But when the active
page is a transient dock page, `_currentlyTransient` is still `true` and
the frame back stack is empty, so `GoHome` can't re-establish the root
page as the frame base.
## Fix
Added `ResetToHome()` to `ShellViewModel`, mirroring the pattern already
used in `WindowHiddenMessage` handling:
1. Clears `_currentlyTransient`
2. Calls `_rootPageService.GoHome()` to reset extension state
3. Sends `PerformCommandMessage` for `_rootPage` — navigating
MainListPage into the frame as the base
In `ShellPage.SummonOnUiThread()`, the `GoHome(false)` call in the
`isPage` branch is replaced with `ViewModel.ResetToHome()`. The root
page is then cleanly in the frame before the hotkey target's
`PerformCommandMessage` navigates on top of it.
Fixes#47994
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Reorder the Pin to Dock dialog content so the configuration controls
(monitor selector, dock section, label options) appear at the top and
the live preview is shown below them. The user now configures the pin
first and sees the resulting preview directly underneath, instead of
staring at the preview and having to scan past it to find the controls.
<img width="515" height="440" alt="image"
src="https://github.com/user-attachments/assets/0d1d0543-2b30-48f5-a1aa-676a165870f5"
/>
## PR Checklist
- [ ] Closes: #xxx
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
## Detailed Description of the Pull Request / Additional comments
XAML-only change to
`src/modules/cmdpal/Microsoft.CmdPal.UI/Dock/PinToDockDialogContent.xaml`.
The new visual order inside the `ScrollViewer`/`StackPanel` is:
1. Monitor selector (still `Visibility=""Collapsed""` by default; shown
when more than one monitor is available)
2. Dock section `Segmented` (Start / Center / End)
3. Label options (`Show title` / `Show subtitle` checkboxes)
4. Divider `Rectangle`
5. Preview `Border`
No logic, bindings, `x:Name` identifiers, event handlers, or `x:Uid`
keys are changed.
## Validation Steps Performed
- Built `Microsoft.CmdPal.UI` (Debug arm64) — clean.
- Verified the Pin to Dock dialog renders with controls on top and the
preview underneath; segmented selection, label-option checkboxes, and
the multi-monitor combo still behave as before.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Fixes#47821
The GPU Performance Monitor widget crashes with
`IndexOutOfRangeException` on systems where GPU performance counters
fail to enumerate (common on Intel Arc and hybrid GPU configurations).
The dock band shows `???` and opening the flyout causes an error.
## Root Cause
`GPUStats.CreateGPUImageUrl()` accessed `_stats[index]` without bounds
checking. When `GetGPUPerfCounters()` finds no matching counter
instances, `_stats` remains empty but callers still pass index 0.
`GetGPUName()`, `GetGPUUsage()`, and `GetGPUTemperature()` already have
proper guards (`if (_stats.Count <= index) return ...`) — this fix adds
the same pattern to the one remaining unguarded method.
## Changes
- `GPUStats.cs`: Add bounds check to `CreateGPUImageUrl()` — return
empty string if index out of range
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Addenda to #47187
that fix only works if the _hwnd is already set. Actually it's crazy it
ever worked.
Tested by disconnecting and reconnecting RDP a couple times, which
pretty consistently reproduces the problem.
## Summary
Fixes#48170 — ShortcutGuide v2 crashes on launch when the bundled
`Manifests` directory is absent from the install path.
### Root Cause
The `Assets\ShortcutGuide\Manifests\*.yml` files were never reaching the
build output directory during the CI solution-level build (`msbuild
PowerToys.slnx /t:Build -graph`). The `CopyToOutputDirectory` metadata
on `<Content>` items does not reliably copy files to a shared
`OutputPath` in this build configuration. As a result, the WiX installer
generator found no yml files to package, and the installed product was
missing the Manifests directory entirely.
At runtime, `PowerToysShortcutsPopulator.Populate()` threw an unhandled
`FileNotFoundException` causing a crash loop.
### Fix (3 layers)
1. **Code resilience** (`Program.cs`, `PowerToysShortcutsPopulator.cs`):
- Wrap `Populate()` in try/catch so a missing manifest degrades
gracefully instead of crashing
- Add `File.Exists` guard before `File.ReadAllText`
2. **Build output** (`ShortcutGuide.Ui.csproj`):
- Add explicit `CopyManifestsToOutputDir` MSBuild target
(`AfterTargets="Build"`) that copies yml files to
`$(OutDir)Assets\ShortcutGuide\Manifests\` — same pattern as the
existing `CopyPRIFileToOutputDir` target
- Keep `<Content Include>` with `CopyToOutputDirectory` as a fallback
for publish scenarios
3. **Installer packaging** (`generateAllFileComponents.ps1`,
`ShortcutGuide.wxs`):
- Add `*.yml` to the file inclusion list
- Add `Generate-FileList` / `Generate-FileComponents` calls for
`ShortcutGuideManifestsFiles`
- Add WiX directory definition and `RemoveFolder` component for the
Manifests directory
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Converts the "Show details" context menu command into a toggle that
switches between "Show details" and "Hide details" with appropriate
icons, and fixes the icon not rendering in the context menu.
Address internal a11y bug.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
## Summary of the Pull Request
Adds **DiskAnalyzer** to the General plugins table in
`doc/thirdPartyRunPlugins.md`.
- **Plugin:**
[Community.PowerToys.Run.Plugin.DiskAnalyzer](https://github.com/thetsaw/PowerToys.Plugin)
- - **Author:** thetsaw
- - **Keyword:** `ds`
- - **License:** MIT
- - **Platforms:** x64 and ARM64
### What it does
Scan any folder or drive to find the largest files and subfolders, view
drive usage with visual progress bars, and navigate your filesystem all
from PowerToys Run.
## PR Checklist
- [x] Plugin has been publicly available
- [ ] - [x] MIT licensed
- [ ] - [x] Releases include x64 and ARM64 zips
- [ ] - [x] plugin.json is correctly formatted
- [ ] - [x] README includes install instructions
## Detailed Description
This is a documentation-only change adding one row to the third-party
plugins table. No source code, binaries, or build files are modified.
## Summary of the Pull Request
This updates PowerToys Settings to remove the obsolete “V2” suffix from
the Shortcut Guide module name. The UI now consistently shows **Shortcut
Guide**.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
- **Settings navigation label**
- Updated `Shell_ShortcutGuide.Content` to `Shortcut Guide`.
- **Module title**
- Updated `ShortcutGuide.ModuleTitle` to `Shortcut Guide`.
- **OOBE title**
- Updated `Oobe_ShortcutGuide.Title` to `Shortcut Guide`.
```xml
<data name="Shell_ShortcutGuide.Content" xml:space="preserve">
<value>Shortcut Guide</value>
</data>
```
## Validation Steps Performed
- N/A for behavior-level validation in this description (change is
limited to localized display strings).
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Both Power Display and Grab and Move have matured beyond their initial
release phase. This removes the "NEW" `InfoBadge` from their navigation
items in Settings, the two parent navigation groups (Windowing &
Layouts, Input / Output) that surfaced the badge when collapsed, and
clears the `IsNew` flag for Power Display in the OOBE shell.
<img width="1867" height="973" alt="image"
src="https://github.com/user-attachments/assets/533f271c-c70f-414f-a76a-43fd9ffbbd44"
/>
<img width="497" height="575" alt="image"
src="https://github.com/user-attachments/assets/fe1e97c3-c806-4f42-a836-76e042630d61"
/>
<img width="1619" height="1027" alt="image"
src="https://github.com/user-attachments/assets/f5db715b-bc69-4505-803a-18a9b2716280"
/>
## PR Checklist
- [x] Closes: #48153
- [x] **Communication:** Tracked by the linked issue
- [x] **Tests:** Markup-only change; Settings.UI builds clean with WinUI
markup compiler (no XAML errors)
- [x] **Localization:** No end-user-facing strings changed
- [ ] **Dev docs:** N/A
- [ ] **New binaries:** N/A
- [ ] **Documentation updated:** N/A
## Detailed Description of the Pull Request / Additional comments
Files touched:
- `src/settings-ui/Settings.UI/SettingsXAML/Views/ShellPage.xaml` —
removed four `<InfoBadge Style="{StaticResource NewInfoBadge}" />`
blocks on `GrabAndMoveNavigationItem`, `PowerDisplayNavigationItem`, and
on the two parent group headers `WindowingAndLayoutsNavigationItem` and
`InputOutputNavigationItem` (the parent badges existed only to surface a
NEW child when the group was collapsed; with no NEW children left in
those groups, the parent badges are now stale).
- `src/settings-ui/Settings.UI/OOBE/ViewModel/OobeShellViewModel.cs` —
flipped `(PowerToysModules.PowerDisplay, true)` to
`(PowerToysModules.PowerDisplay, false)`. Grab and Move was already
`false`.
No other modules or strings affected.
## Validation Steps Performed
- Built `src\settings-ui\Settings.UI\PowerToys.Settings.csproj`
(Release|x64) with MSBuild from VS 18 Enterprise;
`PowerToys.Settings.dll` produced with 0 errors related to this change.
WinUI markup compiler would have aborted before producing the DLL if the
XAML had syntax issues.
- Diff inspected: only the five intended deletions/edits, no collateral
changes.
- Visual run-time verification of the Settings navigation pane is
recommended before merge.
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Cooperative shutdowns of `PowerDisplay.exe` — Runner's `TerminateApp`
NamedPipe message, the `Terminate` named event, tray-quit, Runner-exit
detection, and PowerToys upgrades — all call `Environment.Exit(0)`
immediately. If DDC/CI discovery is mid-flight, that path skips the
`try/finally` that owns `CrashDetectionScope`, leaving `discovery.lock`
on disk. Phase 0 at the next `PowerDisplay.exe` startup then treats this
orphan as evidence of a real crash and auto-disables the module,
surfacing the "PowerDisplay has crashed" InfoBar in Settings UI.
This PR adds an `AppDomain.ProcessExit` safety-net inside
`CrashDetectionScope`. ProcessExit fires for `Environment.Exit` but
**not** for `FailFast` / BSOD / external `TerminateProcess` — exactly
the partition we need: cooperative exit → best-effort delete the lock;
involuntary kill → leave the lock for Phase 0 to detect (original design
intent preserved).
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #48169
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
<!-- no user-facing strings changed -->
- [x] **Dev docs:** Added/updated <!-- inline XML doc on
CrashDetectionScope explains the ProcessExit partition -->
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
### Root cause
`CrashDetectionScope.Begin()` writes `discovery.lock` before DDC/CI
capability fetch and `Dispose()` deletes it when the `using` block
exits. The lock is intentionally designed to survive any code path that
cannot run user-mode cleanup (BSOD, kernel OOM, `TerminateProcess`), so
that the next `PowerDisplay.exe` start can see it and run Phase 0 (write
`crash_detected.flag`, set `enabled.PowerDisplay=false` in global
`settings.json`, signal `AutoDisablePowerDisplayEvent`).
The bug is that several **cooperative** shutdown paths route to
`Environment.Exit(0)` immediately:
| Path | Code |
|---|---|
| Runner's `TerminateApp` NamedPipe | `App.xaml.cs::OnNamedPipeMessage`
→ `Shutdown()` → `Environment.Exit(0)` |
| `Terminate` named event | `App.xaml.cs::OnLaunched` →
`RegisterEvent(..., () => Environment.Exit(0), "Terminate")` |
| Tray-quit | `TrayIconService` callback → `Environment.Exit(0)` |
| Runner-exit detection | `RunnerHelper.WaitForPowerToysRunner` callback
→ `Environment.Exit(0)` |
`Environment.Exit` calls `ExitProcess` under the hood, which terminates
all threads abruptly. Background `Task.WhenAll` doing DDC capability
fetch is killed mid-flight; the `finally` block that calls
`scope.Dispose()` never runs; `discovery.lock` orphans; Phase 0 next
time false-positives.
Concrete repro from logs:
- `15:08:42.510` lock written
- `15:08:42.79` probe monitor #1
- `15:08:46.92` probe monitor #2 (started, not finished — typical probe
takes ~5s)
- `15:08:49.03` `TerminateApp` received → `Environment.Exit(0)` → no
`Dispose` log line
- `15:10:10.03` next startup: Phase 0 sees orphan lock with `pid:17712,
startedAt:2026-05-28T07:08:42Z` → writes `crash_detected.flag` →
auto-disables
### Fix
`CrashDetectionScope.Begin()` now also subscribes to
`AppDomain.CurrentDomain.ProcessExit`. The handler does a best-effort
`File.Delete(_lockPath)` (swallowing exceptions, as required for
ProcessExit handlers). `Dispose()` unsubscribes before deleting. An
`Interlocked.Exchange` guards the race between Dispose and ProcessExit
so only one of the two performs the delete.
ProcessExit's semantics match the cooperative/involuntary partition
exactly:
| Shutdown path | ProcessExit fires? | Behavior after this PR |
|---|---|---|
| `Environment.Exit(code)` (all 4 paths above) | yes | lock deleted by
handler |
| `Environment.FailFast` | no | lock survives → Phase 0 catches it
(correct: explicit FailFast = real failure) |
| BSOD / external `TerminateProcess` / kernel OOM | no | lock survives →
Phase 0 catches it (correct: original design) |
| Discovery completes normally / throws | n/a | `try/finally` calls
`Dispose()` as before; handler unsubscribed first |
### Testability
A new `IProcessExitHook` interface abstracts the subscription so unit
tests can simulate ProcessExit without terminating the test runner.
Production code uses the default `AppDomainProcessExitHook` singleton;
tests inject a fake whose `RaiseExit()` invokes subscribed handlers
synchronously.
### Files touched
-
`src/modules/powerdisplay/PowerDisplay.Lib/Services/IProcessExitHook.cs`
*(new)* — interface + production singleton
-
`src/modules/powerdisplay/PowerDisplay.Lib/Services/CrashDetectionScope.cs`
— subscribe in `Begin`, unsubscribe in `Dispose`, add `OnProcessExit`
handler, expanded class doc
-
`src/modules/powerdisplay/PowerDisplay.Lib.UnitTests/CrashDetectionScopeTests.cs`
*(new)* — 10 unit tests
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
### Automated
10 new unit tests in `CrashDetectionScopeTests`, all passing:
```
Passed Begin_WritesLockFileAtomically
Passed Begin_SubscribesToProcessExit
Passed Dispose_UnsubscribesFromProcessExit
Passed Dispose_DeletesLockFile
Passed ProcessExitFired_BeforeDispose_DeletesLock (core scenario)
Passed ProcessExitFired_AfterDispose_DoesNothing
Passed Dispose_AfterProcessExit_DoesNotThrow
Passed ProcessExitFired_LockFileMissing_DoesNotThrow
Passed Dispose_IsIdempotent
Passed MultipleScopes_DoNotShareState
```
Full `PowerDisplay.Lib.UnitTests` suite: **129 / 132 passing**. The 3
failures (`DetectOrphanAndDisable_RunsFullSequenceWhenOrphanPresent`,
`DetectOrphanAndDisable_HandlesUnknownVersionAsOrphan`,
`DetectOrphanAndDisable_LeavesLockIntactOnSignalFailure`) are
**pre-existing on `main`** — they fail with `REGDB_E_CLASSNOTREG` from
`Constants.AutoDisablePowerDisplayEvent()` (WinRT activation factory not
COM-registered in the test environment). Verified by stashing this PR's
changes and re-running the same 3 tests on baseline `main` — same
failures, same cause, unrelated to this change.
### Manual
1. Reproduced the original false-positive on `main`:
- Enable PowerDisplay → open Settings UI → quickly toggle PowerDisplay
off
- Observe `discovery.lock` left in
`%LOCALAPPDATA%\Microsoft\PowerToys\PowerDisplay\`
- Re-enable PowerDisplay → Phase 0 writes `crash_detected.flag` →
InfoBar appears
2. Repeated the same steps with this branch:
- Toggling PowerDisplay off cleanly deletes `discovery.lock`
(ProcessExit handler ran)
- Re-enabling PowerDisplay shows no InfoBar, no `crash_detected.flag`
created
3. BSOD path is unchanged (verified by inspecting the conditional logic
— `AppDomain.ProcessExit` does not fire for involuntary terminations;
the lock survives just as before).
---------
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
## Summary of the Pull Request
Renames the OOBE welcome/overview hyperlink label from **“Documentation
on Microsoft Learn”** to **“Documentation”** for brevity and
consistency.
Scope is limited to the localized string resource used by the OOBE
overview page.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
- **Resource update (OOBE Overview)**
- Updated `Oobe_Overview_DescriptionLinkText.Text` in
`src/settings-ui/Settings.UI/Strings/en-us/Resources.resw`.
```xml
<data name="Oobe_Overview_DescriptionLinkText.Text" xml:space="preserve">
<value>Documentation</value>
</data>
```
## Validation Steps Performed
- Confirmed the OOBE overview string key now resolves to
**“Documentation”**.
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Return a friendly calculator error when Mages evaluates an expression to
a complex number instead of letting decimal conversion throw.
This fixes the PowerToys Run Calculator result for expressions such as
`sqrt(-1)` by detecting `System.Numerics.Complex` results before decimal
conversion and showing a localized error message instead.
Fixes#43937
## PR Checklist
- [x] Closes: #43937
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [x] **Dev docs:** Added/updated
- [x] **New binaries:** Added on the required places
- [x] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [x] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [x] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [x] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [x] **Documentation updated:** Not required for this bug fix.
## Detailed Description of the Pull Request / Additional comments
The Calculator plugin previously passed complex results from Mages into
`Convert.ToDecimal`, which caused an exception for expressions like
`sqrt(-1)`.
This PR updates the calculator result transformation logic to detect
`System.Numerics.Complex` and return a localized user-facing error
message: `Complex numbers are not supported`.
It also updates calculator query tests to cover both direct keyword and
global query behavior.
## Validation Steps Performed
- Added unit test coverage for `=sqrt(-1)` returning `Complex numbers
are not supported`.
- Added unit test coverage for global query `sqrt(-1)` returning no
result instead of surfacing an unhandled exception.
- Ran `git diff --check`.
- Attempted local build/test with the PowerToys build scripts, but local
validation was blocked by Visual Studio/VC tooling configuration issues
unrelated to this change: `PlatformToolsetVersion` resolves to an empty
value during restore/build.
This removes our last git submodule dependency!
We were using `expected-lite` in one place, which was being compiled out
_anyway_ in favor of using `std::expected`.
## Summary
Migrate `deps/spdlog` from a git submodule to **vcpkg manifest mode**
with an overlay port pinned to the **exact same commit**
(`gabime/spdlog@616866fc`). Replaces the polyfill shim added in #47910
with a proper port-level patch.
This is the follow-up to PR #47928, which I closed after @zadjii-msft /
@DHowett clarified that the intended direction was a single combined
"move to vcpkg **and** apply a patch file" (one change, not two stepping
stones).
## Guidance honored
Per @zadjii-msft (offline):
- ✅ Convert each submodule to vcpkg **one at a time** — this PR is
**spdlog only**. `deps/expected-lite` stays a submodule (separate PR
next).
- ✅ Atomic commit per dep (multiple commits on the branch for review
traceability; squash on merge gives the requested single commit).
- ✅ **Don't bump the version.** Only variable changed: submodule →
vcpkg. Same commit (`616866fc`, v1.8.5 + 38) the submodule pointed at.
Per @DHowett
([review](https://github.com/microsoft/PowerToys/pull/48039#pullrequestreview-4338835150)):
- ✅ No vcpkg submodule — vswhere-first detection via a Terminal-style
`steps-install-vcpkg.yml` template; three-tier `VcpkgRoot` fallback (env
var → VS-shipped → runtime clone pinned to manifest baseline).
## Design
- **Repo-root manifest**: `vcpkg.json` declares only `spdlog`, with
`builtin-baseline` pinned. `vcpkg-configuration.json` registers
`deps/vcpkg-overlays/` as overlay-ports.
- **Overlay port** `deps/vcpkg-overlays/spdlog/`: `vcpkg_from_github(REF
616866fc...)` with bundled fmt preserved (`-DSPDLOG_FMT_EXTERNAL=OFF`);
the MSVC 14.51 fix from #47910 carried as a proper vcpkg patch on
`include/spdlog/fmt/bundled/format.h`.
- **vcpkg integration is global** (set in `Cpp.Build.props`, imported
via `ForceImportBeforeCppProps` for every `.vcxproj`). An earlier
attempt to make vcpkg per-project-opt-in via `deps/spdlog.props` failed
because ~85 PowerToys `.vcxproj` files import `spdlog.props` AFTER
`Microsoft.Cpp.targets`, by which point `vcpkg.props`' `ClCompile` hook
is dead-on-arrival. The trade-off (every C++ project invokes `vcpkg
install` once at build time, ~0.5 s on cache hits, manifest declares
only spdlog so install set is fixed) is documented in the expanded
`Cpp.Build.props` comment.
- **`deps/spdlog.props`** is now a thin shim that only sets the
historical `SPDLOG_*` preprocessor defines for source-compat.
- **`Cpp.Build.targets`** is a new file imported via
`ForceImportAfterCppTargets` to load `vcpkg.targets` after
`Microsoft.Cpp.targets`. A fail-fast `<Target>` errors with a clear
message if `vcpkg.props` can't be found at the resolved `VcpkgRoot`.
- **Removes** `deps/spdlog-msvc-fix/` polyfill, in-tree wrapper
`src/logging/`, spdlog submodule, the single `<ProjectReference>` in
`logger.vcxproj`, plus 3 `.slnf` refs and 2 `.slnx` refs
(`PowerToys.slnx` + `installer/PowerToysSetup.slnx`), plus 3 hard-coded
`..\deps\spdlog\include` entries in `<AdditionalIncludeDirectories>`.
- **CI**: new reusable `.pipelines/v2/templates/steps-install-vcpkg.yml`
(vswhere-first, manifest-baseline-pinned fallback clone, respects
`useVSPreview`). Gated `Cache@2` for `%LOCALAPPDATA%\vcpkg\archives`
keyed on overlay-port contents. Same vcpkg detection added to
`tools\build\build-essentials.ps1` for local devs.
## Verification
Local build matrix (all 4 configs of `logger.vcxproj` and a
representative late-import consumer):
| Config | Result | Notes |
|--------|--------|-------|
| Release \| x64 | ✅ | vcpkg install ~21 s, `logger.lib` produced |
| Debug \| x64 | ✅ | **Validates patch fixes the actual MSVC 14.51 bug**
(`_ITERATOR_DEBUG_LEVEL > 0` → `_SECURE_SCL`) |
| Release \| ARM64 | ✅ | vcpkg cross-installs `arm64-windows-static`
spdlog in ~16 s |
| Debug \| ARM64 | ✅ | **Previously DISABLED for the in-tree spdlog**
(per `<Build Solution="Debug\|ARM64" Project="false" />` in
`PowerToysSetup.slnx`); this migration FIXES that latent gap |
| FancyZonesLib (Release \| x64) | ✅ | Late-import-pattern consumer;
previously broke in v2 |
Full PowerToys CI (x64 + arm64 + CmdPal SDK + all GitHub Actions checks)
green.
**Consumer audit**: 72 `.vcxproj` files reference `logger.vcxproj`; all
72 also import `deps/spdlog.props`. No transitive-link breakage.
## Out of scope (intentional)
- `deps/expected-lite` migration — next PR per "one-at-a-time" rule.
- Remote vcpkg binary cache (Azure Artifacts NuGet feed). Local pipeline
`Cache@2` works for now, but a remote feed survives across pipelines and
is the long-term answer. Happy to split this into a follow-up.
## Notes for review
- Patch in the overlay port is identical content to PR #47928's patch
but regenerated with LF line endings (vcpkg's `vcpkg_apply_patches` is
strict; no `--ignore-whitespace`).
- Once PowerToys eventually bumps spdlog past v1.14 (which ships fmt
10.2 and drops the affected code path), the overlay port can be deleted
and we can use upstream vcpkg's `spdlog` directly.
- Re. official-release pipelines and terrapin / less-restricted network
isolation: VS-shipped vcpkg is the primary path (no network); the
fallback clone is only exercised when VS doesn't ship vcpkg. Happy to
wire terrapin into the fallback as a follow-up if the official build
template needs it.
Closes the work tracked in #47928 (which was closed unmerged).
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Dustin L. Howett <dustin@howett.net>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Some focussed refactoring / simplifying / cleanup / delinting on the
Mouse Without Borders codebase (see [#44508 - [Mouse Without Borders] -
de-linting
codebase](https://github.com/microsoft/PowerToys/issues/44508)) now that
the Common class has been broken down.
This PR does some cleaning up on the ```Logger``` class:
* Uplifting coding style (string interpolation, pattern matching,
```var```, etc)
* Rationalising and simplifying code
* Relocating e.g. IO and UI side effects (writing to disk, displaying
dialog boxes) outside of Logger class
* Removing dead code, tightening visibility of existing code
* Added / updated tests to try to cover as much of the refactoring as
possible to prevent regressions
I've split the changes into lots of small commits - it might be easier
to review the individual commits rather than the whole PR in one go.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
### Run manual tests from [Test Checklist
Template](5bc7201ae2/doc/releases/tests-checklist-template.md (mouse-without-borders)):
* Install PowerToys on two PCs in the same local network:
- [x] Verify that PowerToys is properly installed on both PCs.
- [x] Configure Windows Firewall Rules
- ```netsh advfirewall firewall add rule
name="PowerToys.MouseWithoutBorders - mc" dir=in action=allow
program="C:\src\mc\PowerToys\x64\Debug\PowerToys.exe" enable=yes
remoteip=any profile=any protocol=tcp```
* Setup Connection:
- [x] Open MWB's settings on the first PC and click the "New Key"
button. Verify that a new security key is generated.
- [x] Copy the generated security key and paste it in the corresponding
input field in the settings of MWB on the second PC. Also enter the name
of the first PC in the required field.
- [x] Press "Connect" and verify that the machine layout now includes
two PC tiles, each displaying their respective PC names.
* Verify Connection Status:
- [x] Ensure that the border of the remote PC turns green, indicating a
successful connection.
- [x] Enter an incorrect security key and verify that the border of the
remote PC turns red, indicating a failed connection.
* Test Remote Mouse/Keyboard Control:
- [x] With the PCs connected, test the mouse/keyboard control from one
PC to another. Verify that the mouse/keyboard inputs are correctly
registered on the other PC.
- [ ] Test remote mouse/keyboard control across all four PCs, if
available. Verify that inputs are correctly registered on each connected
PC when the mouse is active there.
- unable to test - only 2 machines available
* Test Remote Control with Elevated Apps:
- note - the main PowerToys.exe must be running as a **non**-admin for
these tests
- [x] Open an elevated app on one of the PCs. Verify that without "Use
Service" enabled, PowerToys does not control the elevated app.
- [x] Enable "Use Service" in MWB's settings (need to run PowerToys.exe
as admin to enable "Use Service", then restart PowerToys.exe as
non-admin). Verify that PowerToys can now control the elevated app
remotely. Verify that MWB processes are running as LocalSystem, while
the MWB helper process is running non-elevated.
- ```get-process -Name "PowerToys.MouseWithoutBorders*" -IncludeUserName
| format-table Id, ProcessName, UserName```
- [x] Process: ```PowerToys.MouseWithoutBorders.exe``` - running as
```SYSTEM```
- [x] Process: ```PowerToys.MouseWithoutBorders.Helper.exe``` - running
as current user
- ```get-service -Name "PowerToys.*" | ft Status, Name, UserName;
get-ciminstance -Class "Win32_Service" -Filter "Name like 'PowerToys%'"
| ft ProcessId, Name```
- [ ] Service: ```PowerToys.MWB.Service``` - running as ```Local
System```
- [x] Toggle "Use Service" again, verify that each time you do that, the
MWB processes are restarted.
- [ ] Run PowerToys elevated on one of the machines, verify that you can
control elevated apps remotely now on that machine.
* Test Module Enable Status:
- [ ] For all combinations of "Use Service"/"Run PowerToys as admin",
try enabling/disabling MWB module and verify that it's indeed being
toggled using task manager.
* Test Disconnection/Reconnection:
- [ ] Disconnect one of the PCs from network. Verify that the machine
layout updates to reflect the disconnection.
- [ ] Do the same, but now by exiting PowerToys.
- [ ] Start PowerToys again, verify that the PCs are reconnected.
* Test Various Local Network Conditions:
- [ ] Test MWB performance under various network conditions (e.g., low
bandwidth, high latency). Verify that the tool maintains a stable
connection and functions correctly.
* Clipboard Sharing:
- [ ] Copy some text on one PC and verify that the same text can be
pasted on another PC.
- [ ] Use the screenshot key and Win+Shift+S to take a screenshot on one
PC and verify that the screenshot can be pasted on another PC.
- [ ] Copy a file in Windows Explorer and verify that the file can be
pasted on another PC. Make sure the file size is below 100MB.
- [ ] Try to copy multiple files and directories and verify that it's
not possible (only the first selected file is being copied).
* Drag and Drop:
- [ ] Drag a file from Windows Explorer on one PC, cross the screen
border onto another PC, and release it there. Verify that the file is
copied to the other PC. Make sure the file size is below 100MB.
- [ ] While dragging the file, verify that a corresponding icon is
displayed under the mouse cursor.
- [ ] Without moving the mouse from one PC to the target PC, press
CTRL+ALT+F1/2/3/4 hotkey to switch to the target PC directly and verify
that file sharing/dropping is not working.
* Lock and Unlock with "Use Service" Enabled:
- [ ] Enable "Use Service" in MWB's settings.
- [ ] Lock a remote PC using Win+L, move the mouse to it remotely, and
try to unlock it. Verify that you can unlock the remote PC.
- [ ] Disable "Use Service" in MWB's settings, lock the remote PC, move
the mouse to it remotely, and try to unlock it. Verify that you can't
unlock the remote PC.
* Test Settings:
- [ ] Change the rest of available settings on MWB page and verify that
each setting works as described.
### Group Policy Tests
See https://learn.microsoft.com/en-us/windows/powertoys/grouppolicy
- [ ] Install *.admx / *.adml and check settings behave as expected
- [ ] I'll expand the list of settings here when I get this far :-)
- [ ] HKEY_LOCAL_MACHINE\SOFTWARE\Policies\PowerToys
- [x] ConfigureEnabledUtilityMouseWithoutBorders
- [x] ```[missing]``` - "Activation -> Enable Mouse Without Borders"
enabled, with GPO warning hidden
- ```reg delete HKEY_LOCAL_MACHINE\SOFTWARE\Policies\PowerToys /v
ConfigureEnabledUtilityMouseWithoutBorders /f```
- [x] ```0``` - "Activation -> Enable Mouse Without Borders" set to
"off" and disabled, with GPO warning visible
- ```reg add HKEY_LOCAL_MACHINE\SOFTWARE\Policies\PowerToys /v
ConfigureEnabledUtilityMouseWithoutBorders /t REG_DWORD /d 0 /f```
- [x] ```1``` - "Activation -> Enable Mouse Without Borders" set to "on"
and disabled, with GPO warning visible
- ```reg add HKEY_LOCAL_MACHINE\SOFTWARE\Policies\PowerToys /v
ConfigureEnabledUtilityMouseWithoutBorders /t REG_DWORD /d 1 /f```
- [ ] MwbClipboardSharingEnabled
- [ ] MwbFileTransferEnabled
- [ ] MwbUseOriginalUserInterface
- [ ] MwbDisallowBlockingScreensaver
- [ ] MwbSameSubnetOnly
- [ ] MwbValidateRemoteIp
- [x] MwbDisableUserDefinedIpMappingRules
- [x] ```[missing]``` - "Advanced Settings -> IP address mapping"
enabled, with GPO warning hidden
- ```reg delete HKEY_LOCAL_MACHINE\SOFTWARE\Policies\PowerToys /v
MwbDisableUserDefinedIpMappingRules /f```
- [x] ```0``` - "Advanced Settings -> IP address mapping" enabled, with
GPO warning hidden
- ```reg add HKEY_LOCAL_MACHINE\SOFTWARE\Policies\PowerToys /v
MwbDisableUserDefinedIpMappingRules /t REG_DWORD /d 0 /f```
- [x] ```1``` - "Advanced Settings -> IP address mapping" disabled, with
GPO warning visible
- ```reg add HKEY_LOCAL_MACHINE\SOFTWARE\Policies\PowerToys /v
MwbDisableUserDefinedIpMappingRules /t REG_DWORD /d 1 /f```
- [x] MwbPolicyDefinedIpMappingRules
- [x] ```[missing]``` - "Advanced Settings -> IP address mapping"
enabled, with GPO warning and GPO values hidden
- ```reg delete HKEY_LOCAL_MACHINE\SOFTWARE\Policies\PowerToys /v
MwbPolicyDefinedIpMappingRules /f```
- [x] ```[empty value]``` - "Advanced Settings -> IP address mapping"
enabled, with GPO warning hidden and GPO values hidden
- ```reg add HKEY_LOCAL_MACHINE\SOFTWARE\Policies\PowerToys /v
MwbPolicyDefinedIpMappingRules /t REG_MULTI_SZ /d "" /f```
- [x] ```[non-empty value]``` - "Advanced Settings -> IP address
mapping" enabled, with GPO warning visible and GPO values visible
- ```reg add HKEY_LOCAL_MACHINE\SOFTWARE\Policies\PowerToys /v
MwbPolicyDefinedIpMappingRules /t REG_MULTI_SZ /d "aaa 10.0.0.1\0bbb
10.0.0.2" /f```
Reopens the change from #46743 (which appears to be broken) on a fresh
branch.
Original author: @daverayment
## Summary
GitHub newcomers can be confused by the current duplicate resolution
message, as it doesn't clearly point to the original referenced issue -
see
https://github.com/microsoft/PowerToys/issues/46347#issuecomment-4103681050.
They may not realise that the #12345 in the duplicate comment is the
relevant link.
This small wording update to the duplicate resolution message tightens
up wording slightly and includes reference to the prior `/dup #nnn`
comment so newcomers don't miss it.
### Before
> Hi! We've identified this issue as a duplicate of another one that
already exists on this Issue Tracker. This specific instance is being
closed in favor of tracking the concern over on the referenced thread.
Thanks for your report!
### After
> We've identified this issue as a duplicate of an existing one and are
closing this thread so discussion stays in one place.<br/><br/>Please
see the comment above for the link to the original tracking issue, and
feel free to subscribe there for updates.
## Validation Steps Performed
N/A - bot reply text change only.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
On the Command Palette Performance Monitor extension's Network widget
page, the **Send** and **Receive** list items both used the generic
`NetworkIcon` (`\uEC05`), making them visually indistinguishable at a
glance.
This PR gives each direction its own glyph:
- Send → up arrow `\uE74A`
- Receive → down arrow `\uE74B`
The redundant `↑`/`↓` characters are removed from the Send/Receive
subtitles since the icons now carry that meaning.
Before vs after:
<img width="918" height="122" alt="image"
src="https://github.com/user-attachments/assets/4af3a2fc-d5a7-4fb5-98c6-f1889c7e80f2"
/>
## PR Checklist
- [x] **Closes:** N/A (no existing issue found)
- [x] **Communication:** I've discussed this with collaborators
- [x] **Tests:** Manually verified
- [x] **Localization:** Updated en-US resw (other locales still contain
the arrow characters and can be translated/updated by the loc pipeline)
## Detailed Description of the Pull Request / Additional comments
Files changed:
- `Icons.cs` – added `NetworkUpIcon` and `NetworkDownIcon`
- `PerformanceWidgetsPage.cs` – set `Icon` on `_networkUpItem` and
`_networkDownItem`
- `Strings/en-US/Resources.resw` – `Send ↑` → `Send`, `Receive ↓` →
`Receive`
## Validation Steps Performed
Local visual verification in Command Palette.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
`ToJsonFromXmlOrCsvAsync` in `AdvancedPaste/Helpers/JsonHelper.cs`
documents that it never throws and returns an empty string on any
failure. The clipboard read at the top of the method
(`clipboardData.GetTextAsync()`) was not wrapped, so a transient
clipboard failure could surface as an exception to callers, contrary to
the documented contract.
This PR:
- Wraps `GetTextAsync()` in a try/catch and returns `string.Empty` on
failure, matching the pattern already used by the JSON/XML/CSV parsing
branches further down in the same method.
- Updates the matching unit test to decode input bytes as UTF-8
(`Encoding.UTF8.GetString(input)`) and consume the awaited task via
`GetAwaiter().GetResult()`, for consistency with sibling tests elsewhere
in the solution.
## Validation
- Local build of `AdvancedPaste.sln`. (Note: my machine has a
pre-existing NuGet SDK resolver issue unrelated to this change — the
same baseline fails on `main` for me. CI should be the source of truth.)
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Adding information about using the Command Palette Visual Studio
solution filter based on
https://github.com/microsoft/PowerToys/issues/47997#issuecomment-4524045763.
## PR Checklist
- [ ] Closes: #47997
## Detailed Description of the Pull Request / Additional comments
I found this really useful and couldn't see a reference to it anywhere
else in the devdocs. Not sure if this is the best way to describe the
process, but I think adding this information somewhere in the debugging
doc would really help newcomers to CmdPal development!
## Summary of the Pull Request
Two crash-correlation aids for the kernel-side DDC/CI BSOD mitigated by
#47734:
1. Log EDID hardware ID (manufacturer + product code, e.g. `DELD1A8`)
during Phase 0 monitor classification, before any DDC/CI capability
fetch enters the BSOD risk window.
2. Show a confirmation dialog before turning the Power Display module on
from the Settings page, so the user understands the BSOD risk before the
first capability fetch runs.
## PR Checklist
- [ ] Closes: #xxx
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
### 1. Phase 0 EdidId logging
`MonitorIdentity.EdidIdFromDevicePath` parses the EDID hardware ID
segment from a DevicePath of the form
``\?\DISPLAY#DELD1A8#5&abc&0&UID12345#{guid}`` and returns ``DELD1A8``.
The 3-letter PNP manufacturer code + 4-hex product code is identical for
every physical unit of the same model, so it identifies the *model*
without leaking per-unit identifiers.
`MonitorManager` logs the EdidId on the existing Phase 0 classification
line. Phase 0 uses `QueryDisplayConfig`, which reads OS-cached EDID and
cannot BSOD, so this line is guaranteed on disk before the crash-prone
Phase 2 capability fetch starts. If a machine crashes during
enumeration, the recovered log identifies every attached model
(including same-model duplicates), which makes it possible to correlate
crash reports to specific monitor models even when the user can't tell
us which monitor caused the crash.
### 2. Enable-module confirmation dialog
`PowerDisplayViewModel.IsEnabled` setter is refactored to follow the
same two-phase pattern already used by `MaxCompatibilityMode`:
- `false → true` does not commit immediately; it kicks off
`ConfirmAndEnableModuleAsync`, which awaits the existing
`DangerousFeatureWarningDialog` (resource prefix
`PowerDisplay_EnableModule`) and either commits or reverts the
ToggleSwitch via `OnPropertyChanged`.
- `true → false` commits unconditionally — we never block a user who
wants to turn the module off.
- App-startup loads via `InitializeEnabledValue()` /
`RefreshEnabledState()` assign the `_isEnabled` field directly,
bypassing the setter, so the dialog never fires on settings restore or
GPO refresh.
- GPO-configured state still short-circuits before any dialog logic.
The dialog reuses the existing `DangerousFeatureWarningDialog` injected
by `PowerDisplayPage.xaml.cs`. The 5 new `PowerDisplay_EnableModule_*`
strings explain that the BSOD is in Windows (not Power Display), that
Power Display will auto-disable itself after a detected crash, and that
the user has to re-enable + dismiss the warning each time.
## Validation Steps Performed
- Built `src/settings-ui` and `src/modules/powerdisplay` locally.
- Unit tests: added `EdidIdFromDevicePath_*` cases to
`MonitorIdentityTests`, all green.
- Settings UI manual: toggling Power Display ON now shows the warning
dialog. Pressing Cancel reverts the ToggleSwitch visually; pressing
Enable commits and the module starts. Toggling OFF does not prompt.
Restarting Settings UI with PowerDisplay enabled does not prompt.
GPO-disabled state still locks the toggle.
- Log inspection: `MonitorManager` Phase 0 log now shows `EdidId=...`
for each path before any capability fetch.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Yu Leng (from Dev Box) <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
## Summary of the Pull Request
Two crash-correlation aids for the kernel-side DDC/CI BSOD mitigated by
#47734:
1. Log EDID hardware ID (manufacturer + product code, e.g. `DELD1A8`)
during Phase 0 monitor classification, before any DDC/CI capability
fetch enters the BSOD risk window.
2. Show a confirmation dialog before turning the Power Display module on
from the Settings page, so the user understands the BSOD risk before the
first capability fetch runs.
## PR Checklist
- [ ] Closes: #xxx
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
### 1. Phase 0 EdidId logging
`MonitorIdentity.EdidIdFromDevicePath` parses the EDID hardware ID
segment from a DevicePath of the form
``\?\DISPLAY#DELD1A8#5&abc&0&UID12345#{guid}`` and returns ``DELD1A8``.
The 3-letter PNP manufacturer code + 4-hex product code is identical for
every physical unit of the same model, so it identifies the *model*
without leaking per-unit identifiers.
`MonitorManager` logs the EdidId on the existing Phase 0 classification
line. Phase 0 uses `QueryDisplayConfig`, which reads OS-cached EDID and
cannot BSOD, so this line is guaranteed on disk before the crash-prone
Phase 2 capability fetch starts. If a machine crashes during
enumeration, the recovered log identifies every attached model
(including same-model duplicates), which makes it possible to correlate
crash reports to specific monitor models even when the user can't tell
us which monitor caused the crash.
### 2. Enable-module confirmation dialog
`PowerDisplayViewModel.IsEnabled` setter is refactored to follow the
same two-phase pattern already used by `MaxCompatibilityMode`:
- `false → true` does not commit immediately; it kicks off
`ConfirmAndEnableModuleAsync`, which awaits the existing
`DangerousFeatureWarningDialog` (resource prefix
`PowerDisplay_EnableModule`) and either commits or reverts the
ToggleSwitch via `OnPropertyChanged`.
- `true → false` commits unconditionally — we never block a user who
wants to turn the module off.
- App-startup loads via `InitializeEnabledValue()` /
`RefreshEnabledState()` assign the `_isEnabled` field directly,
bypassing the setter, so the dialog never fires on settings restore or
GPO refresh.
- GPO-configured state still short-circuits before any dialog logic.
The dialog reuses the existing `DangerousFeatureWarningDialog` injected
by `PowerDisplayPage.xaml.cs`. The 5 new `PowerDisplay_EnableModule_*`
strings explain that the BSOD is in Windows (not Power Display), that
Power Display will auto-disable itself after a detected crash, and that
the user has to re-enable + dismiss the warning each time.
## Validation Steps Performed
- Built `src/settings-ui` and `src/modules/powerdisplay` locally.
- Unit tests: added `EdidIdFromDevicePath_*` cases to
`MonitorIdentityTests`, all green.
- Settings UI manual: toggling Power Display ON now shows the warning
dialog. Pressing Cancel reverts the ToggleSwitch visually; pressing
Enable commits and the module starts. Toggling OFF does not prompt.
Restarting Settings UI with PowerDisplay enabled does not prompt.
GPO-disabled state still locks the toggle.
- Log inspection: `MonitorManager` Phase 0 log now shows `EdidId=...`
for each path before any capability fetch.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Yu Leng (from Dev Box) <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
## Summary of the Pull Request
The File Locksmith IPC layer reads and writes raw UTF-16 (WCHAR) bytes
to `last-run.log`, but all three stream opens were using the default
text mode. On Windows, the CRT translates `0x0A` bytes to `0x0D 0x0A` on
write and collapses `0x0D 0x0A` back to `0x0A` on read. Because each
WCHAR is 2 bytes, any code unit whose little-endian byte pair contains
`0x0A` in the low position (e.g. `U+010A`, `U+0A0D`) is silently
corrupted. The fix opens all three streams in binary mode and adds an
explicit open-failure guard.
```cpp
// IPC.cpp — Writer::start()
// Before
m_stream = std::ofstream(path);
// After
m_stream = std::ofstream(path, std::ios::binary);
// + is_open() guard returning E_FAIL on failure
// NativeMethods.cpp — StartAsElevated() writer
// Before
std::ofstream stream(paths_file());
// After
std::ofstream stream(paths_file(), std::ios::binary);
// NativeMethods.cpp — ReadPathsFromFile() reader
// Before
std::ifstream stream(paths_file());
// After
std::ifstream stream(paths_file(), std::ios::binary);
```
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
Three targeted changes across two files:
1. **`FileLocksmithLib/IPC.cpp` — `Writer::start()`**: switched
`std::ofstream` from text to binary mode and added an `is_open()` check
that returns `E_FAIL` immediately when the file cannot be opened
(previously the try/catch did not catch a silent open failure because
`std::ofstream` does not throw by default).
2. **`FileLocksmithLibInterop/NativeMethods.cpp` —
`StartAsElevated()`**: switched `std::ofstream` from text to binary
mode. This is the elevated-restart writer path; without this fix,
Unicode corruption persisted when File Locksmith relaunched as
administrator.
3. **`FileLocksmithLibInterop/NativeMethods.cpp` —
`ReadPathsFromFile()`**: switched `std::ifstream` from text to binary
mode. This is the symmetric reader-side bug — even with both writers
corrected, the CRT text-mode reader could collapse a `0x0D 0x0A` byte
pair (a valid UTF-16 LE code unit, e.g. U+0A0D GURMUKHI EK ONKAR) into a
single byte, desynchronising the 2-bytes-at-a-time read loop and
corrupting all subsequent path data.
No behaviour change for purely ASCII paths. Paths containing Unicode
code points whose little-endian UTF-16 byte pair spans `0x0D 0x0A` were
silently corrupted in all three code paths before this fix.
## Validation Steps Performed
- Code review: no issues flagged.
- Full diff reviewed: all three stream opens (`ofstream` writer in
`IPC.cpp`, `ofstream` writer in `NativeMethods.cpp`, `ifstream` reader
in `NativeMethods.cpp`) now use `std::ios::binary`, making write and
read paths byte-exact and symmetric.
- Mechanically correct: `std::ios::binary` suppresses Windows CRT
`\n`↔`\r\n` translation; the delimiter `L'\n'` (LE bytes `0x0A 0x00`) is
unambiguous in binary mode and is handled correctly by the existing read
loop.
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: MuyuanMS <116717757+MuyuanMS@users.noreply.github.com>
Co-authored-by: Muyuan Li (from Dev Box) <muyuanli@microsoft.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Problem
Since #47119 (`Refresh check-spelling 0.0.26`, merged 2026-04-23)
refreshed the check-spelling tooling and rewrote
`.github/actions/spell-check/expect.txt` (938 lines / 633 deletions),
the check-spelling bot has been leaving a noisy advisory comment on
**every PR**:
> #### These words are not needed and should be removed
> ABlocked AClient AColumn ACR ADate ADifferent AHybrid ALarger
AModifier ANull AOklab APeriod ARandom ARemapped ASingle ASUS bck …
The same ~150-word list is appended verbatim to every PR the bot looks
at (verified against #48058, #48102, #48104 — the list is identical).
These tokens are residual orphans in `expect.txt` from before the 0.0.26
refresh and no longer match anything in source.
## Fix
Removes exactly the 147 orphan tokens that the bot has consistently
flagged as `now absent` from `.github/actions/spell-check/expect.txt`.
The removed tokens are exclusively the ones the bot itself identified.
All uppercase Win32 / DirectWrite identifiers that are still used in
source (`DWRITE`, `LWIN`, `VCENTER`, `VREDRAW`, etc.) are **preserved**.
## Verification
- Diff is a single file, deletions only: `expect.txt` shrinks from 2343
→ 2196 lines.
- Each of the 4 uppercase Win32 tokens (`DWRITE` line 514, `LWIN` 1074,
`VCENTER` 2105, `VREDRAW` 2144 in the original) remains in the file.
- The check-spelling job on this PR should now post a clean report (no
`should be removed` block).
## Background — which PR introduced the drift
| PR | Date | What it changed |
|----|------|-----------------|
| **#47119** | 2026-04-23 | Refreshed check-spelling to 0.0.26; rewrote
`expect.txt` with 938 line-changes (633 deletions, 305 additions). The
duplicated lowercase/uppercase entries and many obsolete tokens
originate here. |
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <Copilot@users.noreply.github.com>
## Summary
Fixes#46767
Items pinned to the End section of the dock did not animate on startup,
while Start and Center items did. The EndListView had an explicit empty
`ItemContainerTransitions` collection that suppressed all container
transitions. Removing it allows the default WinUI entrance animations to
play, matching Start and Center behavior.
## Changes
- `DockControl.xaml`: Remove empty `TransitionCollection` override from
EndListView
---------
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Fixes the **stable** signed release pipeline failure (Dart build
[`147621162`](https://microsoft.visualstudio.com/Dart/_build/results?buildId=147621162))
at the *"Verify all binaries are signed and versioned"* task:
```
Not Signed: + …\extractedMachineMsi\File\BaseApplicationsFiles_File_PowerAccent.Common.dll
```
## Root cause
PR #47211 (*"[Quick Accent] Move language data to PowerAccent.Common
library, refactor"*) introduced the new managed library:
- `src/modules/poweraccent/PowerAccent.Common/PowerAccent.Common.csproj`
`PowerAccent.Common.dll` is referenced by both `PowerAccent.Core` (ships
in the installer root) and `PowerToys.Settings` (ships in WinUI3Apps;
deduplicated against root by `generateAllFileComponents.ps1`). The DLL
is harvested into the MSI's `BaseApplicationsFiles` component group, but
the PR did not update `.pipelines/ESRPSigning_core.json`, so ESRP never
signs it and `versionAndSignCheck.ps1` correctly fails the build.
This is the same kind of omission that PR #48050 fixed for
`YamlDotNet.dll` introduced by #40834 — a new managed library shipping
in the MSI without a matching signing config entry.
## Fix
One additive line in `.pipelines/ESRPSigning_core.json`, placing
`"PowerAccent.Common.dll"` inside the existing alphabetized PowerAccent
block (next to `PowerAccent.Core.dll`).
```diff
"PowerAccent.Core.dll",
+ "PowerAccent.Common.dll",
"PowerToys.PowerAccent.dll",
```
## Validation
- `git diff` shows exactly one additive line.
- File still parses as valid JSON (`ConvertFrom-Json` round-trip OK).
- Audited the full file list of PR #47211: `PowerAccent.Common.dll` is
the only new shippable assembly it added (the other new project,
`PowerAccent.Common.UnitTests`, is a test project and is not included in
the installer).
- Dedup logic in
`installer/PowerToysSetupVNext/generateAllFileComponents.ps1` keeps only
the root copy when WinUI3Apps and root copies are byte-identical, so a
single root-level entry is sufficient.
## Follow-up
After merge to `main`, cherry-pick / merge into `stable` to unblock the
0.100 release pipeline. This is the third (and hopefully final) PR in
the 0.100 release-pipeline unblock sequence, after #48050 (sign
`YamlDotNet.dll`) and #48054 (remove duplicate
`QuickAccent_SelectedLanguage_Greek_Polytonic` resource).
Co-authored-by: Copilot <Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
* Fix wrong Shortcut for Shortcut Guide V2 (previously always showed
default)
* Fix outdated OOBE description
* Fix process not stopping when exiting via ESC or close button
* Fixed some missing modules
* Removed regex capability and was able to improve the start up time
through this
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Okay I'll admit this is a clanker fix.
There's currently a crash when you navigate to an individual extension
page, that only repro's in Release/AOT. And because of
release&trimming&AOT reasons, that's very very difficult to diagnose.
But the clanker added a shitload of logging, and was able to figure out
that:
We were binding to this `Screenshots` collection. Problem is that we
can't safely bind to an `IReadOnlyList` in trimmed scenarios. C#/AOT is
a wonderful world of horrors.
Fixing this is as simple as swapping it for a `ObservableCollection`.
When we were docked on a screen other than the primary display, then
clicking on a dock item that was a `IPage` would result in the palette
window just not showing up at all.
The fix is to make sure that we use the actual position of the current
monitor in the calculation of the final position of the palette HWND.
## Summary
Adds **77 explicit `AutomationProperties.AutomationId`** declarations (4
→ 81, ~20x) to Command Palette XAML so UI-testing tools (WinAppDriver,
FlaUI, winappCli, Appium) can address controls deterministically instead
of falling back to runtime slugs that change every session. Also covers
the `DataTemplate`-generated long tail (search results, file results,
every extension item, every monitor in Dock settings) by binding
`AutomationId` to existing stable model identifiers via `{x:Bind}`.
Headline: a one-line addition to the four CmdPal `ListItem`
`DataTemplate`s now exposes each instance's underlying `Command.Id`
(e.g. `com.microsoft.cmdpal.builtin.calculator`) as its `AutomationId`.
This is the same pattern as web's `<li data-testid={item.id}>`.
## Why
CmdPal today ships with only **4** explicit
`AutomationProperties.AutomationId` declarations across **44** XAML
files (~303 interactive controls). UI tests are forced to fall back to:
- runtime-generated slugs like `itm-12-9dda` that change every CmdPal
restart, or
- regex-parsing the text output of `winappCli ui inspect`, or
- walking the UIA tree client-side and matching by `Name` (localized,
ambiguous when several controls share a label).
## What changed
### Source changes
- **`Microsoft.CmdPal.UI/ExtViews/ListItemsView.xaml`** — bind
`AutomationProperties.AutomationId="{x:Bind Command.Id, Mode=OneWay}"`
on the root of all 4 `ListItem` `DataTemplate`s (single-row, small-grid,
medium-grid, gallery). Covers **every** runtime list item across all
built-in and third-party providers.
- **`Microsoft.CmdPal.UI/Settings/GeneralPage.xaml`** — 14 IDs on
activation hotkey controls, auto-go-home combo, behaviour toggles,
about-section hyperlinks.
- **`Microsoft.CmdPal.UI/Settings/AppearancePage.xaml`** — 19 IDs on
theme/backdrop/colorization, background image controls, 5 sliders,
behaviour toggles.
- **`Microsoft.CmdPal.UI/Settings/DockSettingsPage.xaml`** — 14 IDs on
enable toggle, theme/colorization/backdrop, background image controls,
plus per-monitor data-bound ID using `DisplayName`.
- **`Microsoft.CmdPal.UI/Settings/ExtensionsPage.xaml`** — 4 IDs on
banner/store hyperlinks plus the per-provider `SettingsCard` (data-bound
to the provider's `Id`, e.g. `com.microsoft.cmdpal.builtin.calculator`).
- **`Microsoft.CmdPal.UI/Settings/ExtensionPage.xaml`** — 5 IDs on
provider enable toggle, alias text/activation, fallback rank link.
- **`Microsoft.CmdPal.UI/Settings/ExtensionGalleryItemPage.xaml`** — 5
IDs on install/cancel/copy-command buttons and author/repo/uninstall
links.
- **`Microsoft.CmdPal.UI/Settings/InternalPage.xaml`** — 7 IDs on the
dev-only "Throw exception" and folder/log-opening buttons.
- **`Microsoft.CmdPal.UI/Pages/ShellPage.xaml`** — 1 data-bound ID on
the `CommandTemplate` button (binds to `CommandViewModel.Id`).
- **`Microsoft.CmdPal.UI.ViewModels/ProviderSettingsViewModel.cs`** —
single-line addition exposing the underlying `CommandProviderWrapper.Id`
so the ExtensionsPage template can bind to it.
### Naming convention applied
`CmdPal_<Page>_<Semantic>` PascalCase, with the existing well-known IDs
left untouched to avoid breaking the `modules/cmdpal/*` checklist tests
already referencing them. `Id` / `TestId` suffixes deliberately omitted
(the attribute is already `AutomationId`); control-type suffixes
(`Toggle` / `Button` / `ComboBox`) omitted because the type lives on the
UIA `type` field already.
### Files NOT touched (deferred)
About 5–7 secondary files (`SettingsWindow.xaml`,
`ExtensionGalleryPage.xaml`, `DockControl.xaml`, `DevRibbon.xaml`, deep
style/viewer XAML) — their interactive controls already carry `x:Name`,
which WinUI auto-promotes to `AutomationId` at runtime, so practical
testability is mostly already there. Can be tightened up in a follow-up
PR.
## Risk
- ✅ Zero runtime cost (`AutomationProperties.AutomationId` is metadata
only)
- ✅ Does not affect localization (`AutomationId` is non-localized by
design)
- ✅ Does not affect user-facing accessibility (screen readers consume
`Name` / `AutomationProperties.Name`, untouched)
- ✅ Does not rename any existing ID; `modules/cmdpal/*` tests
referencing `MainSearchBox` / `PrimaryCommandButton` / `ItemsList` /
etc. remain valid
- ✅ All 51 CmdPal XAML files parse as valid XML (PowerShell `[xml]`
round-trip)
- ✅ All `x:Bind` paths verified against the relevant ViewModel types
(`CommandViewModel.Id`, `DockMonitorConfigViewModel.DisplayName`,
`ProviderSettingsViewModel.Id`)
## Validation requested before merge
XAML compilation was attempted locally but blocked by a missing native
NuGet package (`Microsoft.Windows.ImplementationLibrary`) — needs
`tools\build\build-essentials.cmd` to fully restore first. **Reviewer
please run a clean local build of `Microsoft.CmdPal.UI` (or rely on CI)
to confirm the XAML compiler is happy.**
Once built, a quick smoke check:
```powershell
# After CmdPal restart with the new bits installed:
winapp ui search 'CmdPal_' -w <hwnd> --json # should list all the new IDs
winapp ui invoke 'CmdPal_AppearancePage_OpenCommandPalette' -w <hwnd>
winapp ui invoke 'com.microsoft.cmdpal.builtin.calculator' -w <hwnd> # ListItem template binding
```
## Why this matters beyond our tools
This is the same pattern Microsoft's accessibility and WinUI teams
already recommend for any XAML app intended to be automated. It's the
WinUI equivalent of adding `data-testid` in a web app — universally
accepted as low-risk, high-leverage. WinAppDriver, FlaUI, Appium and any
future automation framework all benefit from the same change.
## Summary
Fixes the **stable** signed release pipeline failure (Dart build
[`147590319`](https://microsoft.visualstudio.com/Dart/_build/results?buildId=147590319))
at the *"Verify all binaries are signed and versioned"* task on both
`Build Release_x64` and `Build Release_arm64`:
```
Not Signed: + …\extractedMachineMsi\File\WinUI3ApplicationsFiles_File_YamlDotNet.dll
```
## Root cause
PR #40834 (*Shortcut Guide V2*) added a `YamlDotNet` `PackageReference`
to:
- `src/modules/ShortcutGuide/ShortcutGuide.Ui/ShortcutGuide.Ui.csproj`
-
`src/modules/ShortcutGuide/ShortcutGuide.IndexYmlGenerator/ShortcutGuide.IndexYmlGenerator.csproj`
Both projects publish into `WinUI3Apps\`, so `YamlDotNet.dll` is
harvested into the MSI by
`installer/PowerToysSetupVNext/generateAllFileComponents.ps1`. The PR
updated `ESRPSigning_core.json` for its own new binaries but missed this
3rd-party transitive dependency, so ESRP never signs it and
`.pipelines/versionAndSignCheck.ps1` correctly fails the build.
## Fix
One additive line to `.pipelines/ESRPSigning_core.json`, placing
`WinUI3Apps\YamlDotNet.dll` next to the other 3rd-party `WinUI3Apps\`
entries (`Google.Apis.*`, `Google.GenAI.dll`, `ReverseMarkdown.dll`,
`SharpCompress.dll`, …).
## Validation
- `git diff` shows exactly one additive line.
- File still parses as valid JSON (`ConvertFrom-Json` round-trip OK).
- Both ShortcutGuide csprojs target
`$(Platform)\$(Configuration)\WinUI3Apps`, so a single
`WinUI3Apps\YamlDotNet.dll` entry covers all uses.
## Follow-up
After merge to `main`, cherry-pick / merge into `stable` so build
`147590319` can be re-queued and the 0.100 release pipeline can proceed.
Co-authored-by: Copilot <Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Adds a built-in monitor blacklist that skips known-bad monitor models at
the DDC/CI discovery stage — before PowerDisplay sends any capabilities
request to the firmware. Matches by EdidId (the PnP manufacturer +
product code from EDID). Two entries pre-populated from existing crash
reports:
| EdidId | Reported in | Notes |
|-----------|-------------|------------------------------------------------------------------------|
| `LTM2C02` | #47556 | Counterfeit-EDID LG 27MR400 (PnP claims Litemax
40" 2011, actual hw is LG 27" 2024) |
| `GSM7714` | #47968 | LG UltraWide HDR WFHD |
Also logs `[EdidId=…] [FriendlyName=…] [DevicePath=…]` immediately
before each `GetCapabilitiesString*` syscall in `DdcCiController`. If
the kernel call BSODs (`win32kfull` stack-cookie overrun), that is the
last log line that survives — adding a new blacklist entry then takes
one PR to `BuiltInMonitorBlacklist.json` instead of a memory-dump
triage.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Mitigates: #47556
- [x] Mitigates: #47968
- [x] **Communication:** This is the "Add blacklist and prevent the API
call on broken monitor" mitigation outlined in [#47556
(comment)](https://github.com/microsoft/PowerToys/issues/47556#issuecomment-4505498427).
The underlying `win32kfull!DdcciGetCapabilitiesStringFromMonitor` stack
overrun is tracked separately by the Windows team.
- [x] **Tests:** 6 new MSTest unit tests under
`PowerDisplay.Lib.UnitTests`. All 120 PowerDisplay tests pass on x64
Debug.
- [ ] **Localization:** N/A — this PR ships no end-user-facing strings
(no UI surface).
- [ ] **Dev docs:** N/A.
- [ ] **New binaries:** N/A. Data ships as an `<EmbeddedResource>`
inside the existing `PowerDisplay.Models.dll`; no new .dll, no installer
change.
- [ ] **Documentation updated:** N/A.
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
### Threat model
Some retail monitor firmwares ship non-conformant DDC/CI capabilities
strings — typically oversized, not NUL-terminated, or otherwise
malformed. When PowerDisplay calls `GetCapabilitiesStringLength` /
`CapabilitiesRequestAndCapabilitiesReply` on these monitors, the request
flows into
`win32kfull!CPhysicalMonitorHandle::DdcciGetCapabilitiesStringFromMonitor`,
which copies the reply into a stack buffer without bounding the length.
The kernel detects its own stack cookie corruption on the epilogue check
and `__fastfail`s with `BUGCHECK 0x139 / 2
(STACK_COOKIE_CHECK_FAILURE)`.
The kernel-side overrun is a Windows defect, not PowerDisplay's, but
PowerDisplay is currently the most widely deployed consumer of this API
on hot-plug. Until the Windows team fixes the kernel, the only safe
mitigation in user space is to **never call** the capabilities API on
monitor models known to trigger it.
### Architecture
| File | Role |
|---|---|
| `PowerDisplay.Models/MonitorBlacklistEntry.cs` | POCO: `{ edidId,
comments }` |
| `PowerDisplay.Models/BuiltInMonitorBlacklist.json` | Data file
(embedded resource) |
| `PowerDisplay.Models/BuiltInMonitorBlacklist.cs` | Lazy-loaded reader,
AOT-safe (source-gen `JsonSerializerContext`); silent fallback to empty
list on any IO/parse failure |
| `PowerDisplay.Lib/Services/MonitorBlacklistService.cs` |
`IsBlocked(monitorId)` — extracts EdidId via
`MonitorIdentity.EdidIdFromMonitorId`, checks a `HashSet<string>` with
`OrdinalIgnoreCase` |
| `PowerDisplay/Helpers/MonitorManager.cs` | Filters QueryDisplayConfig
inventory by EdidId **before** any controller (DDC/CI or WMI) is
dispatched; logs each skip with `[MonitorBlacklist] Skipping ...` |
| `PowerDisplay.Lib/Drivers/DDC/DdcCiController.cs` | Logs EdidId +
FriendlyName + DevicePath immediately before the first capabilities
syscall per-monitor |
Matching is **model-level granularity** (EdidId, e.g. `GSM7714`) rather
than device-level (full DevicePath). One entry covers every physical
port and every machine with the same monitor model.
### Adding new entries
Once the kernel overrun is fixed this list can shrink. Until then, when
a new BSOD surfaces:
1. User reports BSOD, attaches PowerDisplay log
2. The last line before the crash now reads `DDC: probing capabilities
[EdidId=XXXXXX] [FriendlyName='...'] [DevicePath=...]`
3. Add `{ "edidId": "XXXXXX", "comments": "See #issue" }` to
`BuiltInMonitorBlacklist.json`
4. Submit PR, ship in next release
### Scoped out (deliberate)
A user-customized blacklist surface (Settings UI dialog) was prototyped
and hit multiple WinUI 3 dialog parse / measure crashes inside a
`ContentDialog`. The custom-list code path was reverted in favor of
"built-in only" for v1 reliability. UI can be re-added later on top of
this foundation if needed — the data model and service already support
it.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
1. **Unit tests:** `dotnet test` on `PowerDisplay.Lib.UnitTests` →
**120/120 pass** (including 6 new tests for loader + service: built-in
JSON loads, EdidIds are normalized to upper-case, no blank entries,
`Lazy<>` cache returns same instance, `IsBlocked` returns `false` for
empty built-in list and for unidentifiable monitor IDs).
2. **Clean x64 Debug build** via `tools\build\build-essentials.cmd` +
per-project `tools\build\build.cmd`:
- `PowerDisplay.Models` — 0 warning, 0 error
- `PowerDisplay.Lib` — 0 warning, 0 error
- `PowerDisplay` (app) — 0 warning, 0 error
3. **Smoke test (manual):** launched PowerDisplay, confirmed every
connected monitor produces a `DDC: probing capabilities [EdidId=…]
[FriendlyName=…] [DevicePath=…]` log line right before its caps request.
4. **Empty-list case:** `BuiltInMonitorBlacklist.json` with no entries →
`IsBlocked` returns `false` for any input (verified via unit test).
5. **Non-empty case (manual):** temporarily added a connected monitor's
EdidId to the built-in JSON → that monitor disappears from
`MonitorManager.Monitors` and `[MonitorBlacklist] Skipping ...` log line
appears; removed the entry → monitor reappears, all per-monitor settings
intact.
---------
Co-authored-by: Yu Leng (from Dev Box) <yuleng@microsoft.com>
## Summary
Fixes the **"Build PowerToys main project"** failure on every PR/CI
build off current `main` (e.g. Dart build
[`147597426`](https://microsoft.visualstudio.com/Dart/_build/results?buildId=147597426)):
```
WINAPPSDKGENERATEPROJECTPRIFILE: Error PRI175: Processing Resources failed with error: Duplicate Entry.
WINAPPSDKGENERATEPROJECTPRIFILE: Error PRI277: Conflicting values for resource
'Resources/QuickAccent_SelectedLanguage_Greek_Polytonic'
```
## Root cause
`src/settings-ui/Settings.UI/Strings/en-us/Resources.resw` contained two
`<data name="QuickAccent_SelectedLanguage_Greek_Polytonic">` entries:
- Line 3597 — original, added by #47021 (*"[PowerAccent] adding greek
polytonic"*), placed in the alphabetized QuickAccent block.
- Line 6175 — duplicate, appended at the file tail by #47211 (*"[Quick
Accent] Move language data to PowerAccent.Common library, refactor"*).
The two entries are byte-identical (same value `Greek Polytonic`, same
`xml:space="preserve"`). The WinAppSDK PRI generator (`MakePri`) rejects
duplicates, so every build off current `main` fails before reaching the
compile step.
## Fix
Remove the tail duplicate (3-line block right before `</root>`). Keeps
the original entry in its alphabetized location.
## Validation
- `git diff` shows exactly 3 deleted lines.
- `[xml](Get-Content … -Raw)` round-trip succeeds — file is still
well-formed XML.
- `Select-String` count of
`QuickAccent_SelectedLanguage_Greek_Polytonic` in the file is now **1**
(was 2).
- Only `en-us` has this key (other localized `.resw` files are populated
later via Touchdown), so no other file needs touching.
## Note
Discovered while investigating a CI failure unrelated to my other PR
#48050 (signing of `WinUI3Apps\YamlDotNet.dll`). Both fixes are needed
for the 0.100 release pipeline; this PR unblocks PR builds off `main`,
and #48050 unblocks the signed release pipeline once both are merged to
`main` and the next `main → stable` sync happens.
Co-authored-by: Copilot <Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Language data for Quick Accent was previously defined in
**PowerAccent.Core/Languages.cs**, internal to the Quick Accent
application itself. The Settings application had no access to the list
and had to maintain a parallel, manually-synchronised list of the
language names and groups, and there was no single place to look up the
complete set of languages.
This PR resolves this and extracts the language data into a new
**PowerAccent.Common** project with no external or UI dependencies,
making it the single source of truth for both the character popup and
Settings UI.
The implicit and non-obvious ordering of the old **Languages.cs** has
been replaced with explicit ordering of the language groups and
languages list.
A new unit test project for the application has also been added.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #47159
- [x] Closes: #30000
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
The following changes and additions have been made:
New project **PowerAccent.Common** contains:
- **Language.cs** - an enum of Quick Accent's pseudo-ISO identifiers for
each of the supported languages
- **LetterKey.cs** - mirrors the LetterKey enum defined in the
KeyboardListener.idl, and must be kept in sync with that. This exposes
the VK_* data to the Settings UI without it having to reference the
PowerAccentKeyboardService project
- **LanguageGroup.cs** - classifies languages as `Language` (for spoken
languages), `Special` (for sets like Currency and International Phonetic
Alphabet), or `UserDefined` (reserved for future work)
- **LanguageInfo.cs** - a record which combines a language's identity,
group, resource identifier, and character mappings
- **CharacterMappings.cs** - a new version of the old **Languages.cs**,
providing:
- `All` - the canonical registry of every `LanguageInfo` entry
- `DisplayOrder` - the explicit within-group ordering for languages, for
the default popup ordering, i.e. when the user has not got
frequency-based ordering enabled. This is a culturally-neutral
alphabetical ordering by our pseudo-ISO language IDs, and close to the
previous order, minimising any impact to users' muscle memory
- `GroupDisplayOrder` - the explicit ordering of language groups for
both the popup and the Settings UI
- `LanguageLookup` - new _O(1)_ `Language` to `LanguageInfo` dictionary
- `GetCharacters(LetterKey, Language[])` - a deduplicating character
collector and sorter
A significant improvement over the old **Languages.cs** is that language
ordering is now explicit and in one place. Previously, popup ordering
was an implicit side-effect of a large Union chain across all language
mappings, with the enum, the language mapping declarations and the union
all carrying different orderings. This was a source of confusion and
made adding new languages fragile.
The original **Languages.cs** has been deleted. There's now a thin
`CharacterMappings` adapter that casts the `LetterKey` to the managed
equivalent by numeric value. This is another effort to decouple the
Settings UI and Quick Accent and only share the required elements.
In Settings, the `PowerAccentViewModel` class has been updated. It now
derives the language list directly from `CharacterMappings.All`; the
`InitializeLanguages()` call handles localisation, sorting and grouping
in a single place using `GroupDisplayOrder`.
A new unit test project covers `CharacterMappings`, the `LetterKey`
IDL-to-managed code bridge and general language declaration checks. The
application is now much more robust against changes which alter the
declared order, introduce empty/null elements, and so on. It is now
impossible to add a new language and forget a constituent part (the enum
entry, its place in the display order, the character mappings
themselves) without a test failing.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
See new unit test project for comprehensive tests.
Manually confirmed:
- Triggering Quick Accent with no language selected does not report an
error
- Language order is consistent in the popup, no matter what order the
languages are selected in Settings
- The order of languages in Settings is consistent, alphabetically by
localised name
- All declared languages are present in Settings and no resource IDs
were missed
---------
Co-authored-by: Muyuan Li (from Dev Box) <muyuanli@microsoft.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Addresses three critical issues with the PowerToys update experience
that cause user fragmentation across old versions.
### Changes
**1. Fix relaunch after update (Fixes#42004, #43011, #44071)**
- Stage 1 now passes the PowerToys install directory to Stage 2 as an
argument
- After successful install, Stage 2 relaunches `PowerToys.exe` with
`-report_update_success`
- Users will see a 'successfully updated' toast and PT resumes
automatically
**2. Config backup/restore (Fixes#46179)**
- `BackupConfigFiles()` snapshots all JSON configs to `ConfigBackup/`
before update begins
- `RestoreCorruptedConfigs()` checks for null-byte corruption after
install and auto-restores
- Protects Workspaces, FancyZones, Keyboard Manager, and all other
module settings
**3. Enable auto-download by default**
- New installations default `AutoDownloadUpdates` to `true` (was
`false`)
- Existing users' preferences are preserved — this only affects
first-run defaults
- The runner already defaulted to `true`; this aligns the C# settings
model
### Why this matters
The current updater kills all PowerToys processes, runs the installer,
then **exits without relaunching**. Users lose keyboard remappings,
FancyZones layouts, and Awake settings with no indication why. Combined
with auto-download being off by default, most users are multiple
versions behind.
### Testing
- Verified update flow: Stage 1 → Stage 2 → PT relaunches with success
toast
- Config backup creates mirror of all JSON settings before update
- Corruption detection catches null-byte pattern from #46179
- Graceful fallback: if install dir not provided (old Stage 1), logs
warning but doesn't crash
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Muyuan Li (from Dev Box) <muyuanli@microsoft.com>
Co-authored-by: Niels Laute <niels.laute@live.nl>
## Summary of the Pull Request
Shortcut Guide was rendering raw numeric key codes in key visuals for
some shortcuts (notably generated PowerToys shortcuts), showing numbers
where key characters/names were expected. This change normalizes numeric
key handling so display output matches actual key labels.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
- **Scope**
- Updated `ShortcutDescriptionToKeysConverter` in `ShortcutGuide.Ui`
only.
- **Behavior change**
- Numeric key codes are no longer passed through as raw integers by
default.
- Arrow keys (`37/38/39/40`) remain numeric to preserve existing glyph
rendering behavior.
- All other numeric key codes are converted to display names via
`Helper.GetKeyName(...)`.
- **Result**
- Shortcut Guide key caps now show expected characters/key names instead
of numeric values for manifest-provided numeric keys.
```csharp
if (int.TryParse(key, out int keyCode))
{
switch (keyCode)
{
case 38:
case 40:
case 37:
case 39:
shortcutList.Add(keyCode); // keep glyph path
break;
default:
shortcutList.Add(Helper.GetKeyName((uint)keyCode)); // show key label
break;
}
}
```
## Validation Steps Performed
No additional validation details are included in this description.
> [!WARNING]
>
> <details>
> <summary>Firewall rules blocked me from connecting to one or more
addresses (expand for details)</summary>
>
> #### I tried to connect to the following addresses, but was blocked by
firewall rules:
>
> - `o3svsblobprodcus318.vsblob.vsassets.io`
> - Triggering command: `/usr/bin/dotnet dotnet build
/home/REDACTED/work/PowerToys/PowerToys/src/modules/ShortcutGuide/ShortcutGuide.Ui/ShortcutGuide.Ui.csproj
-v minimal` (dns block)
>
> If you need me to access, download, or install something from one of
these locations, you can either:
>
> - Configure [Actions setup
steps](https://gh.io/copilot/actions-setup-steps) to set up my
environment, which run before the firewall is enabled
> - Add the appropriate URLs or hosts to the custom allowlist in this
repository's [Copilot coding agent
settings](https://github.com/microsoft/PowerToys/settings/copilot/coding_agent)
(admins only)
>
> </details>
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: niels9001 <9866362+niels9001@users.noreply.github.com>
Co-authored-by: Noraa Junker <noraa.junker@outlook.com>
This PR:
- Adds a MaxWidth to the SCOOBE and OOBE pages, inline with Fluent
design guidance
- Updates the imagery for PowerToys
- Tweaks some small nits in the General settings page: re-ordering,
adding icons, replacing a ToggleSwitch with a CheckBox
<img width="1774" height="1325" alt="image"
src="https://github.com/user-attachments/assets/d825047a-27be-4e8b-a6ce-7f1a71f39dfe"
/>
The PowerDisplay flyout had no keyboard close path: pressing Tab kept
focus inside the window so it never deactivated, and there was no
KeyDown / KeyboardAccelerator wired up. Handle Escape on RootGrid to
hide the window, matching the behavior of other PowerToys flyouts.
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #48016
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
Co-authored-by: Yu Leng (from Dev Box) <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
## Summary
Corrects four label names in the \VALID_LABELS\ allow-list of the
\uto-label-issues.yml\ workflow to match existing repository labels
exactly.
## Root Cause
The GitHub Actions \ddLabels\ API creates a brand-new label when the
name doesn't exactly match an existing one. The hardcoded list had
typos/mismatches:
| Workflow had | Repo actually has |
|---|---|
| \Product-Power Display\ | \Product-PowerDisplay\ |
| \Product-ColorPicker\ | \Product-Color Picker\ |
| \Product-Command Not Found\ | \Product-CommandNotFound\ |
| \Product-Hosts\ | \Product-Hosts File Editor\ |
## Changes
- Fixed all 4 label strings in \VALID_LABELS\ array
- Deleted the spurious \Product-Power Display\ label that was created by
the mismatch
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Carry per-monitor user preferences from the pre-#47712 Id format onto
the current DevicePath-based Ids by matching on EdidId. Without this,
every upgrade silently resets Enable* toggles (input source, color
temperature, power state) for monitors users had already opted in on,
because the direct-Id lookup in ApplyPreservedUserSettings can never
match the old "DDC_DELD1A8_1" / "WMI_BOE0900_2" keys.
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Yu Leng (from Dev Box) <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Currently, if the user clicks on a dock item which is a Page command,
then we'll attempt to show the cmdpal where the user clicked on a dock
item. This works great. However, if the user right-clicks a dock item to
invoke the context menu, then from that menu clicks on a Page command,
then nothing happens - we don't show the cmdpal window.
Similarly - IInvokableCommand's that return a CommandResult.Confirm - we
never show the cmdpal window so that the confirmation ContentDialog is
shown to the user.
Yes in the long run, these will make more sense to have a better UI for
the "dock flyout" (see #45861). But until then, this fixes these
scenarios.
Closes#45963
very relevant for #47989
## Summary of the Pull Request
This PR introduces real-time settings synchronisation for Image Resizer.
External changes to `settings.json` (via the Settings application or
manual edits to the file) are detected and reloaded immediately without
requiring a restart.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #36943
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
### Reload updates
The reload detection is via an `IFileSystemWatcher` (as file system
operations are abstracted in Image Resizer), which monitors the
`settings.json` file for creation and changes. There is a half second
debounce for changes, as the settings file can rapidly change when the
user is editing the preset name field.
Hot reloading of the properties required refactoring `ReloadCore`, which
replaces the `Sizes` preset collection and updates the Custom and AI
presets, in addition to the application-wide properties like compression
options and the fallback encoder choice.
I changed the combo box data binding from the `SelectedSize` object to
the int `SelectedSizeIndex`. This was required to resolve a specific WPF
issue where reloading the `Sizes` collection would cause issues with
restoring the current combo box selection to the prior value. Binding to
the index decouples the selection state from the object lifecycle during
the reload process.
Selection preservation is based on the preset ID (for user presets) and
preset type (for Custom and AI presets). This ensures that matches are
robust even if the user is in the process of renaming an entry in the
Settings application. If the preset cannot be matched (for example, if
the user deletes the item or changes the ID manually in the file), the
Custom preset is selected.
Selection range checks are maintained from the old code, and additional
checks have been added to ensure that Custom and AI presets will be
present even if they're deleted from the settings file.
The AI preset check from before has been inlined; this guarantees that
the AI resize option will not display if the facility is unavailable on
the current PC, even if it's present in the settings file.
The reload routine is dispatched to the UI thread, as changes involve
updates to the combo box.
### ID recovery
Preset IDs are key to preserving the combo box selection between
reloads, and a couple of changes were necessary to ensure round-tripping
changes were robust.
1. IDs are not reused. The old code could reuse IDs under certain
circumstances, for example if a preset was deleted and re-added via
Settings.
2. The ID Recovery Helper routine sorted the supplied presets by ID
before performing duplicate conflict resolution. This is not required
and it is more natural to assume that the order in the settings file and
the client UI is the source of truth.
New ID assignment is now based on a monotonically increasing ID (seeded
at application start) rather than `Current Maximum ID + 1`. This means
that IDs cannot be reused in the same Settings application session. This
makes the matching process in the client application more reliable.
A small update was made to the ID Recovery Helper to use `HashSet.Add()`
instead of splitting up the check and addition steps (`Add` will return
false if the value already exists), which saves a massive one line of
code.
There were comments about the ID Recovery Helper resolving "empty" or
"invalid" entries; this was inaccurate, as IDs are ints, which must by
definition always have a valid value. The routine only guards against
duplicates, so the comments have been updated to reflect this.
### Miscellaneous
The `Default` Settings property getter previously called `Reload` every
time it was accessed. This is fine when the file is only read at
application startup, but I changed this to lazily instantiate and call
`Reload` a single time.
I refactored duplicate code related to settings file/folder strings, and
also the creation of the Custom and AI size instances.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
Manual testing:
1. Add new preset in the Settings application. The new entry is
reflected in the client application as it is running.
2. Delete new preset. The entry is removed in the client application
list.
3. Edit an existing preset. The property change is reflected in the
client application.
4. Add new preset in the Settings application. Select the new preset in
the client application. Edit the properties of the new preset in the
Settings application and confirm that the updates appear in the client
application.
5. Add new preset in the Settings application. Select the new preset in
the client application. Delete the new preset in Settings. Confirm that
the current preset is removed and the selection changes to the default
in the client application.
6. Change one or more application-wide properties in the settings file
which are represented in the client application, too, e.g. "Make
pictures smaller but not larger" (`imageresizer_shrinkOnly`) or
"Overwrite files" (`imageresizer_replace`). Upon saving, confirm the
checkbox changes immediately in the client application.
7. Edit the `settings.json` file manually by e.g. adding a new Size
preset or editing the width or height property of an existing preset.
Upon saving, the change(s) should be reflected in the client
application.
8. Make a change to an existing or new preset which affects the resize
operation itself, e.g. the dimensions or the "Make pictures smaller but
not larger" setting. Proceed with the resize operation and confirm that
the new changes have been applied.
14 new unit tests have been added to `SetingsTests.cs` to exercise the
`Settings` and `IDRecoveryHelper` functionality.
---------
Co-authored-by: Yu Leng (from Dev Box) <yuleng@microsoft.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
## Summary
Adds the four Google.* DLLs shipped by AdvancedPaste (via
`Microsoft.SemanticKernel.Connectors.Google` for Gemini support) to the
ESRP sign list.
## Problem
The `Verify all binaries are signed and versioned` pipeline step
(`.pipelines/versionAndSignCheck.ps1`) was failing with:
```
Not Signed: + ...\extractedMachineMsi\File\WinUI3ApplicationsFiles_File_Google.Apis.Auth.dll
Not Signed: + ...\extractedMachineMsi\File\WinUI3ApplicationsFiles_File_Google.Apis.Core.dll
Not Signed: + ...\extractedMachineMsi\File\WinUI3ApplicationsFiles_File_Google.Apis.dll
Not Signed: + ...\extractedMachineMsi\File\WinUI3ApplicationsFiles_File_Google.GenAI.dll
```
These DLLs are transitive NuGet dependencies of
`Microsoft.SemanticKernel.Connectors.Google` (referenced by
`src/modules/AdvancedPaste/AdvancedPaste/AdvancedPaste.csproj`) and
deploy to `WinUI3Apps\`, but were never added to
`.pipelines/ESRPSigning_core.json`.
## Fix
Added the four DLLs to the second `SignBatch` (`CP-231522`, the
third-party / OSS-library identity) — same batch used for other
connector DLLs like `OpenAI.dll`, `OllamaSharp.dll` and
`Microsoft.SemanticKernel.Connectors.Ollama.dll`.
## Validation
- `ConvertFrom-Json` on the modified file confirms valid JSON.
- Sign-list entries match the actual deploy paths flagged by
`versionAndSignCheck.ps1`.
- The next ESRP signing run will pick them up; the verify step should
pass.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary
Mitigation for issue #47556 — `KERNEL_SECURITY_CHECK_FAILURE` BSOD
originating in `win32kfull!DdcciGetCapabilitiesStringFromMonitor` when
PowerDisplay calls DDC/CI capability APIs against monitors with
malformed capability strings.
After a detected crash, PowerDisplay auto-disables itself via
`settings.json`, shows an error InfoBar at the top of the PowerDisplay
settings page (page is locked except the Ignore button), so users can
avoid getting stuck in an infinite reboot loop after a crash. And the
user must explicitly dismiss the warning before re-enabling the module.
The actual kernel-side fix is the Windows team's responsibility — this
PR only prevents users from BSOD-ing repeatedly on the same monitor
without warning.
settings page:
<img width="1743" height="1475" alt="image"
src="https://github.com/user-attachments/assets/8cf1b72f-c51a-4955-82d7-213cae49fd4e"
/>
## Mechanism
1. `CrashDetectionScope` IDisposable wraps Phase 2 capability fetch in
`DdcCiController.DiscoverMonitorsAsync`, writing `discovery.lock`
(`WriteThrough` + `Flush(flushToDisk: true)`) before, deleting it on
Dispose.
2. If the process is killed externally (BSOD, FailFast), the lock
survives.
3. On next PowerDisplay.exe startup (Phase 0), `CrashRecovery` detects
the orphan lock and runs a strict fail-fast sequence: write
`crash_detected.flag` → set `enabled.PowerDisplay=false` in global
`settings.json` → signal the new `POWER_DISPLAY_AUTO_DISABLE_EVENT` →
delete the lock (commit point).
4. The runner-loaded `PowerDisplayModuleInterface.dll` runs a one-shot
listener thread that wakes on the event and calls `disable()` to sync
`m_enabled`.
5. `PowerDisplayViewModel` reads the flag at construction and binds
`IsCrashLockActive` to lock the page.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #47556
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This allows users to drag files & URLS to the dock, to immediately
create a bookmark, and pin that bookmark to the dock.
It also updates the bookmark provider's dock bands. If you bookmark a
folder, then pin that to the dock, when you click it on the dock, we'll
default to the "browse" experience, which will open the list of files in
cmdpal, rather than open in explorer.
Had to make miscellaneous changes to make this all a bit faster:
* DirectoryPage didn't load icons smartly, like we do for bookmarks
* I added a different "observable" collection for dock top-level items,
because it caused a flippin CollectionChanged cascade any time a single
provider had more than one item pinned in the dock. Each item from each
provider would cause us to recreate all the dock view models (???) crazy
* I alas had to make `IBookmarksManager` public, so that the UI could
use it. I hate it, but I couldn't figure out a better way. Bookmarks are
a pretty built-in part, so it's _fine i guess_
re: #45584
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
This PR checks if future PRs has added or refines telemetry events, if
so, the bot will add a message to the PR about the needed steps
depending on the PR.
**NOTE**: This PR is submitting a test version, which is only manually
triggered, once tested and confirmed then will move to it being
automatic
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
## Summary of the Pull Request
This PR fixes several issues around the popup selection window's size
and position, selection-related issues which result in flashing or
glitching, and includes more reliable detection of the Shift key.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #44332
- [x] Closes: #44980
- [x] Closes: #35094
- [x] Closes: #40498
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
This PR includes fixes for the Quick Accent's selection window position,
its width measurement, and letter selection-related issues. In addition,
glitches such as the window flashing the selection colour and the window
appearing blank should be reduced or eliminated entirely.
### Popup width bug
When opening Quick Accent from a letter with many mappings, it would
appear too wide for the display. Even though letters could be selected,
they may be entirely off-screen:
<img width="1578" height="134" alt="image"
src="https://github.com/user-attachments/assets/cfcb2ddb-3cf3-47d5-9386-133a2fc70550"
/>
This was because of this flaw in `GetDisplayMaxWidth`, which is used
directly by the popup to set the maximum width of the characters area:
```csharp
// In Selector.xaml.cs
private void SetWindowsSize()
{
this.characters.MaxWidth = _powerAccent.GetDisplayMaxWidth();
}
...
// In PowerAccent.cs
public double GetDisplayMaxWidth()
{
return WindowsFunctions.GetActiveDisplay().Size.Width - ScreenMinPadding;
}
```
`GetActiveDisplay` uses the `GetMonitorInfo` API, which exposes the
working area of the display. It returns its values in _raw unscaled
pixel_ values:
```csharp
public static (Point Location, Size Size, double Dpi) GetActiveDisplay()
{
...
var res = PInvoke.MonitorFromWindow(guiInfo.hwndActive, MONITOR_FROM_FLAGS.MONITOR_DEFAULTTONEAREST);
MONITORINFO monitorInfo = default;
monitorInfo.cbSize = (uint)Marshal.SizeOf(monitorInfo);
PInvoke.GetMonitorInfo(res, ref monitorInfo);
...
return (location, monitorInfo.rcWork.Size, dpi);
}
```
However, the `MaxWidth` property must be a _pre-scaled_ value, i.e. in
logical WPF units not physical pixels. The fix is straightforward:
```csharp
public double GetDisplayMaxWidth()
{
var activeDisplay = WindowsFunctions.GetActiveDisplay();
return (activeDisplay.Size.Width / activeDisplay.Dpi) - ScreenMinPadding;
}
```
### Popup positioning bug
This is related to a subtle DPI issue in `GetActiveDisplay()`:
```csharp
public static (Point Location, Size Size, double Dpi) GetActiveDisplay()
{
GUITHREADINFO guiInfo = default;
guiInfo.cbSize = (uint)Marshal.SizeOf(guiInfo);
PInvoke.GetGUIThreadInfo(0, ref guiInfo);
var res = PInvoke.MonitorFromWindow(guiInfo.hwndActive, MONITOR_FROM_FLAGS.MONITOR_DEFAULTTONEAREST);
MONITORINFO monitorInfo = default;
monitorInfo.cbSize = (uint)Marshal.SizeOf(monitorInfo);
PInvoke.GetMonitorInfo(res, ref monitorInfo);
double dpi = PInvoke.GetDpiForWindow(guiInfo.hwndActive) / 96d;
var location = new Point(monitorInfo.rcWork.left, monitorInfo.rcWork.top);
return (location, monitorInfo.rcWork.Size, dpi);
}
```
Here, the application window's DPI is returned. Unfortunately, the
window may report a value which is different from the monitor's own DPI
value. This will consistently happen if the application is not
Per-Monitor DPI-Aware, and the monitor is not at 100% Scale. The effects
are that the Quick Accent popup can appear misaligned or even off-screen
entirely. Quick Accent can still be used, but the user may not be able
to see what they are selecting.
As Quick Accent is using monitor coordinates for setting its location,
the solution is to use the monitor's own DPI value. The fix is to add
this in place of the `GetDpiForWindow` line:
```csharp
uint dpiRaw = 96; // Safe default
if (PInvoke.GetDpiForMonitor(res, MONITOR_DPI_TYPE.MDT_EFFECTIVE_DPI, out uint dpiX, out _) == 0)
{
dpiRaw = dpiX;
}
double dpi = dpiRaw / 96d;
```
### Selection bugs
After dismissing the Quick Accent window, the `_selectedIndex` state was
not properly reset. The next time the window opened, it could attempt to
scroll to or highlight an index that was out-of-bounds for the new
character set. This could result in glitching, such as the window
flashing the selection colour or the initial selection being incorrect.
In this fix, I:
1. Explicitly set `_selectedIndex` to `-1` when the UI hides.
2. Reset the `SelectedIndex` inside Selector.xaml.cs before updating the
`ItemsSource`.
### Shift key activation
In certain cases, a quick press of Shift could fail to move back through
the character list. In this fix, I:
1. Added a native fallback usign GetAsyncKeyState(VK_SHIFT).
2. Updated `ProcessNextChar()` to evaluate `shiftPressed ||
WindowsFunctions.IsShiftState()`.
3. Updated the multiple `if`s in `ProcessNextChar` to be an if/else
structure, to prevent bugs when more than one trigger key is pressed.
### Support added for multi-codepoint graphemes
The current code loops through each `char` of a mapping, calling
`SendInput` multiple times for multi-char sequences. This will fail for
multi-codepoint graphemes, i.e. where the mapping 'letter' is more than
one UTF-16 codepoint. Those characters may appear as `[]`. The amended
`Insert()` in `WindowsFunctions` appends all characters before calling
`SendInput`.
### Miscelleneous
- Added an `OnDpiChanged` handler for the Selector control, so changing
the DPI of the screen should be picked up automatically. (It's
questionable whether this is essential, as the DPI would have to change
while the control was displayed, but it's worth having for robustness.)
- Now using `SetWindowPos` instead of setting the `Left` and `Top` of
the popup control. Also now initialising the popup offscreen to attempt
to reduce flicker and the occurrence of blank window flashes.
- Changed the `Focusable` property of the characters `ListBox` to
`False`, to attempt to reduce flicker and the window flashing the
selector colour.
- Removed `Width` and added `MinWidth` to the letter control in
Selector.xaml. This allows for wider letters or longer multi-letter
mappings.
- Changed the `VirtualizingStackPanel` to a regular `StackPanel`. We do
not have mappings with enough entries for a virtual control to be
necessary, and using StackPanel seemed to have a positive effect on the
appearance of blank window glitches.
- Added `TextTrimming`, `TextWrapping` and `MaxHeight` to the unicode
description `TextBlock`. This helps support extremely long unicode
descriptions. Again, this will enable us to support longer
multi-character mappings in the future.
- Added CsWin32 to the PowerAccent.UI project, to support the
`SetWindowPos` call.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
See separate doc for full test details:
https://docs.google.com/document/d/19uClcUiv7RUDRlbFhazG-Cmu46oNmrAVoJf9bHSjSJU/edit?usp=sharing
---------
Co-authored-by: Muyuan Li (from Dev Box) <muyuanli@microsoft.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Niels Laute <niels.laute@live.nl>
Adds Greek Polytonic characters set to power accent, based on
https://github.com/microsoft/PowerToys/pull/29709
### PR Checklist
- [x] Closes#46941
- [x] Communication: I've discussed this with core contributors already.
- [ ] Tests: Not sure if there are specific tests for this
- [ ] Documentation updated: Power accent docs
### Detailed Description of the Pull Request / Additional comments
Added all greek polytonic letters to their corresponding english letter
(some duplicated)
(if you wondered about GRC -> ISO 639-3)
### Validation Steps Performed
Compiled and Observed Power Accent
---------
Co-authored-by: Niels Laute <niels.laute@live.nl>
Co-authored-by: Dave Rayment <dave.rayment@gmail.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Renames the issue-triage GitHub Action to **Automatic Triaging on Issue
Creation** and removes the redundant `auto-label-product.yml` workflow.
This consolidates issue labeling/triage under a single workflow surface.
Also syncs this PR branch with the latest `main` via a merge commit to
keep it up to date with upstream.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
- **Workflow rename**
- Updated `.github/workflows/auto-label-issues.yml`:
- `name: Auto-label Issues by Area` → `name: Automatic Triaging on Issue
Creation`
- Updated the manual-trigger comment to reference the new action name.
- **Workflow cleanup**
- Removed `.github/workflows/auto-label-product.yml` to eliminate
overlapping automation.
- **Branch sync requested in PR comments**
- Merged latest `origin/main` into this branch (`4f831bc`) to keep the
PR current.
```yaml
# .github/workflows/auto-label-issues.yml
name: Automatic Triaging on Issue Creation
```
## Validation Steps Performed
- Verified clean merge of `origin/main` into this PR branch (no merge
conflicts).
- Confirmed targeted workflow changes remain present after merge.
- Ran PR validation tooling:
- Code Review completed successfully.
- CodeQL scan timed out in validation tooling.
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: niels9001 <9866362+niels9001@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
This PR reverts #47900: it causes an exception in the Release+AOT build
configuration when retrieving a storyboard from XAML resources and
casting it to a local field.
- [x] Closes: #47970
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
This PR updates the shell command provider to work (almost) exactly like
run. The current shell provider is close, but not technically correct.
It does enumerate files. Sure. But as it turns out, it doesn't enumerate
things **exactly** correctly. It doesn't handle network paths super
well. It doesn't handle NTFS file paths. Basically, there's a lot of
weird edge cases in the way the run dialog enumerates file paths for
suggestions. And the only way to match that is to just use the code from
the old run dialog.
This is code that is taken pretty verbatim from the new run dialog.
Instead of trying to enumerate paths manually and shellexecuting command
lines, We're using the actual APIs that the original run dialog used,
more or less. They've been pretty much ported to C#.
This should make us feel just as correct as the original run dialog did.
And exactly the same as the new Run dialog.
The one major change is the introduction of a static item at the top of
the list for running the command that the user typed. This command is
used to just immediately take whatever is in the search box and fire it
off as the command the user typed. This is essentially what happens with
the run dialog. When you press the button, we run the command in the
text box.
See: [The new Run dialog: faster, cleaner, and more capable - Windows
Command
Line](https://devblogs.microsoft.com/commandline/the-new-run-dialog-faster-cleaner-and-more-capable/)
Honestly, most of this PR is just deleting the files we no longer need
from the shell list provider and adding the tests from the OS side here.
I also had to update CsWinRT for this.
## Summary
Adds a Battery widget to the CmdPal Performance Monitor dock, exposing
live charge percentage, charging/AC status, and estimated time
remaining. The dock-band icon updates each tick to reflect the current
charge level and charging state, matching the existing
CPU/RAM/GPU/Network widget pattern.
Closes#47218
## Detail
- New `SystemBatteryUsageWidgetPage` in `PerformanceWidgetsPage.cs`,
mirroring the existing widget pattern.
- Data source: `GetSystemPowerStatus` via CsWin32 (AOT-safe), wrapped in
`BatteryStats` and surfaced through `SystemData` / `DataManager` on the
shared 1 s timer.
- Live dock-band icon: `Icons.cs` caches 23 `IconInfo` instances
(`BatteryIcons[0..10]`, `BatteryChargingIcons[0..10]`,
`BatteryUnknownIcon`); alloc-free `BatteryGlyph()` selects the glyph per
tick. Codepoints `0xEBA0`–`0xEBB5` / `0xEC02` (Segoe Fluent Icons
MobBattery family).
- Adaptive card template `SystemBatteryTemplate.json` registered as
`<None Update … PreserveNewest>`, consistent with the other PerfMon
templates.
- New `Battery_*` strings under `Strings\en-US\Resources.resw`.
- Handles `GetSystemPowerStatus` sentinels: `BatteryLifePercent == 255`,
`BatteryLifeTime == 0xFFFFFFFF`, `BatteryFlag == 0xFF`, `0x80` (no
battery), `0x08` (charging).
## Screenshots
<img width="1660" height="401" alt="image"
src="https://github.com/user-attachments/assets/6e26f10a-267b-4f6b-b43a-b2c63c092a6d"
/>
When the battery is at 100%, the widget will always show the "full" icon
(`MobBatteryCharging10`).
<img width="537" height="37" alt="image"
src="https://github.com/user-attachments/assets/d925a328-b9b6-4760-a488-f936910a1715"
/>
Otherwise and when on AC, it will show the "charging" variant of the
icon:
<img width="85" height="61" alt="image"
src="https://github.com/user-attachments/assets/065e3a53-c210-4d65-8ca2-02dccf63edf6"
/>
The icon also always reflects the current charging state: It resolves to
the correct glyph in both charging and discharging states, rounded to
the nearest 10% (so 0%, 10%, … 100%).
## How tested
- `MSBuild CommandPalette.slnf /p:Configuration=Release /p:Platform=x64
/m /restore` — green, 0 errors, 0 warnings.
- Launched dev `Microsoft.CmdPal.UI.exe`, opened PerfMon dock, confirmed
Battery widget renders charge %, status, time remaining, and that the
dock-band glyph matches the current charge level (verified on AC, on
battery, and while charging).
## Summary
Fixes#38317 — "Too many tags on a list item looks hilarious"
When CmdPal list items have many tags (e.g., GitHub issue labels), the
tag pills dominated the row and pushed the title text out of view.
### Changes
**Two-pronged fix:**
1. **Cap visible tags at 3** — ViewModel exposes `VisibleTags` (first 3)
and `OverflowTagText` (`+N`) properties. Original `Tags`/`HasTags`
bindings are preserved (no contract breaks).
2. **Overflow badge** — A `[+N]` badge appears when more than 3 tags
exist, reusing existing `Tag*` theme resources for visual consistency.
Includes `AutomationProperties.Name` for accessibility.
### Files Changed
| File | Change |
|------|--------|
| `ListItemViewModel.cs` | Added `MaxVisibleTags`, `VisibleTags`,
`OverflowTagCount`, `HasOverflowTags`, `OverflowTagText` properties and
`UpdateVisibleTags()` |
| `ListPage.xaml` | `ItemsRepeater` binds `VisibleTags`, wrapped in
`StackPanel` with overflow `Border` badge |
### Validation
- [x] Build clean (`Microsoft.CmdPal.UI.ViewModels` +
`Microsoft.CmdPal.UI`)
- [x] Handles 0, 1-3, and 4+ tags correctly
- [x] No ABI/contract breaks (original `Tags`/`HasTags` preserved)
- [x] Overflow badge reuses existing theme resources
- [x] Accessibility: `AutomationProperties.Name` on overflow badge
### Follow-up
- Overflow badge accessible name could be enriched (e.g., "+3 more tags"
instead of "+3")
### Screenshots
<img width="972" height="545" alt="image"
src="https://github.com/user-attachments/assets/327dcdd0-2e62-435a-9b07-0432189cc947"
/>
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Sets `IsLoading = true` before querying open windows in
`WindowWalkerListPage.GetItems()` and resets it in a `finally` block, so
the UI shows a loading indicator while Window Walker results are being
fetched.
Closes#38314
## PR Checklist
- [x] Closes: #38314
- [ ] **Communication:** I've discussed this with core contributors
already.
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
## Detailed Description
The Window Walker extension was not setting `IsLoading` during its
search operation. This meant users got no visual feedback that a search
was in progress. This follows the same pattern used by other extensions
(AllAppsPage, IndexerPage, etc.).
## Validation Steps Performed
- Verified the change follows existing patterns in AllAppsPage.cs and
IndexerPage.cs
- The try/finally ensures IsLoading is always reset even if Query throws
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Enables dragging dock bands between monitors in edit mode. Previously,
drag-and-drop only worked within bands on the same monitor's dock.
Closes#47920
## PR Checklist
- [x] Closes: #47920
- [ ] **Communication:** I've discussed this with core contributors
already.
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
## Detailed Description
The existing drag-and-drop implementation used a local `_draggedBand`
field per `DockControl` instance, which meant the target DockControl on
another monitor had no reference to the dragged band. This PR fixes that
by:
1. **DataPackage properties** — `DragItemsStarting` now stores the band
ID and source monitor device ID in `DataPackage.Properties`, making the
identity available to any drop target in the same process.
2. **Cross-window drop acceptance** — `DragOver` and `DragEnter` now
check for `DockBandId` in `DataView.Properties` in addition to the local
`_draggedBand` field.
3. **CrossMonitorBandDropMessage** — A new `WeakReferenceMessenger`
message coordinates removal from the source dock after a successful
cross-monitor drop.
4. **ViewModel methods** — `DockViewModel.AcceptBandFromMonitor()`
creates a new band ViewModel on the target dock using the same factory
pattern as `AddBandToSection`. `RemoveBandById()` handles cleanup on the
source dock.
Both methods call `EnsureMonitorForked()` to fork per-monitor band
settings from global when needed, maintaining the existing customization
model.
## Validation Steps Performed
- Verified existing same-monitor drag-and-drop behavior is preserved
(local path unchanged)
- Verified cross-monitor handler properly stores/retrieves DataPackage
properties
- Reviewed against existing patterns in DockControl (AddBandToSection,
MoveBandWithoutSaving)
---------
Co-authored-by: root <root@io.bbq>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Clicking **Reset** in the Quick Access hotkey shortcut dialog crashes
PowerToys Settings. The root cause is that `C_ResetClick` sets the
`HotkeySettingsProperty` DependencyProperty to `null`, which causes the
WinUI3 XAML runtime to dereference a null pointer during two-way binding
property-change notification — a native E_POINTER crash (0x80004003).
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
### Root Cause
`C_ResetClick` previously did:
```csharp
hotkeySettings = null;
SetValue(HotkeySettingsProperty, null);
```
The `HotkeySettingsProperty` is bound two-way (`{x:Bind Mode=TwoWay}`)
to the ViewModel. `SetValue(null)` synchronously triggers the binding
chain: ViewModel setter → `NotifyPropertyChanged` → XAML runtime reads
back the DP value → null pointer → **native E_POINTER crash** in
`CoreMessagingXP.dll`.
### Fix
Use `new HotkeySettings()` instead of `null` — matching the existing
`C_ClearClick` pattern that never crashes:
```csharp
// C_ResetClick — fixed
hotkeySettings = new HotkeySettings();
SetValue(HotkeySettingsProperty, hotkeySettings);
SetKeys();
lastValidSettings = hotkeySettings;
shortcutDialog.Hide();
```
An empty `HotkeySettings` (`IsEmpty() == true`) semantically means "no
shortcut configured" — the same intent as null, but the XAML binding
chain always has a valid object to dereference.
Also added null-conditional guards in `OpenDialogButton_Click` as
defense-in-depth:
- `HotkeySettings?.GetKeysList() ?? new List<object>()` ensures `c.Keys`
is never null, preventing `{x:Bind Keys.Count}` in
`ShortcutDialogContentControl.xaml` from throwing.
- `hotkeySettings?.HasConflict ?? false` and
`hotkeySettings?.ConflictDescription` guard the remaining property
accesses.
- A null guard in `Hotkey_KeyDown` prevents a crash if the user presses
keys after Reset.
## Validation Steps Performed
- Reproduced crash: Settings → General → Quick Access shortcut → Edit →
Reset → crash (before fix)
- Verified fix: Same steps → no crash, dialog closes cleanly, shortcut
shows as empty
- Verified reopening dialog after Reset works without crash
- Verified pressing keys in dialog after Reset works without crash
- Built and tested with full Runner IPC (debug PowerToys.exe → Settings)
---------
Co-authored-by: Muyuan Li (from Dev Box) <muyuanli@microsoft.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: MuyuanMS <116717757+MuyuanMS@users.noreply.github.com>
Adds a **"Show file preview tooltip"** toggle to Peek's Behavior
settings, letting users disable the metadata tooltip (filename, type,
date modified, size) shown on hover over the preview.
## Changes
- **`PeekProperties`** — new `ShowFilePreviewTooltip: BoolProperty`
(default `true`)
- **`IUserSettings` / `UserSettings`** — expose the setting; updated by
the existing file watcher on settings change
- **`FilePreview.xaml`** — `ImagePreview` and `VideoPreview` now bind
`ToolTipService.ToolTip` directly to `InfoTooltip` (the same
attached-property form already used by `AudioControl`). The previous
explicit `<ToolTip Content="{x:Bind InfoTooltip}"/>` element form left a
`ToolTip` instance permanently attached, so nulling the content still
produced an empty popup on hover. With the attached-property form, a
`null` `InfoTooltip` detaches the tooltip entirely and no popup is
shown.
- **`FilePreview.xaml.cs`** — new `ShowFilePreviewTooltip`
DependencyProperty; sets `InfoTooltip = null` when disabled; re-triggers
`UpdateTooltipAsync` when re-enabled with a file loaded; `infoTooltip`
field changed to `string?`
- **`MainWindow.xaml.cs`** — reads setting from `IUserSettings` and
pushes it to `FilePreviewer.ShowFilePreviewTooltip` on each
`Initialize()` call, picking up the latest user preference each time
Peek activates
- **`PeekViewModel`** — `ShowFilePreviewTooltip` property for two-way
Settings UI binding
- **`PeekPage.xaml`** — toggle in the Behavior group, consistent with
existing toggles
- **`Resources.resw`** — localizable strings for header and description
## PR Checklist
- [x] Closes: #46621
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
Setting defaults to `true` — no behavior change for existing users.
The tooltip is suppressed by setting `InfoTooltip` to `null` while bound
through `ToolTipService.ToolTip` as an attached property. A `null` value
on the attached property prevents WinUI from creating a tooltip popup at
all (this is the same pattern `AudioControl` already used and behaves
correctly with null). The earlier explicit `<ToolTip>` element form did
*not* behave this way — the `ToolTip` instance stayed attached and an
empty popup appeared on hover — so the XAML was switched to the
attached-property form on `ImagePreview` and `VideoPreview`. The custom
tooltip placement logic (top/bottom based on cursor Y) is unaffected.
## Validation Steps Performed
- Verified toggle appears in Peek Settings > Behavior
- Toggling off fully suppresses the hover tooltip on image, video, and
audio previews (no empty bubble)
- Toggling back on restores tooltip with file metadata on next file load
- Default `true` preserves existing behavior
<img width="688" height="99" alt="image"
src="https://github.com/user-attachments/assets/369d0ba7-fc9a-495e-9603-f4b2b95ba68e"
/>
<img width="692" height="305" alt="image"
src="https://github.com/user-attachments/assets/cb5dfc69-a445-46b0-9c0a-041fbf4c89c2"
/>
<img width="684" height="73" alt="image"
src="https://github.com/user-attachments/assets/8088caa7-9536-4384-91dc-dbdbbe0ae755"
/>
<img width="686" height="430" alt="image"
src="https://github.com/user-attachments/assets/0d9d2d6f-bc42-4497-89c5-06ac9c18a2b7"
/>
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: niels9001 <9866362+niels9001@users.noreply.github.com>
Co-authored-by: Niels Laute <niels.laute@live.nl>
Co-authored-by: Muyuan Li (from Dev Box) <muyuanli@microsoft.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
`NavigateToString` throws `ArgumentException` when previewing Markdown
files containing many multi-byte UTF-8 characters (e.g., CJK) — file
size exceeds 2MB but character count stays under 1.5M, bypassing the
guard.
## Summary of the Pull Request
The size guard in `MarkdownPreviewHandlerControl` used
`markdownHTML.Length` (character count / UTF-16 code units), but
WebView2's `NavigateToString` limit is measured in **bytes**. A string
with 700K CJK characters has only 700K `.Length` units but ~2.1MB of
UTF-8 bytes — enough to crash the API while passing the old check.
**Changes:**
- **`MarkdownPreviewHandlerControl.cs`**: Replace character-count guard
with UTF-8 byte count:
```csharp
// Before
if (markdownHTML.Length > 1_500_000)
// After
if (System.Text.Encoding.UTF8.GetByteCount(markdownHTML) > 1_500_000)
```
When the byte threshold is exceeded, content is written to a temp file
and loaded via `_browser.Source` instead of `NavigateToString` —
existing fallback path, now correctly triggered.
- **`MarkdownPreviewHandlerTest.cs`**: Added 3 regression tests to
prevent this class of bug from recurring:
1. Multi-byte UTF-8 content (CJK, <1.5M chars but >2MB bytes) →
temp-file navigation path
2. Small ASCII content within both thresholds → `NavigateToString` path
3. Large ASCII content exceeding 1.5M chars → temp-file navigation path
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
The existing fallback (write HTML to a temp file, navigate via
`_browser.Source`) was already correct and handles arbitrarily large
content safely. The only bug was in the guard condition that decides
when to use it — it measured the wrong unit (characters vs. bytes).
Single-byte ASCII content is unaffected; only multi-byte Unicode content
was under-counted.
The tests use reflection to read the private `_localFileURI` field.
Since this field is set synchronously before `Controls.Add(_browser)`,
the check is race-free: once the wait loop exits with `Controls.Count >
0`, `_localFileURI` is guaranteed to have its final value.
## Validation Steps Performed
- Verified the fix by reasoning through the byte math: a file with 700K
CJK characters → `Length` = 700K (passes old check) → UTF-8 bytes ≈
2.1MB (fails new check → uses temp file path, no crash).
- Added 3 targeted unit tests in `UnitTests-MarkdownPreviewHandler`
covering the multi-byte threshold boundary; all 14 tests in the suite
pass (14/14).
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: MuyuanMS <116717757+MuyuanMS@users.noreply.github.com>
Co-authored-by: Muyuan Li (from Dev Box) <muyuanli@microsoft.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
1. Subscribes to GUID_CONSOLE_DISPLAY_STATE so PowerDisplay rescans
monitors when the console display wakes from sleep — previously, woken
monitors stayed unrecognized until the user manually re-triggered
discovery.
2. Locks the PowerDisplay UI immediately on wake to block stale
interactions before the rescan completes.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #47951
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
This PR moves storyboard that handles showing and hiding breadcrumbs in
Settings windows to a XAML resources.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
## Summary of the Pull Request
This adds `rand()` and `randi()` functions to Command Palette's
Calculator, making it consistent with Run.
It also expands upon the return values from `ToWStringFullPrecision()`,
so NaN, ParseError and +/-infinity results are passed back to the
caller, improving the specificity of the error message display.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #47707
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
The two new functions have been added to **ExprtkEvaluator.cpp**,
alongside `sign()` and `factorial()`. As they need to handle the state
of the RNG, they're slightly more complex in implementation. I used the
Mersenne Twister RNG with a uniform distribution, and the instances are
marked `static thread_local` in case the engine moves to multithreaded
evaluation in the future.
It's possible for the RNG to return a value out of the range of
`double`, and this is caught and `quiet_NaN()` is returned. To prevent
this being caught as a generic parse error, I updated
`ToWStringFullPrecision()` to distinguish between `NaN`, expression
parsing errors and infinity values. This should improve the accuracy of
error messages for other expressions, too.
Finally, I corrected a comment in **CalculateEngine.cs,** which still
referred to the Mages calculation engine. The log/ln mapping is the same
for both engines, so the comment was still accurate except for this
reference.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
Unit tests were added to exercise the new functions. All Calculator
tests pass:
<img width="375" height="59" alt="image"
src="https://github.com/user-attachments/assets/5a33e1ed-a4fd-4d53-b9ba-6b44000f1bf4"
/>
Confirmed that error messages are displaying correctly for the
newly-exposed result types:
**Not a number**
<img width="787" height="128" alt="image"
src="https://github.com/user-attachments/assets/8c73dcf6-122b-4af8-bf1a-62284842433a"
/>
<img width="786" height="145" alt="image"
src="https://github.com/user-attachments/assets/fe14338c-1160-4aae-83dd-5ca3491ae59e"
/>
**+/- Infinity**
<img width="898" height="137" alt="image"
src="https://github.com/user-attachments/assets/20cfacda-72a7-44bb-a875-af7be39ee7e2"
/>
**Parser failure**
<img width="607" height="139" alt="image"
src="https://github.com/user-attachments/assets/7d7120b2-a2cf-45b6-ab89-79af4051fa50"
/>
<img width="587" height="140" alt="image"
src="https://github.com/user-attachments/assets/2dc7a365-7ee6-4379-8b3f-47b3912e6891"
/>
## Summary of the Pull Request
This fixes Calculator functions with multiple arguments in cultures
where the number group separator and list separator are identical, e.g.
**en-US** and **en-GB**.
It maintains existing parsing behaviour for other cultures where the
separators differ, e.g. **de-DE**.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #47726
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
The root issue was in the number translation step. In cultures such as
en-US, commas were being consumed as part of numeric tokens and were
treated as number group separators, which broke functions whcih took
multiple arguments. For example, `max(1,2)` would be translated as
`max(12)` and `pow(2,3)` as `pow(23)`. The number translator's result
would be passed on to ExprTK, which could sometimes still interpret the
input and would give a result (`12` in the case of `max(1,2)`); for
cases like `pow(2,3)`, it would surface an error, as the expression
`pow(23)` is invalid.
This fix resolves the ambiguity in two ways:
- It uses stricter grouped number matching when the culture's list
separator and number group separator are the same.
- It preserves separator characters for multi-argument functions. This
means that expressions like `max(123,456)` are correctly interpreted as
two arguments inside the function call, while `123,456` outside a
function call is still interpreted as a grouped number.
Care has been taken to preserve grouped numbers inside single-argument
functions, e.g. `ceil(123,456.23)`.
The more permissive parsing for cultures which do not have this
ambiguity has been retained. (Arguably this is too loose, but that's
something to consider separately.)
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
Added and updated unit tests, with **en-US** used as the representative
'ambiguous culture'. Coverage includes:
- `max()`, `min()` and `pow()`
- Grouped numbers in single-argument functions like `ceil()`, `floor()`,
`round()`, `log()` and `sin()`.
- Nested expressions
- Spacing around function calls
- Scientific notation
- Hexadecimal, binary and octal literals
All tests pass:
<img width="696" height="52" alt="image"
src="https://github.com/user-attachments/assets/ed89fbb9-70e0-4cf7-8e13-12f1f36b6037"
/>
Manual testing was also performed to confirm:
- Multi-argument functions now evaluate correctly
<img width="243" height="136" alt="image"
src="https://github.com/user-attachments/assets/b5c96954-ee6d-4842-b599-561ccbd10607"
/>
- Grouped numbers inside single-argument functions still work
<img width="356" height="135" alt="image"
src="https://github.com/user-attachments/assets/918425b0-cce4-4708-855b-c8b4916e6a4a"
/>
- Nested expressions and spacing variations are handled correctly
<img width="606" height="137" alt="image"
src="https://github.com/user-attachments/assets/12be5aa9-ba33-4000-96d5-444a2932bbe7"
/>
<img width="482" height="135" alt="image"
src="https://github.com/user-attachments/assets/aea2ebce-7c88-469e-b9e4-bdb7099ef538"
/>
## Summary of the Pull Request
This PR fixes WMC1506 warnings introduced by the extension gallery PR.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
## Summary of the Pull Request
This PR filters URIs from extension gallery and allows only HTTP/HTTPS
URIs as links for the installation page.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
This pull request updates the logging behavior in the `LogMessage`
method to ensure that log messages are categorized and logged according
to their severity (Error, Warning, or Info), instead of always using
debug-level logging.
Logging improvements:
* Updated the `LogMessage` method in `AppExtensionHost.cs` to log
messages using `CoreLogger.LogError`, `CoreLogger.LogWarning`, or
`CoreLogger.LogInfo` based on the `MessageState` of the incoming
message, improving log clarity and severity categorization.
Previously, any logging sent to LogMessage (which is exposed in the
toolkit) would only be written when debugging. Any error/warning/info
messages sent as part of the normal operation would be bypassed because
it only used `CoreLogger.LogDebug`
Now extension developers can log errors/warnings and use PowerToys log
export / bug report feature to help their users provide them with
actionable data to make their extensions better.
(this PR is an updated version of #43784)
This PR adds a new type of page to Command Palette:
The `ParametersPage`.
This allows extensions to create commands that require a set of
parameters
before invoking the command. Previously, extensions could create
commands with a
form page to use an adaptive card for parameter input, but that was a
relatively
heavyweight UX.
Instead, the `ParametersPage` allows extensions to define a set of
lightweight
inputs, which allows for a more streamlined experience.
The parameters page is made up of a set of "runs". Each run represents a
single element in the search box. Runs can be either:
* A label run: a static piece of text
* An value run: some input for the user to provide a value. These fall
into several categories:
* String input
* Command Input
* `IInvokableCommand`s become buttons in the search box
* `IListPage`s become a list input to pick from (**these will be added
in a follow-up PR**)
There are a ton of samples included.
I also added all my draft notes in the drafts folder, to see how we got
here.
I'd skip reviewing those.
Furthermore, I added the "dumb" token support, where an extension can
opt in to
having tokens in the search box, delimited by ZWSP characters.
The XAML styling was fixed by Niels a few months back
Closes#40948
---------
Co-authored-by: Niels Laute <niels.laute@live.nl>
## Summary
Adds a **Hide app descriptions** toggle setting to the Command Palette
All Apps extension. When enabled, the descriptive subtitle text next to
app results is hidden for a cleaner look.
### Changes
| File | Change |
|------|--------|
| `AllAppsSettings.cs` | New `HideAppDescriptions` `ToggleSetting`
(default: `false`), following the existing `EnableStartMenuSource`
pattern |
| `AllAppsPage.cs` | Conditionally clears `Subtitle` on each
`AppListItem` in `GetPrograms()` when setting is enabled |
| `Resources.resx` / `Resources.Designer.cs` | Resource strings for the
setting label and description |
| `AllAppsPageTests.cs` | Two new tests verifying subtitle visibility
with setting on/off |
### Validation
- [x] Build clean (exit code 0)
- [x] All 14 unit tests pass (including 2 new tests)
- [x] Setting defaults to `false` (preserves current behavior)
- [x] Empty subtitle renders gracefully (no placeholder shown)
- [x] No ABI breaks — all changes scoped to All Apps extension
Closes#46634
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Fixes#46810 — When navigating backward in CmdPal (Esc/Backspace), the
bottom command bar retained stale commands from the previous page until
the user changed selection.
## Root Cause
In `ListPage.xaml.cs` `OnNavigatedTo()`, the back-navigation path sets
selection using `SuppressSelectionChangedScope()`, which prevents
`Items_SelectionChanged` from firing. This means `PushSelectionToVm()`
is never called, so `UpdateCommandBarMessage` is never sent to
`CommandBarViewModel`, and the command bar displays stale state.
## Fix
Added `PushSelectionToVm()` after the selection restoration block inside
the back-navigation dispatcher callback. This is safe because:
- `PushSelectionToVm()` is idempotent (guards with
`ReferenceEquals(_lastPushedToVm, li)`)
- Handles both selected items (pushes to VM) and no selection (sends
null)
- Triggers `UpdateCommandBarMessage` which refreshes the command bar
## Validation
### Manual Test Steps
| Scenario | Expected |
|----------|----------|
| **Esc navigation** — Navigate into nested page, press Esc | Bottom bar
commands update immediately to parent page |
| **Backspace navigation** — Navigate into nested page, press Backspace
| Bottom bar commands update immediately to parent page |
| **Forward navigation** — Navigate forward into a page | Commands
update (no regression) |
| **Selection change** — Change selection on any page | Commands update
(no regression) |
| **No flicker** — Navigate back and observe command bar | Single clean
update, no double-fire |
| **Empty page** — Navigate back to page with no selectable items |
Graceful handling, no crash |
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Fixes the pluralization bug in Command Palette settings where extensions
with a single command displayed "1 commands" and "1 fallback commands"
instead of the correct singular forms.
### Changes
**`ProviderSettingsViewModel.cs`** — Rewrote the `ExtensionSubtext`
property to use tuple pattern matching that selects the correct
singular/plural format string based on whether command count and
fallback count equal 1.
**`Resources.resx`** — Added 4 new resource strings for singular form
combinations:
- `builtin_extension_subtext_singular` — "{0}, {1} command"
- `builtin_extension_subtext_with_fallback_singular_command` — "{0}, {1}
command, {2} fallback commands"
- `builtin_extension_subtext_with_fallback_singular_fallback` — "{0},
{1} commands, {2} fallback command"
- `builtin_extension_subtext_with_fallback_singular_both` — "{0}, {1}
command, {2} fallback command"
**`ProviderSettingsViewModelPluralizationTests.cs`** — 17 new test cases
covering all singular/plural combinations for command and fallback
command counts.
### Validation
- [x] Matches existing `CompositeFormat` pattern used elsewhere in
CmdPal
- [x] Follows `.editorconfig` and StyleCop conventions
- [x] All files within `CommandPalette.slnf` scope
Closes#47110
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Diagnostic / prototype fix for the LNK2038 C++/WinRT version mismatch
that has been failing PowerToys CI on every batched-CI run since commit
`59eefd9581` (5/14):
`
SettingsAPI.lib(settings_objects.obj): error LNK2038: mismatch detected
for 'C++/WinRT version':
value '2.0.250303.1' doesn't match value '2.0.250303.5' in main.obj
[src/modules/GrabAndMove/GrabAndMove/GrabAndMove.vcxproj]
`
## Root cause
GrabAndMove.vcxproj does not import the `Microsoft.Windows.CppWinRT`
NuGet package, so `main.cpp` picks up `<winrt/Windows.Foundation.h>`
(included transitively via `SettingsAPI/settings_objects.h` ->
`common/utils/json.h`) from the **Windows SDK's in-box CppWinRT**
instead of the repo-pinned NuGet version.
After the SHINE-VS18-Latest agent image picked up a newer Windows SDK
shipping `CppWinRT 2.0.250303.5`, `main.obj` began emitting that version
via `#pragma detect_mismatch`, while `SettingsAPI.lib` continued to be
built against the pinned NuGet `2.0.250303.1`. The linker rejects the
mix.
This was masked while the agent SDK happened to ship a matching CppWinRT
version, and surfaced after #47470 (Bump WindowsAppSDK to 2.0.1) plus
the agent image roll.
## Fix
Mirror the canonical CppWinRT NuGet wiring used by every other native
vcxproj in the repo (see `src/common/SettingsAPI/SettingsAPI.vcxproj`
for the reference pattern):
- Add `packages.config` pinning `Microsoft.Windows.CppWinRT
2.0.250303.1`.
- Import the props after `Microsoft.Cpp.Default.props`.
- Import the targets in an `ExtensionTargets` `ImportGroup`.
- Add `EnsureNuGetPackageBuildImports` for restore-time validation.
## Validation
- Local x64/Release build of GrabAndMove.vcxproj clean (linked against
SettingsAPI.lib without LNK2038).
- (Local SDK on the dev box already ships matching CppWinRT
2.0.250303.1, so the LNK2038 cannot reproduce locally; the CI pool agent
has the newer SDK that exposes the latent issue.)
- Awaiting PowerToys CI to confirm fix on the agent image.
## Related
- #47470 (Bump WindowsAppSDK to 2.0.1) — preceded but did not directly
cause this; just changed which CppWinRT was sitting in the include path.
- Failing CI runs: 319304, 319351, 319593 (all on shine-oss PowerToys CI
definition 3).
## Summary of the Pull Request
Prevents users from clearing the name field in ImageResizer size preset
edit dialog. Empty names made the UI confusing without causing errors.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
Added validation guard in `ImageSize.Name` property setter:
```csharp
public string Name
{
get => _name;
set
{
if (!string.IsNullOrWhiteSpace(value))
{
SetProperty(ref _name, value);
}
}
}
```
Invalid assignments (empty, null, whitespace) are silently ignored,
preserving the existing value. This matches the existing pattern used
for `FileName` validation in `ImageResizerViewModel`.
TwoWay binding in UI causes the TextBox to revert when users attempt to
clear the field—standard behavior for required fields.
## Validation Steps Performed
- Added unit test `ImageSizeNameShouldNotBeSetToEmptyOrNull()` covering
all rejection and acceptance cases
- Verified silent rejection behavior matches `FileName` property pattern
> [!WARNING]
>
> <details>
> <summary>Firewall rules blocked me from connecting to one or more
addresses (expand for details)</summary>
>
> #### I tried to connect to the following addresses, but was blocked by
firewall rules:
>
> - `i1qvsblobprodcus353.vsblob.vsassets.io`
> - Triggering command: `/usr/bin/dotnet dotnet build
/home/REDACTED/work/PowerToys/PowerToys/src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.Registry/Helpers/ResultHelper.cs
/home/REDACTED/work/PowerToys/PowerToys/src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.Registry/Helpers/RegistryHelper.cs
/home/REDACTED/work/PowerToys/PowerToys/src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.Registry/Enumerations/TruncateSide.cs
/home/REDACTED/work/PowerToys/PowerToys/src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.Registry/Constants/KeyName.cs
/home/REDACTED/work/PowerToys/PowerToys/src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.Registry/Constants/MaxTextLength.cs
/home/REDACTED/work/PowerToys/PowerToys/src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.Registry/Commands/CopyRegistryInfoCommand.cs
/home/REDACTED/work/PowerToys/PowerToys/src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.Registry/Commands/OpenKeyInEditorCommand.cs
/home/REDACTED/work/PowerToys/PowerToys/src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.Re
/home/REDACTED/work/PowerToys/PowerToys/src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.Registry/CopyType.cs
/home/REDACTED/work/PowerToys/PowerToys/src/modules/cmdpal/ext/Microsoft.CmdPal.Ext.Registry/Icons.cs`
(dns block)
>
> If you need me to access, download, or install something from one of
these locations, you can either:
>
> - Configure [Actions setup
steps](https://gh.io/copilot/actions-setup-steps) to set up my
environment, which run before the firewall is enabled
> - Add the appropriate URLs or hosts to the custom allowlist in this
repository's [Copilot coding agent
settings](https://github.com/microsoft/PowerToys/settings/copilot/coding_agent)
(admins only)
>
> </details>
<!-- START COPILOT ORIGINAL PROMPT -->
<details>
<summary>Original prompt</summary>
>
> ----
>
> *This section details on the original issue you should resolve*
>
> <issue_title>[Settings: ImageResizer] Edit size dialog: Add validation
against empty name</issue_title>
> <issue_description>There is one small thing we should add. (But I
don't know how.)
> The data in the name field of the edit dialog should be validated
against `.NullOrEmpty`!
> _Originally posted by @htcfreek in
https://github.com/microsoft/PowerToys/pull/13285#issuecomment-924113251_
>
> **Detailed description:**
> While users edit an image size in the ImageResizer settings on PT's
settings ux they are able to clear the name of the image size. This
doesn't lead to errors/exceptions, but it also doesn't make sense to
have sizes without a name. So we should prevent user from doing this in
the edit dialog.
>
> _**Note:**_
> This is the second and last change that is needed to prevent the users
from having incorrect image size settings (#8698). But I think this is
not as relevant, because a missing name won't break image resizing._
>
> ---
>
> cc: @niels9001 </issue_description>
>
> <agent_instructions>Implement a proper fix for this issue. Do NOT
create placeholder stub files. Create actual working code that addresses
the issue described. Follow PowerToys coding guidelines and
conventions.</agent_instructions>
>
> ## Comments on the Issue (you are @copilot in this section)
>
> <comments>
> <comment_new><author>@TheJoeFin</author><body>
> does this issue still happen with v0.73.0?
/needinfo</body></comment_new>
> </comments>
>
</details>
<!-- START COPILOT CODING AGENT SUFFIX -->
- Fixesmicrosoft/PowerToys#13336
<!-- START COPILOT CODING AGENT TIPS -->
---
✨ Let Copilot coding agent [set things up for
you](https://github.com/microsoft/PowerToys/issues/new?title=✨+Set+up+Copilot+instructions&body=Configure%20instructions%20for%20this%20repository%20as%20documented%20in%20%5BBest%20practices%20for%20Copilot%20coding%20agent%20in%20your%20repository%5D%28https://gh.io/copilot-coding-agent-tips%29%2E%0A%0A%3COnboard%20this%20repo%3E&assignees=copilot)
— coding agent works faster and does higher quality work when set up for
your repo.
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: yeelam-gordon <73506701+yeelam-gordon@users.noreply.github.com>
## Summary of the Pull Request
Two related installer changes to (1) eliminate genuinely-unused
dependencies and (2) deduplicate shared WinAppSDK files between
`<install>\` and `<install>\WinUI3Apps\` to shrink the installer
download.
### 1. Remove unused dependencies (~11 MB savings per output location)
- **System.Data.SqlClient**: Removed from MouseWithoutBorders projects
and the central `Directory.Packages.props` pin. It was a transitive
dependency of `Microsoft.Windows.Compatibility` but PowerToys has zero
SQL database usage.
- **Unused `using` import**: Removed `using
System.ServiceModel.Channels` from MouseWithoutBorders `Program.cs` (no
WCF usage).
- **MFC / C++ AMP / OpenMP DLLs**: Added `RemoveUnusedVCRuntimeDlls`
target in `Directory.Build.targets` to clean up `mfc140*`, `mfcm140*`,
`vcamp140*`, and `vcomp140*` DLLs that leak from the VC++
Redistributable tree but are not imported by any PowerToys binary
(verified with `dumpbin /dependents` across all installed binaries).
Also excluded MFC DLLs from installer file collection.
### 2. WinAppSDK file deduplication (build-time only; install-time uses
copy)
**Background**: The `WinUI3Apps` subfolder must remain a real directory
because MSIX sparse package registration applies DACL changes to the
`ExternalLocation` folder (PR #47177). Flattening is not viable.
**Build-time** (`generateAllFileComponents.ps1`): computes the SHA256
intersection of root and `WinUI3Apps` files, and for each file that is
also present in the BaseApplications WXS file list, removes the
duplicate from the WinUI3Apps WXS component list and writes its name to
a `hardlinks.txt` manifest. The BaseApplications cross-check ensures we
never deduplicate a file the MSI does not actually deploy at the install
root, which would otherwise leave both copies missing post-install. The
manifest is written as UTF-8 without BOM (via
`[System.IO.File]::WriteAllLines` with `UTF8Encoding($false)`) so its
encoding is identical regardless of the build host's PowerShell version.
This step produces the **MSI download-size win** (~97 MB smaller cab;
LZX:21 was already deduplicating most byte-identical content
automatically inside the cab).
**Install-time** (`CreateWinAppSDKHardlinksCA` custom action):
- Reads `hardlinks.txt` after `InstallFiles` as a raw byte stream and
converts each line to a `std::wstring` via `MultiByteToWideChar(CP_UTF8,
MB_ERR_INVALID_CHARS, ...)`. Avoids `std::wifstream`'s ANSI-codepage
codecvt so non-ASCII paths can never be silently mangled.
- For each entry, computes `(installDir / fileName).lexically_normal()`
and `(winui3Dir / fileName).lexically_normal()`, then verifies via
`std::mismatch` that each resolved path is still rooted at its
respective folder. Manifest entries containing `..`, absolute paths, or
alternate-stream syntax are logged and skipped.
- Materialises each validated entry from `<install>\<name>` into
`<install>\WinUI3Apps\<name>` via `fs::copy_file` (overwrite_existing).
- Reports the per-file copy / failure counts to the install log. If
every entry failed (`created == 0 && failed > 0`), the CA escalates to
`E_FAIL` so the install does not silently succeed with an unusable
WinUI3Apps tree.
`DeleteWinAppSDKHardlinksCA` removes the materialised copies before
`RemoveFiles` on uninstall, using the same UTF-8 reader and per-entry
containment check.
**WiX sequencing**: `CreateWinAppSDKHardlinks` runs
`After="InstallFiles"` with `Condition="NOT Installed OR
WIX_UPGRADE_DETECTED OR REINSTALL"` so a `msiexec /fa` repair refreshes
the deduplicated copies (otherwise `RemoveFiles` would orphan them).
#### Why copy and not hard-link
A hard-linked variant of this CA was originally proposed but caused a
Monaco preview-handler regression. Hard-links share an NTFS inode (and
therefore one DACL) between `<install>\<file>` and
`<install>\WinUI3Apps\<file>`. The MSIX sparse-package registrations for
PowerRename / ImageResizer / FileLocksmith / NewPlus run after the dedup
CA and propagate the `WinUI3Apps` parent's rich DACL (Capability SID, 5×
Package SIDs, 5× conditional SYSAPPID ACE, RC SID) onto the shared
inode. The root path then also exposes the rich DACL, which trips a
kernel "stricter access evaluation" path that blocks the LOW-IL
`prevhost.exe` from `LoadLibrary`-ing `hostfxr.dll` (and the rest of the
.NET runtime), turning the Monaco preview pane blank for `.json` / `.md`
/ `.cs` / `.xaml` / `.svg` / `.xml` files.
`fs::copy_file` creates a **fresh inode** for the WinUI3Apps copy. The
root inode keeps its simple DACL (`SY:F + BA:F + owner:F` + inherited
`BU:RX`) so LOW-IL `prevhost.exe` can still load it — Monaco preview
works. The WinUI3Apps copy inherits the WinUI3Apps parent's rich DACL
via normal NTFS inheritance (matches 0.99.1 behaviour exactly) — MSIX
context-menu shells continue to work.
#### Trade-off
| Metric | Hard-link variant (rejected) | This PR (file copy) | 0.99.1
(no dedup) |
|---|---|---|---|
| MSI size | ~296 MB | ~296 MB | ~393 MB |
| On-disk after install | ~2,475 MB | ~2,772 MB | ~2,772 MB |
| DACL contamination risk | YES (broke Monaco) | NO | NO |
The on-disk savings (~297 MB) are given up in exchange for eliminating
the DACL contamination risk; the **installer download savings (~97 MB)**
are preserved by the build-time WiX/cab dedup.
#### Edge cases handled
- Empty duplicate list: `hardlinks.txt` always written, CA handles
empty.
- All files duplicated: `Generate-FileComponents` returns early for
empty list.
- File stripped from BaseApplications by an earlier build step:
BaseApplications cross-check skips it during dedup so neither copy goes
missing.
- Manifest entry escapes install root (`..`, absolute path): rejected
per-entry, install continues.
- Manifest line is non-UTF-8: rejected per-entry, install continues.
- Source missing at install time: per-entry skip, install continues.
- All copies fail: install aborts loudly via `E_FAIL` (catastrophic-case
escalation).
- Upgrade or `msiexec /fa` repair: CA fires (`NOT Installed OR
WIX_UPGRADE_DETECTED OR REINSTALL`).
**MSI repair risk**: Burn bundle uses `SuppressRepair=yes` and
`MajorUpgrade` (full uninstall + reinstall) for all version upgrades, so
the standard upgrade path is unaffected. The `OR REINSTALL` clause
covers power users running `msiexec /fa` directly.
## PR Checklist
- [x] **Communication:** Discussed approach via PRs #46866, #47177,
#46745
- [ ] **Tests:** Installer infrastructure only — no runtime behaviour
changes
- [ ] **Localization:** N/A
- [ ] **Dev docs:** N/A
- [ ] **New binaries:** N/A
## Detailed Description of the Pull Request / Additional comments
Based on the approach from PR #46745 by @yeelam-gordon, rebased onto
latest main and switched from hard-links to file copies after the DACL
contamination root cause was identified. Hardening (UTF-8 read, path
containment, catastrophic-case escalation, REINSTALL repair,
BaseApplications-filtered dedup) added in response to review feedback.
These changes are purely build/installer infrastructure — no runtime
behaviour changes to any PowerToys module.
## Validation Steps Performed
Validated on a 0.99.4 / 0.99.5 local install (per-user
`%LocalAppData%\PowerToys`):
- ✅ `dumpbin /dependents` across the installed PowerToys tree confirmed
zero binaries import `mfc140*`, `mfcm140*`, `vcamp140*`, or `vcomp140*`
— the cleanup target removes ~11 MB of genuinely unused VC runtime DLLs.
- ✅ `System.Data.SqlClient` has zero call-sites in PowerToys source.
- ✅ Local installer build produces a 296 MB MSI (down from 393 MB
pre-dedup, ~97 MB cab savings purely from the build-time WiX dedup).
- ✅ MSI table inspection (`wix msi decompile`) confirms the deferred CAs
are present (`CreateWinAppSDKHardlinks`, `DeleteWinAppSDKHardlinks`) and
the `hardlinks.txt` File row is registered.
- ✅ MSI table inspection confirms .NET runtime DLLs (`hostfxr.dll`,
`coreclr.dll`, `hostpolicy.dll`, `clretwrc.dll`, `Accessibility.dll`,
`backup_restore_settings.json`) appear ONLY in
`BaseApplicationsFiles_File_*`, NOT in `WinUI3ApplicationsFiles_File_*`
— proving the build-time dedup worked.
- ✅ Post-install verification: deduplicated files materialised at both
root and WinUI3Apps with byte-identical SHA256 hashes, and `fsutil
hardlink list` returns link-count == 1 for each — proving the
install-time copy approach worked, not hard-link.
- ✅ DACL on root .NET runtime DLLs is clean: no Package SID, no
Capability SID, no SYSAPPID conditional ACE, no `ALL APPLICATION
PACKAGES` ACE — Monaco preview load path is safe.
- ✅ DACL on WinUI3Apps copies has the rich MSIX inheritance —
context-menu shells continue to work (matches 0.99.1).
- ✅ All four MSIX sparse packages (PowerRename, ImageResizer,
FileLocksmith, NewPlus) registered after install.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Fixes the `Auto-label Issues by Area` workflow, which currently logs
`GITHUB_TOKEN is not set; skipping.` on every run and never applies
labels.
Failing run on issue #47818:
https://github.com/microsoft/PowerToys/actions/runs/25722067064/job/75525452318
## Root cause
`actions/github-script@v7` consumes its `github-token` input only to
authenticate the injected `github` Octokit object. It does **not**
export that value to `process.env.GITHUB_TOKEN`. The inline script reads
`process.env.GITHUB_TOKEN` to authorize a direct `fetch()` against
`https://models.inference.ai.azure.com/chat/completions`, so the token
check at the top of `labelIssue()` always fails and the function returns
early before calling the model.
## Fix
Add a step-level `env:` block exposing `GITHUB_TOKEN` to the Node
process running the inline script. The existing `with.github-token`
input is preserved so the injected `github` Octokit continues to
authenticate.
`yaml
- name: Apply area labels with AI
uses: actions/github-script@v7
env:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
with:
github-token: ${{ secrets.GITHUB_TOKEN }}
`
5 lines added, 0 removed.
## Security
- `secrets.GITHUB_TOKEN` is the workflow's built-in ephemeral token
(auto-issued per run, auto-revoked at job end). Not a PAT.
- Scope is already constrained by `permissions: models: read, issues:
write` at the top of the workflow. No widening.
- Exposure is unchanged: the token is already loaded into the same Node
process by `actions/github-script` via `github-token:`. The `env:`
mapping just lets the script body read what the action already has in
the same process.
- The token is sent only to GitHub's own GitHub Models inference
endpoint, which is the documented use of `models: read`.
- Triggers are safe: `issues: opened/reopened` (issue body is
JSON-encoded into the request body, never interpolated into a shell) and
`workflow_dispatch` (write-access required).
- Token is never logged.
## Validation
After merge, re-run the workflow on issue #47818 via Actions ->
"Auto-label Issues by Area" -> Run workflow, and confirm logs show
`Model response: ...` instead of `GITHUB_TOKEN is not set; skipping.`
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Adds the **Extension Gallery** to Command Palette — a built-in page
where users can discover, browse, and install community extensions
without leaving the app.
https://github.com/user-attachments/assets/e4565333-b970-4085-9e40-5cfd207e533b
## How it works
### 1. The extension author's side
Extensions are listed in the external repo
**[`microsoft/CmdPal-Extensions`](https://github.com/microsoft/CmdPal-Extensions)**.
To get an extension into the in-app gallery, an author opens a PR there
that adds a single entry to `extensions.json`. Nothing in PowerToys
itself needs to change. A typical entry looks like:
```json
{
"id": "contoso.sample",
"title": "Sample Extension",
"description": "Short blurb shown in the list and detail view.",
"author": { "name": "Contoso", "url": "https://github.com/contoso" },
"homepage": "https://github.com/contoso/sample",
"iconUrl": "https://.../icon.png",
"screenshotUrls": ["https://.../screenshot-1.png"],
"tags": ["sample"],
"installSources": [
{ "type": "winget", "id": "Contoso.SampleExtension" },
{ "type": "msstore", "id": "9P..." },
{ "type": "url", "uri": "https://github.com/contoso/sample/releases/latest" }
],
"detection": { "packageFamilyName": "Contoso.SampleExtension_8wekyb..." }
}
```
- `id`, `title`, `description`, `author.name`, and at least one
`installSources` entry are required; everything else is optional.
- `installSources` can mix and match `winget` / `msstore` / `url`. The
gallery shows an install button for the first source it can handle
(WinGet preferred) and exposes any remaining sources as links.
- `detection.packageFamilyName` lets CmdPal recognise an
already-installed packaged extension before any WinGet lookup resolves,
so the "Installed" badge appears instantly.
Once the PR is merged into `CmdPal-Extensions`, every running copy of
CmdPal picks the new entry up the next time its feed cache expires
(within 4 hours) or when the user clicks **Refresh**.
### 2. What CmdPal does with it
`ExtensionGalleryService` (in `Microsoft.CmdPal.Common`) owns the whole
pipeline:
1. **Resolve the feed URL.** Default is
`https://raw.githubusercontent.com/microsoft/CmdPal-Extensions/refs/heads/main/extensions.json`.
A hidden setting (`GalleryFeedUrl`) lets developers point at a custom
URL or a `file://` path for local testing.
2. **Fetch** the feed through `ExtensionGalleryHttpClient`, which wraps
`HttpCachingClient` — a conditional-GET + on-disk cache layer built on
`HttpClient` (ETag / `If-None-Match`, 30 s timeout, UA
`PowerToys-CmdPal/1.0`).
3. **Parse** with the source-generated `GallerySerializationContext`
into a strongly-typed `GalleryRemoteIndex` (`{ "extensions": [ ... ]
}`). Entries without an `id` are dropped.
4. **Normalize** relative `iconUrl` / `screenshotUrls` against the feed
URL (useful for local `file://` feeds).
5. **Localize icons.** Each HTTP icon URL is pulled through the same
cache and rewritten to a local `file://` URI before the view model binds
to it, so the list renders instantly on subsequent loads and works
offline.
6. **Prune** cached resources that are no longer referenced, but only
after a successful forced refresh.
The gallery page itself is built on top of `ExtensionGalleryViewModel`,
with `ExtensionGalleryItemViewModel` handling per-entry concerns —
install/update/uninstall (via the shared WinGet service),
installed-state detection, and joining in-flight install progress so the
global `WinGetOperationsButton` in the top bar stays in sync.
### 3. Caching + offline behaviour
The cache lives under
`ApplicationData.Current.LocalCacheFolder\GalleryCache\` when CmdPal
runs packaged, or
`%LOCALAPPDATA%\Microsoft\PowerToys\Microsoft.CmdPal\Cache\GalleryCache\`
when unpackaged.
| Resource | TTL |
|-----------------|----------|
| `extensions.json` feed | 4 hours |
| Icons (per URL) | 24 hours |
Each fetch returns a `GalleryFetchResult` whose flags drive the UI:
- `FromCache` — cache was still fresh, no network call was made.
- `UsedFallbackCache` — network failed; the last-known-good cached copy
was served instead. The page shows a "showing cached data" info bar.
- `RateLimited` — origin returned `429` and no fallback was available.
The page shows a rate-limit error.
`RefreshAsync` (wired up to the gallery's refresh button) forces a fresh
conditional GET, then prunes any cached files that the new feed no
longer references.
### 4. WinGet install flow
- `installSources[type=winget].id` is handed to the shared WinGet
service for install/update/uninstall.
- In-flight operations are surfaced by `WinGetOperationsButton` in the
top bar with per-operation progress.
- `detection.packageFamilyName` is consulted first so that the gallery
can show "Installed" / "Update available" without waiting on WinGet
metadata.
### Top-level command cleanup
- Removed the separate "Find extensions from WinGet" and "Find
extensions from the Store" top-level commands — the gallery replaces
both.
- Renamed the gallery command to **"Find and install Command Palette
extensions"** and gave it the extensions puzzle-piece icon.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## New projects / areas
| Area | What |
|------|------|
| Microsoft.CmdPal.Common | Gallery models, `ExtensionGalleryService`
(fetch + cache), HTTP caching layer (`HttpCachingClient`,
`FileSystemHttpResourceCacheStore`), WinGet service abstractions and
implementations |
| Microsoft.CmdPal.UI.ViewModels | `ExtensionGalleryViewModel`,
`ExtensionGalleryItemViewModel`, WinGet operation view models, gallery
sort options |
| Microsoft.CmdPal.UI | `ExtensionGalleryPage.xaml`,
`ExtensionGalleryItemPage.xaml`, `IconCarouselControl`,
`WinGetOperationsButton`, service registrations |
| Microsoft.CmdPal.Ext.WinGet | Streamlined — removed the two redundant
"find extensions" top-level commands, kept the general WinGet search
page |
| Tests | Unit tests for gallery service, gallery view models, WinGet
services |
| Docs |
[`doc/devdocs/modules/cmdpal/extension-gallery/extension-gallery.md`](https://github.com/microsoft/PowerToys/blob/dev/jpolasek/f/46628-cmdpal-extension-gallery/doc/devdocs/modules/cmdpal/extension-gallery/extension-gallery.md)
— dev reference for the runtime, caching, and feed shape |
## Validation Steps Performed
- Gallery loads and displays extensions from the remote index
- Search, sort, and filtering work as expected
- WinGet install/update/uninstall flow works end-to-end with progress
tracking
- Loading state correctly hides all content until data is fetched
- Offline / cache-fallback path surfaces the info bar as expected
- Spell-check CI workflow passes
---------
Co-authored-by: Niels Laute <niels.laute@live.nl>
Co-authored-by: niels9001 <9866362+niels9001@users.noreply.github.com>
## Summary of the Pull Request
The Calculator components in Command Palette and Run produced incorrect
results or errors for `log` and `ln` inputs where spaces exist between
the function name and the argument list. This is because input
validation allowed for those spaces, but this was not respected in the
log mapping code which transforms user input into a string for
consumption by the expression evaluator engine.
The result of this is discrepancy was that:
- Natural log would be called instead of log base 10 for `log (n)`.
- An error would be shown for `ln (n)`.
For example:
<img width="556" height="166" alt="image"
src="https://github.com/user-attachments/assets/d2110292-ca8f-4635-bdef-9fa4f2211deb"
/>
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #47759
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
The issue was with this code:
a650504640/src/modules/launcher/Plugins/Microsoft.PowerToys.Run.Plugin.Calculator/CalculateEngine.cs (L56-L58)
This maps user input such as `log(10)` or `ln(10)` to the functions the
back-end expression evaluator understands. For Mages and ExprTK, this is
mapped to `log10(n)` for base 10 logs or `log(n)` for natural logs.
Unfortunately, the input validator regex allows for spaces between the
function name and the start of the argument list, and the string replace
does not. When spaces exist:
- For `log (n)` - the log10 match is missed and the expression is
interpreted as a natural log, producing incorrect results.
- For `ln (n)` - the string is passed as-is to the back-end. Neither
Mages nor ExprTK recognise it, so an error is returned.
The fix is to replace the string replacement code with two regexes.
These are both forgiving of spaces between the function name and
argument list:
For log base 10: `"log(?![0-9])\\s*\\("`
For natural log: `"ln\\s*\\("`
Both are culture-agnostic and have `IgnoreCase` set. I took the
opportunity to remove the "en-US" culture from the
`DivisionByZeroRegex`, as it is not required.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
Added unit tests which ensure the log and ln functions perform
identically with or without spaces. Confirmed all unit tests for Command
Palette and Run still pass.
Manually confirmed that `log` and `ln` now produce correct results in
both applications.
## Problem
The scheduled UI Test Automation pipeline
([Dart/161438](https://microsoft.visualstudio.com/Dart/_build?definitionId=161438))
has been failing daily since 2026-04-30 because PR #41280 (`.NET 10
Upgrade`) updated the main build template (`job-build-project.yml`) for
.NET 10 / VS 2026 but the parallel changes were missed in the UI Test
Automation templates.
Symptoms in recent runs (e.g. build
[#20260512.1](https://microsoft.visualstudio.com/Dart/_build/results?buildId=146803928)):
```
error NETSDK1045: The current .NET SDK does not support targeting .NET 10.0.
Either target .NET 9.0 or lower, or use a version of the .NET SDK that supports .NET 10.0.
```
…emitted ×179 across every csproj during the *Restore solution-level
NuGet packages* step on both `Build UI Tests Only Release_x64` and
`…_arm64` jobs.
## Changes
This PR mirrors the two pipeline changes that PR #41280 already applied
to the main CI templates.
**1. Install the .NET 10 SDK on the build agent**
- File: `.pipelines/v2/templates/job-build-ui-tests.yml`
- Bump the pinned `steps-ensure-dotnet-version.yml` parameter from
`version: '9.0'` → `'10.0'`.
- Matches `job-build-project.yml`, which already installs the .NET 10
SDK alongside 6.0/8.0.
**2. Pin the build agent image to VS 2026 (MSBuild 18)**
- File: `.pipelines/v2/templates/pipeline-ui-tests-official-build.yml`
- Add `demands: ImageOverride -equals SHINE-VS18-Latest` to the agent
pool spec.
- The .NET 10 SDK (≥ 10.0.300) requires MSBuild 18, which only ships
with VS 2026. Without this demand, the SHINE pool selects a default
image with VS 2022 / MSBuild 17 and restore fails with `MSB4236: The SDK
'Microsoft.NET.Sdk' specified could not be found`.
- Mirrors the unconditional pattern in `pipeline-ci-build.yml` (lines
56–67), which applies the same demand across both `SHINE-INT-L` and
`SHINE-OSS-L` pools.
## Validation
Queued pipeline 161438 against this branch:
| Build | Commit | Result |
| --- | --- | --- |
|
[`146821554`](https://microsoft.visualstudio.com/Dart/_build/results?buildId=146821554)
| `4dd13b9` (SDK fix only) | `NETSDK1045` gone ✅ — exposed VS 2022 /
MSBuild 17 mismatch |
|
[`146825355`](https://microsoft.visualstudio.com/Dart/_build/results?buildId=146825355)
| `97cadbb` (SDK + agent demand) | Both errors gone ✅ — agent now
`Visual Studio\18\Enterprise\` |
After this PR, the pipeline reaches NuGet restore and now hits a
separate, **pre-existing** issue from the April 13–29 failure window:
`401 Unauthorized` on the `shine-oss/PowerToysPublicDependencies`
Artifacts feed for newly-released `Microsoft.NETCore.App.*/8.0.27`
packages. That requires feed-admin action (granting the Dart pipeline
build identity "save from upstream" on the feed, or pre-seeding those
package versions) and is **out of scope for this PR**.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary
Two related changes to the `release-note-generation` agent skill:
### 1. Step 3 reviews: use the local agent instead of
`mcp_github_request_copilot_review`
Step 3.1 previously instructed the agent to call
`mcp_github_request_copilot_review` for every milestone PR so that the
`CopilotSummary` column in `sorted_prs.csv` would be populated by the
GitHub-side Copilot bot.
When this skill is driven from a CLI / coding agent, that request comes
from a bot identity, and the GitHub API rejects it (`Bot reviewers
cannot be requested`). The PR ends up with no Copilot review and
`CopilotSummary` stays empty.
`references/step3-review-grouping.md` has been rewritten to instead have
**the local agent** that is running the skill perform the review itself:
- Fetch each PR's diff with a non-mutating tool
(`mcp_github_pull_request_read` `get_diff` / `get_files`, or `gh pr
diff`).
- Produce a 1-3 sentence user-facing summary in the same style as a
Copilot PR review.
- Write the summary directly into the `CopilotSummary` column of
`Generated Files/ReleaseNotes/sorted_prs.csv`, preserving row order and
skipping rows that already have a non-empty summary.
Step 3.2 (re-run `dump-prs-since-commit.ps1`) is demoted to optional,
with a note that re-running the dump will overwrite the
locally-generated summaries.
`SKILL.md` was updated to match: front-matter description, "When to
Use", workflow diagram, the 3.1-3.3 row in the summary table,
prerequisites (no longer requires "GitHub Copilot code review enabled
for the org/repo"; mentions MCP for fetching diffs), and the
troubleshooting row for empty `CopilotSummary` now points at Step 3.1
with the bot-rejection caveat.
### 2. Vendor `prepare-release-assets.ps1` into the skill
Added `scripts/prepare-release-assets.ps1` -- previously kept in
OneDrive at `Tools/prepare-release.ps1`. Renamed because the script does
more than download installers: it also pulls per-arch symbol archives,
computes SHA256, and emits the **Installer Hashes** markdown table for
the GitHub release page. "Release assets" captures all of that.
Header `.SYNOPSIS` / `.DESCRIPTION` / `.EXAMPLE` blocks were updated to
reflect the new filename and the symbol-archive behavior (the original
synopsis only mentioned installers).
`SKILL.md` registers the new script in the "Available Scripts" table,
lists Azure CLI + the `azure-devops` extension as a prerequisite (only
when running this script), adds a "Prepare GitHub release assets" entry
to "When to Use", and adds a troubleshooting row for the most common
failure (`Failed to acquire ADO access token` -> `az login`).
## Files changed
| File | Change |
|------|--------|
| `.github/skills/release-note-generation/SKILL.md` | Updated
description, prerequisites, workflow, scripts table, troubleshooting |
|
`.github/skills/release-note-generation/references/step3-review-grouping.md`
| Rewritten to use local-agent review; demoted refresh step |
|
`.github/skills/release-note-generation/scripts/prepare-release-assets.ps1`
| New (vendored from OneDrive) |
## Validation
- PowerShell parser ([`Parser]::ParseFile`) reports no errors on the new
script.
- Documentation-only / scripts-only change -- no product code touched,
so the standard PowerToys build / test gates do not apply.
- The change preserves the existing CSV schema (`Id, Title, Labels,
Author, Url, Body, CopilotSummary, NeedThanks`), so downstream Step 3.3
(`group-prs-by-label.ps1`) and Step 4 summarization continue to work
without modification.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
1. Adds an opt-in Max compatibility mode in PowerDisplay's Advanced
settings. When enabled, DDC discovery probes monitors that don't
advertise capabilities, picking up displays that would otherwise be
skipped.
2. Toggling the setting triggers an immediate rescan via a new
RescanPowerDisplayMonitorsEvent IPC event from Settings to PowerDisplay.
3. Hides the brightness slider on monitors that lack VCP 0x10.
Also fixed animation issue #47868
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #47878
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Haiku 4.5 <noreply@anthropic.com>
This pull request introduces per-monitor dock customization support and
refactors how dock band settings are managed to enable independent
layouts on different monitors. The changes add a new
`DockMonitorConfigViewModel` for monitor-specific configuration, update
`DockViewModel` to handle per-monitor band lists and settings, and
refactor band movement and ordering logic to respect per-monitor
overrides.
**Per-monitor dock customization:**
* Added `DockMonitorConfigViewModel` to encapsulate the configuration
and state for each monitor, exposing properties for binding and
persisting changes using `ISettingsService`.
* Updated `DockViewModel` to track an optional `MonitorDeviceId`,
enabling docks to be associated with a specific monitor and to expose
per-monitor settings and methods.
[[1]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06L17-R33)
[[2]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06L41-R56)
**Band management refactor for per-monitor settings:**
* Refactored band retrieval and update logic in `DockViewModel` to use
new helper methods (`GetActiveBands`, `WithActiveBands`) that select and
modify either global or per-monitor band lists as appropriate.
* Updated band movement and ordering methods (`SyncBandPosition`,
`MoveBandWithoutSaving`, `SaveBandOrder`) to operate on the correct band
lists for each monitor, ensuring that changes apply to the intended
scope (global or per-monitor).
[[1]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06L205-R399)
[[2]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06L250-R413)
[[3]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06L263-R424)
[[4]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06L280-R437)
[[5]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06L290-R447)
[[6]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06L300-R465)
[[7]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06L329-R491)
**Resource management:**
* Implemented `IDisposable` on `DockViewModel` to clean up event
handlers and prevent resource leaks.
[[1]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06L17-R33)
[[2]](diffhunk://#diff-b661a9311de64dd1123860e858f1f4963f05ccee06b5bd218916635495b2ff06R85-R235)
Closes#46939
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: Niels Laute <niels.laute@live.nl>
Co-authored-by: Mike Griese <migrie@microsoft.com>
## Summary of the Pull Request
The system-wide CPU counter in the Command Palette Performance monitor
used `% Processor Utility`, which is scaled by `% Processor Performance`
and is intentionally unbounded above 100% when cores boost above their
nominal base frequency. Under load this produced values like 144% in the
dock, as reported in #46381.
Switch to `% Processor Time` — the same counter Task Manager renders —
which is naturally bounded to 0-100%.
The per-process branch is unaffected; it already divides by
`Environment.ProcessorCount` and continues to use `% Processor Time`, so
individual process readings remain correct.
## PR Checklist
- [x] **Closes:** #46381
- [x] **Communication:** I've discussed this with the team/maintainers
via the linked issue.
- [x] **Tests:** Manually tested locally. Built
`Microsoft.CmdPal.Ext.PerformanceMonitor` and the full
`CommandPalette.slnf` (Debug|x64). Verified the built DLL contains `%
Processor Time` and no longer contains `% Processor Utility`.
- [x] **Manual tests:** Ran the rebuilt dev build under sustained CPU
load; the system CPU figure now stays bounded to 100%.
- [x] **Localization:** Not applicable — counter name is an OS API
string, not user-visible text.
## Detailed Description of the Pull Request / Additional comments
`% Processor Utility` is documented by Microsoft as deliberately
unbounded; it represents an "effective" utilization that accounts for
turbo/boost frequencies. It is well-suited for billing/capacity contexts
but not for a "% of CPU used" UI element where users expect a 0-100%
reading consistent with Task Manager and Resource Monitor.
`_procPerformance` (the `% Processor Performance` counter) is retained
because it is still used by `CpuSpeed` to compute the live frequency
display, so removing it would regress the speed readout.
### Before
System CPU dock showed values >100% under load (reported 144% in the
issue).
### After
System CPU dock is bounded to 0-100%, matching Task Manager.
## Summary of the Pull Request
ZoomIt derives three recording hotkeys from one base key via XOR:
fullscreen (base), crop (base XOR Shift), window (base XOR Alt). When
Alt is the sole modifier, `base XOR Alt = 0`, registering a
modifier-less hotkey that captures every bare keypress (e.g., pressing
`5` triggers window recording).
**`Zoomit.cpp`** — 4 hotkey registration sites:
- Guard `RECORD_CROP_HOTKEY` registration behind `(g_RecordToggleMod ^
MOD_SHIFT) != 0`
- Guard `RECORD_WINDOW_HOTKEY` registration behind `(g_RecordToggleMod ^
MOD_ALT) != 0`
```cpp
// Before (all 4 sites):
registerHotkey( RECORD_CROP_HOTKEY, ( g_RecordToggleMod ^ MOD_SHIFT ) | MOD_NOREPEAT, ... );
registerHotkey( RECORD_WINDOW_HOTKEY, ( g_RecordToggleMod ^ MOD_ALT ) | MOD_NOREPEAT, ... );
// After:
if ( g_RecordToggleMod ^ MOD_SHIFT ) {
registerHotkey( RECORD_CROP_HOTKEY, ( g_RecordToggleMod ^ MOD_SHIFT ) | MOD_NOREPEAT, ... );
}
if ( g_RecordToggleMod ^ MOD_ALT ) {
registerHotkey( RECORD_WINDOW_HOTKEY, ( g_RecordToggleMod ^ MOD_ALT ) | MOD_NOREPEAT, ... );
}
```
**`ZoomItViewModel.cs`** — `RecordToggleKeyCrop` /
`RecordToggleKeyWindow` computed properties:
- Return `null` when the derived hotkey would have no modifier keys, so
the Settings UI omits the inapplicable shortcut description rather than
displaying a bare-key shortcut.
## PR Checklist
- [ ] Closes: #xxx
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
ZoomIt registers crop and window recording variants by XOR-ing the base
modifier with `MOD_SHIFT` and `MOD_ALT` respectively. This design breaks
when the base modifier equals the XOR target — the result is `0`, and
`RegisterHotKey` with no modifiers intercepts every bare keypress of
that key system-wide.
The fix is symmetric across all 4 registration sites (standalone
`registerHotkey()` in `RegisterAllHotkeys`, the legacy dialog OK path,
and two `WM_CREATE`-adjacent paths): skip the derived hotkey
registration when the computed modifier is zero. The Settings UI
ViewModel mirrors this logic by returning `null` for the affected
computed properties, causing the converter to emit an empty string
instead of a modifier-less shortcut label.
## Validation Steps Performed
- Code review confirmed fix is applied consistently across all 4
`RegisterHotKey` call sites in `Zoomit.cpp`
- Verified `HotkeySettingsToLocalizedStringConverter` returns
`string.Empty` for `null` input — no display regression in Settings UI
- Confirmed the default `Ctrl+5` hotkey is unaffected (`MOD_CONTROL ^
MOD_ALT = MOD_CONTROL | MOD_ALT ≠ 0`)
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: MuyuanMS <116717757+MuyuanMS@users.noreply.github.com>
Co-authored-by: Muyuan Li (from Dev Box) <muyuanli@microsoft.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Two FancyZones Editor strings have misleading Japanese translations due
to ambiguous English phrasing:
- **"Space around zones"** → 「ゾーン周りのスペース」 (generic space/room) — should
be 「ゾーン周りの余白」 (margin/padding)
- **"Highlight distance"** → 「距離を強調表示」 (distance to visually emphasize)
— should be 「隣接するゾーンの検知距離」 (detection distance for adjacent zones)
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
Changes to `FancyZonesEditor/Properties/Resources.resx`:
- **`Distance_adjacent_zones`** — Rewrote existing comment to explicitly
state this is a *detection/snapping proximity distance* (px at which a
nearby zone activates), not a visual-emphasis concept. Prevents
translators from reaching for 強調表示 ("visually highlight/emphasize").
- **`Space_Around_Zones`** — Added comment clarifying this is
*padding/margin* (empty gap in pixels) between zones, not a generic
space/area. Guides translators toward 余白 over スペース.
- **`Show_Space_Zones`** — Added comment clarifying this is a toggle for
enabling/disabling the padding display.
```xml
<data name="Distance_adjacent_zones" xml:space="preserve">
<value>Highlight distance</value>
<comment>The pixel distance at which an adjacent zone highlights (lights up) when a window is dragged near it. This is about detection/snapping range proximity, not about making something visually stand out.</comment>
</data>
<data name="Space_Around_Zones" xml:space="preserve">
<value>Space around zones</value>
<comment>The size (in pixels) of the padding or margin (empty gap) surrounding each zone. This is about blank space/whitespace between zones, not an area or region.</comment>
</data>
```
## Validation Steps Performed
Verified `Resources.resx` is well-formed XML and that the modified
entries render correctly in the resource file. No runtime behavior is
changed — comments are translator-only metadata consumed by the
Touchdown Build localization pipeline.
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: niels9001 <9866362+niels9001@users.noreply.github.com>
The Korean translation of "Activation modifier key"
(`GrabAndMove_ModifierKey.Header`) was rendered as "정품인증 보조키" ("product
authentication/genuine certification modifier key") — conflating feature
activation with Windows product licensing. The correct translation is
"활성화 보조 키".
## Change
- **`src/settings-ui/Settings.UI/Strings/en-us/Resources.resw`**:
Expanded the `<comment>` on `GrabAndMove_ModifierKey.Header` to
explicitly disambiguate "Activation" for translators:
```xml
<comment>Drop-down to choose which modifier key (Alt or Win) activates Grab And Move drag and resize.
"Activation" here means to enable/trigger the feature (활성화 in Korean),
NOT product authentication/genuine certification (정품인증 in Korean).</comment>
```
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
The word "Activation" is ambiguous in Korean: it can refer to Windows
product license activation (정품인증) or feature enablement (활성화). The
localization pipeline picked the wrong interpretation. Adding an
explicit in-source comment with both the correct and incorrect Korean
terms guides the translation tooling and human reviewers to use "활성화 보조
키" going forward.
## Validation Steps Performed
- Verified the updated comment appears correctly in the `.resw` file.
- No runtime behavior changes; comment-only modification to the resource
file.
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: niels9001 <9866362+niels9001@users.noreply.github.com>
The German locale translated the system tray tooltip `AppName` ("Power
Display") to "Leistungsanzeige", despite the Settings UI keeping the
utility name untranslated. The affected resource entries had no
translator guidance, so "Power Display" was treated as a localizable
phrase rather than a fixed product name. This PR adds consistent "do not
translate" comments across all touchpoints in both resource files.
## Summary of the Pull Request
- `src/modules/powerdisplay/PowerDisplay/Strings/en-us/Resources.resw`:
Added `Product name, do not translate.` comment to the `AppName` entry
(system tray tooltip)
- `src/settings-ui/Settings.UI/Strings/en-us/Resources.resw`: Added
`{Locked="Power Display"}` comments to all 15 entries containing "Power
Display" — including the module title, enable/toggle/open/launch
strings, OOBE title/description/activation text, LearnMore link,
QuickProfiles description, group header, and flyout header — following
the same convention already used for FancyZones and other product names
in that file
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [x] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
The system tray tooltip is populated via `GetString("AppName")` in
`TrayIconService.cs`. Because the resource had no comment, translators
localized "Power Display" as a common phrase ("performance display")
rather than treating it as a fixed product name.
To ensure consistency across all touchpoints, `{Locked="Power Display"}`
comments have been added to every entry in the Settings UI resource file
that contains the product name, matching the convention already in use
for `Shell_PowerDisplay.Content` (`Product name: Navigation view item
name for Power Display`) and for other utilities such as FancyZones
(`{Locked="FancyZones"}`).
## Validation Steps Performed
Verified that all `AppName` and Settings UI resource entries containing
"Power Display" now carry the appropriate translator comment. No runtime
behavior changes — the English values are unchanged; the comments solely
guide translators to leave the product name as-is in all locales.
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: niels9001 <9866362+niels9001@users.noreply.github.com>
The nittiest of nits: "No shortcuts to show.." message should be "No
shortcuts to show." when no modules are enabled.
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
This one is pretty self-explanatory.
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
(Manual testing.)
<img width="240" height="141" alt="image"
src="https://github.com/user-attachments/assets/a036d345-dc1d-4040-9111-187619453dcd"
/>
## Summary of the Pull Request
Adds the Mouse Without Borders "Refresh connections" action to both the
Quick Access tray flyout and the Settings Dashboard, so users can
trigger a reconnect without navigating into MWB settings.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
### Quick Access flyout (`Settings.UI.Controls`)
- **`QuickAccessViewModel`**: Added `MouseWithoutBorders` to
`InitializeItems()`; tooltip shows the configured `ReconnectShortcut`.
- **`QuickAccessLauncher`**: Added `case ModuleType.MouseWithoutBorders`
— signals `MWBReconnectEvent` (same named Windows event used by the
CmdPal `MWBReconnectCommand`).
### Settings Dashboard (`Settings.UI`)
- **`DashboardViewModel`**: Added `GetModuleItemsMouseWithoutBorders()`
returning a `DashboardModuleButtonItem` wired to a new
`MouseWithoutBordersReconnectClicked` handler that fires
`MWBReconnectEvent`. Added the module to the `GetModuleItems()` switch.
Added `using System.Threading` and `using PowerToys.Interop`.
No new strings — reuses existing
`MouseWithoutBorders_ReconnectButton.Text` ("Refresh connections") and
`MouseWithoutBorders_ReconnectTooltip.Text` for the button label and
description.
## Validation Steps Performed
- Verified `MWBReconnectEvent` is the same event used by
`MWBReconnectCommand` in the CmdPal extension.
- Confirmed `MouseWithoutBorders.png` icon exists in
`Assets/Settings/Icons/`.
- Confirmed `PowerToys.Interop` is already referenced by
`PowerToys.Settings.csproj`.
> [!WARNING]
>
> <details>
> <summary>Firewall rules blocked me from connecting to one or more
addresses (expand for details)</summary>
>
> #### I tried to connect to the following addresses, but was blocked by
firewall rules:
>
> - `o3svsblobprodcus318.vsblob.vsassets.io`
> - Triggering command: `/usr/bin/dotnet dotnet build
Settings.UI/PowerToys.Settings.csproj -c Debug` (dns block)
>
> If you need me to access, download, or install something from one of
these locations, you can either:
>
> - Configure [Actions setup
steps](https://gh.io/copilot/actions-setup-steps) to set up my
environment, which run before the firewall is enabled
> - Add the appropriate URLs or hosts to the custom allowlist in this
repository's [Copilot coding agent
settings](https://github.com/microsoft/PowerToys/settings/copilot/coding_agent)
(admins only)
>
> </details>
<!-- START COPILOT ORIGINAL PROMPT -->
<details>
<summary>Original prompt</summary>
>
> ----
>
> *This section details on the original issue you should resolve*
>
> <issue_title>Add Refresh Connections (from Mouse Without Borders) to
Quick Access</issue_title>
> <issue_description>### Description of the new feature / enhancement
>
> As per subject ... add Refresh Connections (_from Mouse Without
Borders_) to Quick Access
>
> ### Scenario when this would be used?
>
> There are a number of scenarios (_in my use case_) where I have one of
my machines connected to a business VPN .... which in 90% of the cases
triggers a small network drop resulting in the machine dropping off
mouse w/o borders - a refresh connections fixes it - but is currently a
bit annoying to get to.
>
> ### Supporting information
>
> _No response_</issue_description>
>
> ## Comments on the Issue (you are @copilot in this section)
>
> <comments>
> </comments>
>
</details>
<!-- START COPILOT CODING AGENT SUFFIX -->
- Fixesmicrosoft/PowerToys#45935
<!-- START COPILOT CODING AGENT TIPS -->
---
🔒 GitHub Advanced Security automatically protects Copilot coding agent
pull requests. You can protect all pull requests by enabling Advanced
Security for your repositories. [Learn more about Advanced
Security.](https://gh.io/cca-advanced-security)
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: niels9001 <9866362+niels9001@users.noreply.github.com>
Co-authored-by: Niels Laute <niels.laute@live.nl>
## Summary
Adds a GitHub Action workflow that **automatically applies `Product-*`
labels** to issues, reducing manual triage effort.
### How it works
**Two-tier approach:**
1. **Deterministic mapping** — Parses the structured "Area(s) with
issue?" dropdown from bug report templates and maps selections to the
correct `Product-` label via a hardcoded lookup table.
2. **AI inference (Copilot fallback)** — When no product is resolved
from the structured field (e.g., feature requests without the area
field), calls GitHub Models API (`gpt-4.1-mini`) to infer the product
from issue title + body.
### Trigger modes
| Trigger | Use case |
|---------|----------|
| `issues: [opened]` | Auto-labels every new issue |
| `workflow_dispatch` (single) | Test on one specific issue with dry-run
|
| `workflow_dispatch` (batch) | Process all open issues missing Product-
labels |
### Safety features
- **Label validation** — checks each label exists in the repo before
applying
- **Dry-run mode** — logs what would happen without modifying issues
- **Concurrency control** — prevents duplicate runs
- **Conservative AI prompt** — only labels products the issue is
*primarily* about
### Testing performed
Tested locally against 10 real issues with `Needs-Triage` and no
`Product-*` label:
- **6/10 resolved deterministically** (correct labels applied via `gh
issue edit`)
- **4/10 tested AI inference** via GitHub Models API:
- #47482 (CmdPal Dock) → `Product-Command Palette` ✅
- #47474 (grab and move + fancy zones) → `Product-FancyZones` ✅
- #47476 (modular download) → `[]` (correctly abstained) ✅
- #47478 (Quick Access pinning) → improved prompt to avoid over-labeling
### Files changed
| File | Purpose |
|------|---------|
| `.github/workflows/auto-label-product.yml` | The GitHub Action |
| `.github/policies/resourceManagement.yml` | Removed redundant
Workspaces-only regex rule |
| `tools/Test-AutoLabelProduct.ps1` | Local PowerShell test script for
dry-run testing |
### Mapping validated against actual repo labels
Confirmed all label names in the mapping exist in the repo via `gh label
list --search "Product-"`.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
Adds a GitHub Actions workflow that fires on every new/reopened issue
and uses `gpt-4o-mini` (via GitHub Models) to classify the issue and
apply the correct `Product-*` / `Area-*` label(s) automatically. Also
adds a `workflow_dispatch` path so maintainers can manually backfill
labels on existing untriaged issues by supplying a comma-separated list
of issue numbers.
## PR Checklist
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
**New file:** `.github/workflows/auto-label-issues.yml`
- **Trigger:** `issues: [opened, reopened]` for the automatic path;
`workflow_dispatch` with an `issue_numbers` input (comma-separated) for
manual backfill
- **Model call:** issue title + body (≤4 000 chars) → `gpt-4o-mini` at
the GitHub Models inference endpoint (`models.inference.ai.azure.com`);
uses `GITHUB_TOKEN` — no extra secrets required; `temperature: 0` for
deterministic output
- **Safety:** model output is filtered through a hard-coded allow-list
of 37 `Product-*` / `Area-*` labels before any API call — hallucinated
labels are dropped silently
- **Permissions:** `models: read` + `issues: write` only (same
minimal-permission pattern as the existing dedup workflow)
- **Concurrency:** per-issue group for automatic events; per-run-ID for
manual dispatch — prevents races without blocking unrelated runs
## Validation Steps Performed
- Manually triggered via `workflow_dispatch` against several untriaged
issues; step logs confirmed correct JSON output from the model and
matching labels applied via the GitHub Issues API.
- Verified the allow-list filter correctly discards any label not in the
predefined set.
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: niels9001 <9866362+niels9001@users.noreply.github.com>
Co-authored-by: Niels Laute <niels.laute@live.nl>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Adds an explicit Phase 0 classification step in `MonitorManager` that
uses `DISPLAYCONFIG_VIDEO_OUTPUT_TECHNOLOGY` (from `QueryDisplayConfig`)
to label every connected display as **internal** (built-in) or
**external** before any controller runs. Each controller is then
dispatched a strictly-scoped target list:
- **WMI controller** → only internal displays
- **DDC/CI controller** → only external displays
Two practical wins:
1. **Performance** — DDC/CI's ~4-second I2C capabilities probe (per
monitor) is now skipped entirely for internal laptop panels, which never
respond to DDC/CI in the first place. On a typical laptop with one
built-in panel + one external monitor, discovery is noticeably faster.
2. **Layering** — `WmiController` no longer reaches into the
`Drivers/DDC/` namespace to call
`DdcCiNative.GetAllMonitorDisplayInfo()`. Both controllers receive their
input from `MonitorManager` via a single `QueryDisplayConfig` call.
Strict classification is enforced: a display classified as internal but
not returned by `WmiMonitorBrightness` is dropped + logged (Warning),
with **no fallback to DDC/CI**. This is a deliberate design choice — the
spec discusses the trade-off in detail.
Adds a Phase 0 classification log (Info level) so misclassifications are
diagnosable from logs alone:
```
[DisplayClassification] Found 2 displays:
[Path 1] \\.\DISPLAY1 / "Built-in display": OutputTechnology=0x80000000 → Internal
[Path 2] \\.\DISPLAY2 / "Dell U2723QE": OutputTechnology=10 → External
[DisplayClassification] Summary: 1 internal, 1 external
```
The classification rule (in `DisplayClassifier.IsInternal`) is
deliberately conservative — misclassifying an external display as
internal would silently drop it from DDC/CI discovery with no fallback,
so we err on the side of external:
- **Internal**: the bare `INTERNAL` flag (`0x80000000`) alone, the
`INTERNAL` flag combined with a documented embedded subtype
(`DISPLAYPORT_EMBEDDED` 11 or `UDI_EMBEDDED` 13), or one of those
embedded subtypes on its own
- **External**: everything else, including the `INTERNAL` flag combined
with an undocumented subtype (HDMI, DP_EXTERNAL, MIRACAST, etc.)
- LVDS (6) is intentionally **not** classified internal — the [Microsoft
docs](https://learn.microsoft.com/en-us/windows/win32/api/wingdi/ne-wingdi-displayconfig_video_output_technology)
describe it only as a connector type, not as an internal-display marker
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Haiku 4.5 <noreply@anthropic.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Adding missing Grab And Move and Power Display events
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
## Summary of the Pull Request
Bumps `Microsoft.WindowsAppSDK` (and split sub-packages) from
**`1.8.260209005`** to public NuGet''s **`2.0.1`** GA, plus the
centrally-pinned `Microsoft.Web.WebView2` to **`1.0.3719.77`** to
satisfy the new transitive requirement from
`Microsoft.WindowsAppSDK.WinUI 2.0.12`.
This is the minimum mechanical version-bump set required for the
WinAppSDK 2.0 upgrade ΓÇö no API migration or behavioral changes.
Subsequent commits on this branch will address breaking-change
migrations as they surface.
NuGet source-of-truth used to pick sub-package versions: WinAppSDK 2.0.1
catalog `dependencyGroup`
(https://api.nuget.org/v3/catalog0/data/2026.04.29.22.25.36/microsoft.windowsappsdk.2.0.1.json).
## Detailed Description of the Pull Request / Additional comments
### Scope
| Package | Pinned |
|---|---|
| `Microsoft.WindowsAppSDK` | `2.0.1` |
| `Microsoft.WindowsAppSDK.Foundation` | `2.0.20` |
| `Microsoft.WindowsAppSDK.AI` | `2.0.185` |
| `Microsoft.WindowsAppSDK.Runtime` | `2.0.1` |
| `Microsoft.Web.WebView2` | `1.0.3719.77` |
## Validation Steps Performed
`tools\build\build-essentials.cmd` results on the temp branch (x64
Debug, VS 2026 Insiders 14.51.36231):
| Stage | Errors | Warnings | Time |
|---|---|---|---|
| `PowerToys.slnx` NuGet restore | **0** | 0 | 1:05 |
| Native C++ `src\runner\runner.vcxproj` | **0** | 0 | 4:08 |
| Managed `src\settings-ui\Settings.UI\PowerToys.Settings.csproj` |
**0** | 0 | 3:35 |
Exit code: **0**. The end-to-end `build-essentials` flow passes locally
ΓÇö the WinAppSDK 2.0 upgrade is verified to compile against the runner
+ Settings.UI projects.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
The brightness / contrast / volume sliders in the PowerDisplay flyout
previously committed to hardware (DDC/CI) only on `PointerCaptureLost`
plus arrow-key `KeyUp`. That doesn't work reliably on touchscreens —
tap-on-track issues no pointer capture, and the Slider / Thumb doesn't
always bubble `PointerCaptureLost` on touch — so the value the user
dialed in never reached the monitor.
This PR replaces that wiring with the simpler approach we actually
wanted: a TwoWay `Value` binding plus a debounced commit in the
view-model. `ValueChanged` runs the setter on every move, but a 200 ms
`DispatcherQueueTimer` (restarted on each change) collapses the entire
interaction into a single DDC/CI write once the user stops sliding.
Mouse, touch, tap-on-track, and held arrow keys all use the same code
path.
It also adds a small `SliderExtensions` helper (Toolkit-style attached
properties) so the brightness / contrast / volume sliders can be
adjusted by mouse-wheel scroll. The wheel event is marked handled so the
parent `ScrollViewer` does not also scroll, and the existing 200 ms
debounce coalesces wheel input into a single DDC/CI write.
## PR Checklist
- [x] Closes: #47724
- [x] **Communication:** I've discussed this with core contributors
already.
- [x] **Tests:** Manual smoke; no automated coverage exists for the
slider flyout.
- [x] **Localization:** No new end-user-facing strings.
- [x] **Dev docs:** N/A
- [x] **New binaries:** N/A
- [x] **Documentation updated:** N/A
## Detailed Description of the Pull Request / Additional comments
### Debounced commit
- `MonitorViewModel.Brightness/Contrast/Volume` setters now: assign
field → `OnPropertyChanged()` → `ScheduleCommit(...)` which (re)starts a
per-metric `DispatcherQueueTimer`. On `Tick` the closure calls
`SetBrightnessAsync(_brightness)` (etc.), reading the latest field value
so intermediate drag values are coalesced into one hardware write.
- `SetBrightnessAsync` / `SetContrastAsync` / `SetVolumeAsync` (used by
hotkeys, profile load, hardware refresh) keep their immediate-apply
behavior — they bypass the public setters, so a TwoWay source-update
from one of those paths does **not** reschedule a commit.
- `MonitorViewModel.Dispose()` stops the three timers. Pending writes
are intentionally dropped: `Dispose` only runs when the monitor is gone
(unplug / refresh) or the app is shutting down, in which case a hardware
write would race a half-torn-down `MonitorManager`.
- `AppConstants.UI.SliderCommitDebounceMs = 200` is the single source of
truth for the delay.
- Removed the six
`Handle{Brightness|Contrast|Volume}{PointerCaptureLost|KeyUp}` methods
and the now-unused `IsArrowKey` helper.
- `using DispatcherQueueTimer =
Microsoft.UI.Dispatching.DispatcherQueueTimer;` alias avoids ambiguity
with `Windows.System.DispatcherQueueTimer`.
### Wheel-scrollable sliders
- New `PowerDisplay.Helpers.SliderExtensions` — a Toolkit-style `public
static class <Control>Extensions` with two attached properties:
- `SliderExtensions.IsMouseWheelEnabled` (bool) — opt in to wheel input.
- `SliderExtensions.MouseWheelChange` (double, default `NaN` → falls
back to the slider's own `Slider.SmallChange`) — per-notch delta.
- Handler computes `notches = MouseWheelDelta / 120`, clamps to
`[Minimum, Maximum]`, sets `Value`, and marks `e.Handled = true` so the
enclosing `MainScrollViewer` doesn't also scroll.
- Wired on the brightness, contrast, and volume sliders with
`MouseWheelChange="5"` (20 notches for a full sweep).
- Body has zero PowerDisplay-specific dependencies (only
`Microsoft.UI.Xaml.*`) so the helper is straightforward to lift into the
WinUI Community Toolkit later.
## Validation Steps Performed
- `tools/build/build.cmd` against
`src/modules/powerdisplay/PowerDisplay/PowerDisplay.csproj` — exit code
0.
- Manual smoke on a touchscreen device:
- Touch drag thumb 50 → 80, lift finger → monitor commits once at 80.
- Tap on the track at 25 % → monitor commits once at 25 %.
- Manual smoke with mouse:
- Drag 50 → 80, release → exactly one DDC/CI write at 80.
- Manual smoke with keyboard:
- Hold Right arrow on the brightness slider → repeated nudges collapse
into a single write 200 ms after key release.
- Manual smoke with mouse wheel:
- Hover brightness slider, scroll wheel → value moves by 5 per notch,
monitor commits once after scrolling stops.
- Wheel over a slider does **not** scroll the flyout's
`MainScrollViewer`; wheel outside the sliders still scrolls.
- Wheel past `Minimum` / `Maximum` clamps at the boundary.
- Hotkey-driven brightness change still applies immediately (regression
check on the `SetBrightnessAsync` path).
- `DisplayChangeWatcher`-driven hardware refresh does not schedule a
spurious commit.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This PR solves the following issues:
- Narrator announced the Resize button as 'button', now it's 'resize'.
- The app window was still 'WinUI Desktop', it's now 'Image Resizer'
Binding is hard.
The tooltip is bound to `.Tooltip`. But we forgot to raise a property
changed event for `Tooltip` when title or subtitle change.
Closes: issue filed on teams
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Fixes#47458 — on monitors that advertise a non-100 raw VCP maximum
(e.g. Samsung S27DG602SN reports 50), the DDC controller was writing the
slider's 0–100 percentage directly as the raw VCP value. Anything above
the device's max was silently clamped, so only the bottom half of the
brightness / contrast / volume sliders had any visible effect.
The same root cause affects all three continuous-range VCPs (0x10
brightness, 0x12 contrast, 0x62 volume).
- Capture the device-reported `max` at discovery into new
`Monitor.{Brightness,Contrast,Volume}VcpMax` fields.
- Add `VcpFeatureValue.FromPercentage(percent, max, min)` as the
symmetric inverse of the existing `ToPercentage()`.
- Scale percent → raw inside `DdcCiController.Set{Brightness,Contrast,
Volume}Async` before each `SetVCPFeature` call.
- WMI is unaffected — Windows already standardises WMI brightness to
0–100.
The whole stack above the controller (ViewModel, MonitorManager, slider,
settings persistence, IPC) stays in the percent domain. The only place
percent ↔ raw conversion happens is the controller boundary.
#47458
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Fixes#47665 — users report the per-monitor "Show input source control"
/ "Show power state control" toggles silently revert to off over time.
Three independent log captures pin the root cause to two layers in
`MainViewModel`:
- **Data-loss layer**: `SaveMonitorsToSettings` rebuilds
`settings.Properties.Monitors` from the currently-discovered list and
only re-adds entries with `IsHidden=true`. Whenever monitor discovery
transiently fails (DDC `GetPhysicalMonitors` NULL handle, DDC `empty
capabilities string`, WMI 4200 — all common in production logs),
affected monitors get **deleted** from settings.json. Next successful
discovery initialises them with the post-#47303 defaults
(`EnableInputSource=false`, `EnableColorTemperature=false`,
`EnablePowerState=false`), and `ApplyPreservedUserSettings` cannot
recover what is no longer there.
- **Identity layer**: `Monitor.Id` is
`{Source}_{EdidId}_{MonitorNumber}` — `EdidId` is not unique for
identical-model monitors, and `MonitorNumber` is OS-assigned and changes
after sleep/wake / GPU reset / display reorder. Even fixing the
data-loss layer would still apply preserved settings to the wrong
physical monitor for users with multiple identical displays.
This PR addresses both layers:
- **30-day retention** — `MonitorSettingsRebuilder` (in
`PowerDisplay.Lib`) replaces the inline `IsHidden`-only loop.
Currently-discovered monitors get a fresh `LastSeenUtc` stamp;
missing-but-recent (< 30 days) entries are preserved with all `Enable*`
flags intact; missing-and-stale entries are dropped with a single info
log. `IsHidden=true` entries are still preserved unconditionally.
- **DevicePath-based Id** — `Monitor.Id` is now the Windows `DevicePath`
minus the trailing device-class GUID (e.g.
`\\?\DISPLAY#DELD1A8#5&abc&0&UID12345`). The middle PnP-instance segment
is unique per (physical device × physical port) and stable across
reboots, sleep/wake, and OS-level reordering —
`MonitorDisplayInfo.DevicePath` already carries this value from
`QueryDisplayConfig`; it just wasn't being used as the Id.
The order of commits keeps the app behaviourally identical to today
through the first 8 commits (only adds dormant helpers / fields /
wiring); the actual ID-format flip is the very last commit.
Bisect-friendly.
#47599
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [x] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: Yu Leng <yuleng@microsoft.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
Exposes the 16:9 aspect ratio toggle for the ZoomIt screen region
recording (default: `Ctrl`+`Shift`+`5`) in the settings UI.
<img width="1047" height="817" alt="image"
src="https://github.com/user-attachments/assets/4117a6a1-9f57-44c5-9a21-1b1b65bc2992"
/>
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
Checked that the toggle works and that it's compatible with the
standalone ZoomIt, too.
## Summary
Fixes#47645.
The ADMX policy definition (`src/gpo/assets/PowerToys.admx`) declares
`revision="1.20"` and `<resources minRequiredRevision="1.20"/>`, but the
en-US ADML language resource (`src/gpo/assets/en-US/PowerToys.adml`) was
still at `revision="1.19"`. Group Policy Editor refuses to load the
PowerToys template because the language resource does not meet the
minimum required revision, producing the error reported in the issue.
## Change
Bump `revision` on the `policyDefinitionResources` root element of
`PowerToys.adml` from `1.19` to `1.20` so it matches the ADMX
`minRequiredRevision`.
## Validation
- Diffed against `src/gpo/assets/PowerToys.admx` to confirm the
revisions now align (both `1.20`).
- Documentation/asset-only change; no code build required.
## PR Checklist
- [x] Closes#47645
- [x] Tests added/passed — N/A (asset-only revision bump)
- [x] Requires documentation to be updated — No
- [x] I know how to test these changes — Open `gpedit.msc` and confirm
PowerToys policies load without the resource-version error.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
ZoomItBreak.vcxproj contained a leftover Import of
Microsoft.Windows.ImplementationLibrary.targets pinned to version
1.0.231216.1, plus the matching EnsureNuGetPackageBuildImports guard.
The project has no packages.config and its source files
(BreakTimer.cpp/.h, ZoomItBreakScr.cpp) include no wil headers, so the
import was dead code copy-pasted from ZoomIt.vcxproj when the
screensaver project was introduced in #46506.
Until PR #41280 (.NET 10 upgrade), the sibling ZoomIt project restored
WIL 1.0.231216.1 via its own packages.config, so the folder
coincidentally existed and the stale Import silently succeeded. PR
#41280 bumped that dependency to 1.0.260126.7, leaving nothing in the
repo that requests the old version. NuGet no longer creates the
1.0.231216.1 folder, the EnsureNuGetPackage target fires, and the
official build fails with:
ZoomItBreak.vcxproj(227,5): Error : This project references
NuGet package(s) that are missing on this computer ...
Microsoft.Windows.ImplementationLibrary.1.0.231216.1\...targets
Fix: drop the unused import and guard target. Verified by clean rebuild
of x64 Release and ARM64 Release; ZoomItBreak64.scr / ZoomItBreak64a.scr
produced with empty error logs.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
This adds the capability to include a webcam in the ZoomIt recording.
Also, the trim editor now supports appending multiple clips with
selectable transitions.
<img width="451" height="570" alt="image"
src="https://github.com/user-attachments/assets/025a7840-2e40-424c-af57-5ed523af8646"
/>
Also fixed some minor bugs in the ZoomIt settings within PowerToys and
added the options there, too.
<img width="1556" height="962" alt="image"
src="https://github.com/user-attachments/assets/dfe4209c-1b59-47c3-9170-36832d37f880"
/>
There was a bug in the microphone selection, fixed it together with the
webcam selection dialog, too.
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [x] Closes: #47230
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
---------
Co-authored-by: markrussinovich <markrussinovich@users.noreply.github.com>
Co-authored-by: Copilot <copilot@github.com>
Updated instructions for downloading the .exe file and release notes
link.
<!-- Enter a brief description/summary of your PR here. What does it
fix/what does it change/how was it tested (even manually, if necessary)?
-->
## Summary of the Pull Request
<!-- Please review the items on the PR checklist before submitting-->
## PR Checklist
- [ ] Closes: #xxx
<!-- - [ ] Closes: #yyy (add separate lines for additional resolved
issues) -->
- [ ] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
<!-- Provide a more detailed description of the PR, other things fixed,
or any additional comments/features here -->
## Detailed Description of the Pull Request / Additional comments
<!-- Describe how you validated the behavior. Add automated tests
wherever possible, but list manual validation steps taken as well -->
## Validation Steps Performed
## Summary of the Pull Request
Fixes `tools\build\build-common.ps1` so a clean
`tools\build\build-essentials.cmd` (or `build.ps1`) run discovers and
uses **Visual Studio 2026 Insiders** without any manual
`Enter-VsDevShell` preamble. Today the script lands on the first VS 2022
BuildTools instance it finds (which lacks the C++ workload) and every
native `.vcxproj` errors with `MSB4086: $(PlatformToolsetVersion)
evaluates to ""` during NuGet restore.
Two scoped commits, **only `tools\build\build-common.ps1` is touched**
(+58 / −36 net).
## PR Checklist
- [ ] Closes: #xxx
- [x] **Communication:** I''ve discussed this with core contributors
already. If the work hasn''t been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places (n/a — build-script
change only)
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
### What''s broken
`build-common.ps1`''s `vswhere` lookup runs **without `-prerelease`**,
so VS 2026 Insiders / Preview installs are invisible to it. The explicit
fallback path list also only contains VS 2022 entries. On a machine that
has VS 2022 BuildTools (typical for CI hosts and many dev boxes) but
only has VS 2026 in *Insiders* form, the script:
1. Picks `C:\Program Files (x86)\Microsoft Visual
Studio\2022\BuildTools` because it''s a candidate `vswhere` returned and
it''s in the fallback list.
2. Enters its DevShell — but BuildTools has no C++ workload installed,
so `$(VCToolsInstallDir)` and `$(PlatformToolsetVersion)` are unset.
3. NuGet restore over `PowerToys.slnx` fans out to every `.vcxproj` and
they each hit `Microsoft.CodeAnalysis.targets(401,15): error MSB4086: A
numeric comparison was attempted on "$(PlatformToolsetVersion)" that
evaluates to "" instead of a number, in condition
"''$(PlatformToolsetVersion)''<''120''".` The build dies before any
project compiles.
This was reproduced today against `origin/main` — confirming this is a
pre-existing breakage independent of any in-flight feature work.
### What this PR changes
Commit **`30acf72c` — Fix build scripts to discover VS 2026 / Insiders
installations**
- Adds `-prerelease` to `vswhere` calls, tried **before** the stable
lookup so prerelease VS is preferred when it''s the only one with a
working C++ workload.
- Adds VS 2026 year-name and internal-version (`18\Insiders`) paths to
the explicit fallback list.
- Keeps VS 2022 paths as the final fallback so existing setups keep
working.
- Priority order: `prerelease+VC tools` → `prerelease` → `stable+VC
tools` → `stable`.
Commit **`18b27209` — build: simplify VS environment initialization with
VS2022/VS2026 support**
- Adds `-prerelease` to `vswhere` (consolidates with the above; the
prerelease query subsumes the stable one now).
- Restricts the SKU query to `Community` / `Professional` / `Enterprise`
(explicitly excludes `BuildTools`) so the script can no longer
accidentally pick a SKU without the C++ workload.
- Reduces vswhere from **4 calls to 2**.
- Removes the destructive BuildTools env-var cleanup block (no longer
needed once vswhere refuses to return BuildTools in the first place).
- Adds `VsDevCmd.bat` exit-code validation so a partial DevShell init
fails loudly instead of silently producing a half-initialized
environment.
- Adds VS 2022 / VS 2026 Preview entries to the explicit fallback list.
### Why two commits instead of a squash
The first commit is the minimum-viable fix that addresses the reported
MSB4086 failure. The second commit is a follow-up cleanup that''s only
safe **once** prerelease is preferred and BuildTools is excluded.
Splitting them keeps each commit independently revert-able if a
regression shows up on a specific dev environment shape.
## Validation Steps Performed
| Step | Result |
|---|---|
| Reproduce on `main`: cold `tools\build\build-essentials.cmd` in a
non-DevShell PowerShell | **MSB4086** on
`Microsoft.CommandPalette.Extensions.vcxproj`,
`Microsoft.Terminal.UI.vcxproj`, `PowerToys.MeasureToolCore.vcxproj`,
`PowerRenameUI.vcxproj` — confirmed broken |
| With this branch checked out: same cold `build-essentials.cmd`
invocation | `[VS] vswhere found: ... C:\Program Files (x86)\Microsoft
Visual Studio\Installer\vswhere.exe` → `[VS] Checking candidate:
C:\Program Files\Microsoft Visual Studio\18\Insiders` → `[VS] Entered
Visual Studio DevShell at C:\Program Files\Microsoft Visual
Studio\18\Insiders` — VS 2026 Insiders selected directly, restore step
proceeds past the MSB4086 wall |
| Manual workaround pre-init via `Import-Module
Microsoft.VisualStudio.DevShell.dll` + `Enter-VsDevShell` | Now
**unnecessary** — the script self-initializes correctly |
After this fix, the next failure mode the build hits is unrelated NuGet
feed coverage for in-flight WinAppSDK upgrade work, which is the gated
dependency this PR was extracted from to keep this change atomic.
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
## Summary of the Pull Request
.NET 10 Upgrade. Requires Visual Studio 2026.
## PR Checklist
- [x] **Communication:** I've discussed this with core contributors
already. If the work hasn't been agreed, this work might be rejected
- [ ] **Tests:** Added/updated and all pass
- [ ] **Localization:** All end-user-facing strings can be localized
- [ ] **Dev docs:** Added/updated
- [ ] **New binaries:** Added on the required places
- [ ] [JSON for
signing](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ESRPSigning_core.json)
for new binaries
- [ ] [WXS for
installer](https://github.com/microsoft/PowerToys/blob/main/installer/PowerToysSetup/Product.wxs)
for new binaries and localization folder
- [ ] [YML for CI
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/ci/templates/build-powertoys-steps.yml)
for new test projects
- [ ] [YML for signed
pipeline](https://github.com/microsoft/PowerToys/blob/main/.pipelines/release.yml)
- [ ] **Documentation updated:** If checked, please file a pull request
on [our docs
repo](https://github.com/MicrosoftDocs/windows-uwp/tree/docs/hub/powertoys)
and link it here: #xxx
## Detailed Description of the Pull Request / Additional comments
- Upgraded target framework from `net9.0` to `net10.0` across all
projects
- Removed redundant package references now included by default in .NET
10
- Updated package versions to .NET 10 releases
- Modernized regex usage with source generators for better performance
- Added `vbcscompiler` to the spell-check allowlist
(`.github/actions/spell-check/expect.txt`)
## Validation Steps Performed
<!-- START COPILOT CODING AGENT TIPS -->
---
🔒 GitHub Advanced Security automatically protects Copilot coding agent
pull requests. You can protect all pull requests by enabling Advanced
Security for your repositories. [Learn more about Advanced
Security.](https://gh.io/cca-advanced-security)
---------
Co-authored-by: Jeroen van Warmerdam <jeronevw@hotmail.com>
Co-authored-by: Copilot <copilot@github.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
The influx of Copilot-originated pull requests has made it impossible
for our CI to keep up; this is in part because for some reason Copilot
is treated as a team member and therefore automatically scheduled for CI
on the build pool.
2026-04-29 13:27:41 -05:00
1207 changed files with 383350 additions and 16699 deletions
When creating or editing Shortcut Guide keyboard shortcut manifest files, follow the schema and naming conventions in the spec:
- [WinGet Manifest Keyboard Shortcuts schema](<../doc/specs/WinGet Manifest Keyboard Shortcuts schema.md>) – manifest file format, field definitions, file naming, and the `+` prefix convention for apps without a WinGet package
reply:Hi! We've identified this issue as a duplicate of another one that already exists on this Issue Tracker. This specific instance is being closed in favor of tracking the concern over on the referenced thread. Thanks for your report!
reply:We've identified this issue as a duplicate of an existing one and are closing this thread so discussion stays in one place.<br/><br/>Please see the comment above for the link to the original tracking issue, and feel free to subscribe there for updates.
- closeIssue
- removeLabel:
label:Needs-Triage
@@ -266,16 +266,5 @@ configuration:
- addReply:
reply:Hi! Your last comment indicates to our system, that you might want to contribute to this feature/fix this bug. Thank you! Please make us aware on our ["Would you like to contribute to PowerToys?" thread](https://github.com/microsoft/PowerToys/issues/28769), as we don't see all the comments. <br /><br />_I'm a bot (beep!) so please excuse any mistakes I may make_
Thank you for contributing to PowerToys. We've detected that this PR might include a new or modified telemetry event. After this PR is merged, please follow these next steps:
- [ ] Reach out to Jessica (${REVIEWER_MENTION}) to follow up on the next steps: https://aka.ms/pt-telemetry-process
Thank you for contributing to PowerToys. We've detected that this PR might include a new or modified telemetry event. Please ensure the following before merging:
- [ ] Add your telemetry events to [DATA_AND_PRIVACY](https://github.com/microsoft/PowerToys/blob/main/DATA_AND_PRIVACY.md).md within this PR.
- [ ] Reach out to Jessica (${REVIEWER_MENTION}) to follow up on the next steps: https://aka.ms/pt-telemetry-process`;
description: "Verify a single PowerToys module's release checklist items end-to-end. Drive each checkbox via UIA / Named Events / settings.json edits / clipboard inspection / GPO / SendInput. Output a structured PASS / FAIL / BLOCKED verdict per item with evidence (FAIL distinguishes product defects from stale/ambiguous checklist items). Combine standard winapp ui mechanics (see references/winapp-ui-testing.md) with PT-specific recipes and the helper .ps1 files shipped with this skill."
license: Complete terms in LICENSE.txt
---
## When to use this skill
Use this skill when you need to **verify every checklist item for a single PowerToys module** for a release sign-off — e.g. "verify all 18 Color Picker items", "verify all 88 Command Palette items". Each item produces a PASS / FAIL / BLOCKED verdict with evidence (UIA enumeration, log line, settings.json diff, screenshot, etc.).
The **checklist to verify is supplied with the task** (the calling prompt points you at the module's checklist file). This skill is the *how* — the drive techniques, helpers, taxonomy, and reporting format — independent of any specific checklist.
## Required reads (in order)
1.**`references/winapp-ui-testing.md`** — the **prerequisite** UIA mechanics doc (winapp ui verbs, scripted batch testing, file pickers, accessibility audits, screenshots, click-vs-invoke, PostMessage, SendInput cb=40, stunted-UIA recovery, settings-mutation safety contract). **Read this first** — this skill assumes you know its content and only adds PT-specific extensions.
2.**This `SKILL.md`** — the PT-specific playbook: the 3-bucket drive-technique selector (Step 2), classification taxonomy, critical pitfalls, helper-script catalog.
3.**`references/modules/<module>.md` IF IT EXISTS** — per-module entry-paths, item-by-item recipes, common BLOCKED traps, fixture lists, source citations. **Always check `references/modules/` first.** If no profile exists, fall back to this SKILL.md and create one after you finish (template in `references/modules/README.md`).
4.**`references/explorer-context-menu-flow.md` IF your module registers an Explorer right-click entry** (PowerRename, File Locksmith, Image Resizer, New+, Preview Pane, RegistryPreview) — shared synthetic-right-click + UIA-invoke + multi-file-selection flow + module-caption table. Helper: `scripts/pt-explorer-contextmenu.ps1`.
5.**`references/pre-flight.md`** — pre-flight checks, bootstrap snippet, state-hygiene cleanup, final wrap-up, hard rules.
7.**`references/environment-setup.md`** — RDP/sleep/screensaver/session-attachment gotchas. Cite in BLK-ENV verdicts.
8.**`references/release-checklist/<module>.md`** — the checklist for the module under test (one file per module; see `references/release-checklist/index.md` for the full list). Each item carries `[ADMIN: …]` + `[CLARITY: …]` metadata. **This file IS the set of items to verify.**
## Helper scripts shipped with this skill
| File | Purpose |
|---|---|
| `scripts/pt-shared-events.ps1` | `Invoke-PtSharedEvent`, `Test-PtSharedEvent`, `Get-PtSharedEventCatalog` — 56-entry friendly-name map for PT Named Events (CmdPal.Show, AOT.Pin, PowerLauncher.Invoke, LightSwitch.Toggle, ZoomIt.Draw, ...). The deterministic, foreground-free, UIPI-immune way to trigger a module. |
| `scripts/pt-sendinput-chord.ps1` | `Send-PtChord`, `Wait-PtHotkeyAccepted` — last-resort SendInput hotkey injection with the cb=40 fix. Use only when the module has no Named Event and the hotkey itself is the test subject. |
| `scripts/pt-foreground-guard.ps1` | `Test-PtForeground`, `Force-PtForeground`, `Assert-PtForegroundOrAbort` — guard helpers to ensure target window IS foreground before SendInput, so keys don't leak to caller's terminal. |
| `scripts/pt-admin-probe.ps1` | `Test-PtAdmin`, `Test-ProcessElevated`, `Test-PtRunnerAdmin` — TokenElevation probes to verify your session and the PT runner have the right elevation for the test. |
| `scripts/pt-explorer-com.ps1` | `Get-PtExplorerWindows`, `Open-PtExplorerAtPath`, `Select-PtExplorerFiles`, `Invoke-PtPeekWithExplorerSelection`, `Test-PtInteractiveDesktop` — drive Explorer via Shell COM to set up multi-file selections, then trigger Peek/FZ/PowerRename/Image Resizer/Workspaces via their hotkeys. **Use this for Peek L706-L709, L719-L720 and any test that needs an Explorer file selection.** |
| `scripts/pt-explorer-contextmenu.ps1` | `Test-PtDesktopInteractive`, `Open-PtExplorerContextMenu`, `Invoke-PtContextMenuItem`, `Get-PtContextMenuItems` — open Win11's real context menu via synthetic right-click (with retry), then UIA-invoke a menu item by name. **Canonical user-flow path for File Locksmith / Image Resizer / PowerRename / New+ menu-presence + launch tests.** Needs an unlocked interactive desktop. See `references/explorer-context-menu-flow.md` for the full write-up, stability notes, and per-module captions. |
| `scripts/pt-shell-verbs.ps1` | `Get-PtShellVerbs`, `Invoke-PtShellVerb`, `Reset-PtShellComCache` — enumerate + invoke CLASSIC HKCR shell verbs via Shell.Application COM. **NOT for PT context-menu modules on Win11** (PT registers via IExplorerCommand, not classic — use `pt-explorer-contextmenu.ps1` for those). Useful for non-PT verbs (Open/Edit/Send-to/third-party) and as a negative check that PT verbs are NOT classic-shadowed. |
| `scripts/pt-state.ps1` | `Get-PtSettings`, `Get-PtModuleSettings`, `Get-CmdPalSettings`, `Get-PtRunnerLogTail`, `Test-PtModuleEnabled`, `Test-PtModuleProcess`, `Restart-PtRunner`, `Backup-PtModuleSettings`, `Restore-PtModuleSettings` — common state checks. |
| `scripts/pt-nonelevated.ps1` | `Start-PtNonElevated`, `Invoke-PtNonElevatedCapture` — launch an exe at **Medium IL (non-elevated)** from an elevated agent shell via a one-shot `RunLevel Limited` scheduled task. Required for elevation-visibility tests (a non-elevated module must NOT see higher-integrity processes; e.g. File Locksmith L649/L650). Verify the result with `Test-ProcessElevated`. |
Dot-source them **all** at once in your bootstrap (the `Get-ChildItem` loop loads every helper — see **Step 1 — Bootstrap**):
```powershell
$skill='<this skill folder>'# the folder containing SKILL.md, e.g. <PT-repo>\.github\skills\powertoys-module-verification
# The checklist items to verify are supplied with the task (see the calling prompt).
# Read that module's checklist file and iterate its items (see Step 6 — Verifier loop).
```
## Step 2 — Drive techniques
Every checklist item boils down to ONE of three intents. **Pick the bucket from the verb in the item, then use the best technique inside it.** Stop at the first technique that works.
| Intent | Verb-cues in the checklist item | Bucket |
|---|---|---|
| Change a setting | "default is X", "setting persists", "is enabled/disabled by default", "value Y is accepted" | §2.A |
| Interact with a UI element | "click X", "toggle X", "type into Y", "X button is visible", "selecting Z does W" | §2.B |
| Trigger a module action | "pressing hotkey X opens Y", "module launches", "Z happens when invoked" | §2.C |
### §2.A — Change a setting (single technique)
Edit the JSON file the module reads, wait for the file-watcher debounce, assert, then restore from backup. Zero external tools.
> For shell-extension modules (PowerRename, File Locksmith, Image Resizer, New+) edit the **module-owned** file under `%LOCALAPPDATA%\Microsoft\PowerToys\<Module>\`, then `Restart-PtRunner` (and on stubborn handlers, restart Explorer). See pitfall #18 below.
>
> If you need to flip the *enabled* bit for a whole module, debounce isn't enough — call `Restart-PtRunner` after the write.
### §2.B — Interact with a UI element (2 techniques, most-reliable first)
Invoke goes through UIA InvokePattern COM IPC — no foreground steal, no UIPI. See references/winapp-ui-testing.md §CRITICAL — invoke vs click.
#### B2. PostMessage WM_KEYDOWN/CHAR — when UIA can't reach the target
For elevated targets, AppX windows with stunted UIA trees, or keystrokes that UIA `set-value` can't dispatch (arrow-key ListView nav, Enter to commit). See references/winapp-ui-testing.md §CRITICAL — Keystroke input that bypasses UIPI (PostMessage). Esc is often filtered by WinUI 3 raw-input hook — use BackButton invoke instead.
| **Proves** | The action fires (the path *downstream* of the hotkey). **Not** that the chord is bound. | The full path: real keys → runner hook → action. The **only** method that proves the chord binding itself. |
| **Robustness** | Highest — no foreground, no input desktop, UIPI-immune; works headless / RDP-minimized. | Lowest — needs an attached input desktop (else `BLK-ENV`), steals foreground, can't inject OS-reserved chords (Win+L / Win+Tab). |
| **Precondition** | Owning module process is running (the event only exists while it is). | Attached input desktop + foreground. |
**Pick by what the item asserts:** for "does action Y happen" use C1; for "pressing chord X triggers Y" or "the rebind takes effect", C1 is insufficient (it bypasses the chord) — use C2, or C1 *plus* a runner-log line proving the chord was accepted.
#### C1. Named Event signal — preferred
```powershell
Invoke-PtSharedEvent-Name'CmdPal.Show'# opens CmdPal without keyboard
Invoke-PtSharedEvent-Name'AOT.Pin'# pins foreground window via AOT
Invoke-PtSharedEvent-Name'PowerLauncher.Invoke'# opens PT Run
No synthetic input — it's a `SetEvent` on the kernel event the module waits on, the same downstream path the runner's hotkey handler signals. Verify the side effect via UIA (`winapp ui list-windows -a <module>`), a log line (`Get-PtRunnerLogTail`), or settings.json diff (`Get-PtModuleSettings`). The event only exists while the owning process runs, so `Test-PtSharedEvent` doubles as an "is the module alive" check.
#### C2. SendInput chord — last resort / chord-binding verification
Real synthetic keys. Loud (steals foreground) and fragile, but the only way to prove the activation chord is actually bound. The runner's global keyboard hook catches the chord regardless of focus, so the precondition is just an **attached input desktop** (pitfall #13; on a detached desktop `SendInput` returns `ACCESS_DENIED` and the keys vanish → mark `BLK-ENV`).
if(-not$line){throw'Runner did not log hotkey invocation'}
```
> **Rare fallback — a module that uses its own `RegisterHotKey` and exposes no Named Event.** Post `WM_HOTKEY` (`0x312`) straight to its message window (find the HWND via `EnumWindows`+`GetClassName` through `Add-Type` — same P/Invoke pattern as `pt-foreground-guard.ps1`). **No current PT module needs this:** ZoomIt — the obvious candidate — also waits on Named Events (`ZoomIt.Zoom`, `ZoomIt.Draw`, …; source: `Zoomit.cpp` `CreateEventW(ZOOMIT_ZOOM_EVENT)`), so drive it with C1.
> **Different case — sending keys *into* a specific focused window** (e.g. a CmdPal alias like `=` / `<` / `>` that `winapp ui set-value` can't trigger because it bypasses TextChanged; see pitfalls #4 and #6). Here the keystrokes go to whatever currently has focus, so you must bring the target window foreground first:
> ```powershell
> Assert-PtForegroundOrAbort -AppId Microsoft.CmdPal.UI # -AppId = the window you're typing INTO
> Send-PtChord -Key 0xBB # '=' (no modifiers) to trigger the calculator alias
> ```
> The `-AppId` is whatever window you're targeting — it's **not** CmdPal-specific. CmdPal is just the worst offender: its AppX foreground-lock drops focus after the first `SetForegroundWindow`, so without the guard the keys silently leak to your terminal.
> Verdict decisions (PASS if behavior matches spec; **FAIL** if the product is wrong *or* the checklist item is stale/ambiguous; BLOCKED if you couldn't run the check after ≥2 entry-paths) live in **Step 3 — Classification taxonomy** below. Don't put verdict logic in Step 2.
## Step 3 — Classification taxonomy
### Verdicts (assign exactly ONE per item)
| Verdict | Meaning |
|---|---|
| **PASS** | You drove/observed the behavior and it matched the spec. **A pass is a pass — there is no PASS sub-type.** Record *how* you verified in the item's **Category** field as free text, e.g. "full UIA flow + asserted popup", "settings.json round-trip", "runner-log line", "Shell COM / IExplorerCommand", "screenshot pixel-diff", "output matches fixture", "process spawn/exit", "module CLI", "admin GPO write". |
| **FAIL** | The item is **red** — something is wrong and action is required. Treat the checklist as test code: a test fails because **the product is wrong***or***the test/checklist is wrong**. Record the **cause** in the **Category** field: <br>• **product** — behavior contradicts a valid spec → file a product bug (repro + expected-vs-actual + screenshot/log + build version). <br>• **checklist** — the item itself is broken: *stale* (feature was removed/deprecated — cite the source grep proving it's gone) or *ambiguous* (`[CLARITY: VAGUE-*]`, no definable pass/fail criterion — quote the original wording). Fix the checklist, not the product. |
| **BLOCKED** | Couldn't run the check in this environment / with this toolset *after ≥2 entry-paths* — inconclusive, like a skipped test. **Not red against the product.** Tag exactly one concrete reason below. |
### BLOCKED reasons
Different failure reasons stay distinct because each drives a different remediation.
| Reason | When |
|---|---|
| `BLK-ENV` | This specific shell can't drive it (non-interactive / Session 0, RDP-minimized, missing Explorer windows) but a normal interactive desktop CAN. Triggers a "re-run on an interactive desktop" recommendation. Cite `references/environment-setup.md`. |
| `BLK-HARDWARE` | Needs hardware this session lacks — multi-monitor, 2 physical PCs (MWB), real camera / battery / game-mode, or live screen/device capture. State the specific shortfall in **Category**. |
| `BLK-DRAG-REQUIRED` | Needs a real mouse-drag gesture; synthetic drag is insufficient (e.g. FancyZones snap). |
| `BLK-DESTRUCTIVE` | Reboot, hibernate, install/uninstall, or mid-session AppX uninstall — would damage the run environment. |
| `BLK-VISUAL-RENDER` | The thing to verify is a rendered surface UIA can't see — WinUI3 islands, WebView2, or Explorer-side context-menu rendering/localization. Needs pixel/OCR or a manual eyeball. |
| `BLK-OVERLAY-INPUT-BLOCK` | Overlay both blocks input and excludes itself from capture (`BlockInput` + `WDA_EXCLUDEFROMCAPTURE`, e.g. ZoomIt draw mode) — can neither drive nor screenshot it. |
| `BLK-EXTERNAL-APP` | Needs a 3rd-party tool, a real API key, or a system locale change. |
**Rule of thumb**: in your report, separate the two FAIL causes — *product* FAILs are bugs to file; *checklist* FAILs are items to rewrite or prune. `BLOCKED` is only for a concrete, named obstacle (cite it), never a substitute for effort. If a large share of a module's items are checklist-FAILs, the checklist needs an overhaul before re-verifying.
## Step 4 — Report format
**See `references/reporting-format.md` for the full template** (per-item table, summary, step-table rules, anti-patterns, worked example). Don't paraphrase; copy the templates literally. This includes a mandatory **§G Retrospective** — a self-reflection on the *run itself*: list every friction encountered (classified by source — `SKILL-UNCLEAR` / `WINAPP-TOOL-BUG` / `WINAPP-DOC-UNCLEAR` / `HELPER-FLAW` / `PT-PRODUCT` / `ENVIRONMENT` — with severity + minutes/attempts cost + a suggested fix), or write `Everything was smooth — no friction encountered.` if there was none. This is how the skill improves run over run, so don't skip it.
## Step 5 — State hygiene (CRITICAL)
**See `references/pre-flight.md` §State hygiene** for the backup/restore pattern and cleanup commands. Always wrap mutations in `try { ... } finally { Restore-* }`.
## Module-specific quick reference
**Look for `references/modules/<module>.md` FIRST.** Each per-module profile contains paths, entry-paths, item-by-item recipes, common BLOCKED traps, fixture lists, and source citations specific to that module.
Catalog: see `references/modules/README.md`. Currently authored: `peek.md`, `power-rename.md`, `file-locksmith.md`, `image-resizer.md`.
If your module has NO profile yet:
1. Fall back to the generic drive-stack in §2 above.
2.**For Explorer-context-menu modules** (PowerRename / File Locksmith / Image Resizer / New+ / Preview Pane / RegistryPreview): read **`references/explorer-context-menu-flow.md`** first — it has the synthetic-right-click + UIA-invoke pattern with stability rules and module-caption table. Per-module profiles cite it and only document module-specific quirks. The canonical helper is `scripts/pt-explorer-contextmenu.ps1`.
3. After finishing the verification, **create the profile** using the template in `references/modules/README.md` so the next agent benefits from what you learned.
Quick one-liners for modules without dedicated profiles (will be moved to per-module files as they're authored):
- **FancyZones**: `Invoke-PtSharedEvent -Name 'FancyZones.ToggleEditor'`. Snap-drag tests are usually `BLK-DRAG-REQUIRED`; settings verify via settings.json round-trip.
4. Checklist item itself is broken — feature removed from source, or spec too ambiguous to judge → FAIL, cause=checklist (cite the source proof / quote the wording)
5. Couldn't drive it after ≥2 entry-paths → BLOCKED with a concrete reason (§3)
6. Record verdict + evidence + cleanup
7. Next item
```
When done, run state hygiene cleanup, write the report **including the §G retrospective**, archive the workspace (Step 7), and exit.
## Step 7 — Archive the workspace to the sign-off folder (do this LAST)
The live run works out of `%TEMP%`, but the **final deliverable must live in the module sign-off archive** so reports persist and sync via OneDrive:
```powershell
# After the report is written AND the artifact-existence check passes:
$signoff="$env:OneDrive\PowerToys\Module-Signoff"# e.g. C:\Users\<you>\OneDrive - Microsoft\PowerToys\Module-Signoff
Print the **moved** report path (under `…\PowerToys\Module-Signoff\`) as the last line — never the `%TEMP%` path.
## Invocation & placeholders
This skill auto-activates when you ask to verify a PowerToys module's checklist (e.g. "verify all Color Picker items"). **One module per run** — never chain multiple modules into one report. Resolve these placeholders for the module under test:
| Placeholder | Substitute with |
|---|---|
| `<Module>` | Exact display name, e.g. `Color Picker`, `Command Palette`, `PowerToys Run`, `FancyZones` (see `references/release-checklist/index.md`). |
| `<module>` | Lowercase-kebab-case for file lookup, e.g. `color-picker`, `command-palette`, `power-rename` — used for BOTH `references/release-checklist/<module>.md` (checklist) and `references/modules/<module>.md` (profile, if any). |
**Execution order:**`references/pre-flight.md` → per item, the §2 drive-stack (this file) → `references/reporting-format.md` per-item table → Step 6 verifier loop → `references/pre-flight.md` §Final wrap-up → Step 7 archive → print the final report path.
## What NOT to do
- Do NOT chain multiple modules in one report — one module per run.
- Do NOT mark an item BLOCKED without a concrete, named obstacle (see §3 and `references/pre-flight.md` §Hard rules).
- Do NOT invent steps for a VAGUE checklist item — if the spec is too ambiguous to judge, that is FAIL (cause=checklist), not a guess.
- All other rules (foreground guard, always restore mutated state, etc.) live in `references/pre-flight.md` §Hard rules — follow them.
## Critical pitfalls (PT-specific)
*Reference, not a sequential step — skim before you start and consult while driving. Numbered for cross-reference only.*
1.**PT runner does NOT auto-pickup edits to master `settings.json`** (top-level `enabled.<Module>` flags). Call `Restart-PtRunner`.
2.**Each module's own `<Module>\settings.json` IS hot-reloaded** via per-module file watcher (~3s debounce). **EXCEPTION — shell-extension/context-menu modules do NOT read this file; see pitfall #18.**
3.**PT Run setting key has a space**: `"PowerToys Run"` not `PowerToysRun`.
4.**CmdPal AppX foreground from external CLI is unreliable** — Windows foreground-lock blocks `SetForegroundWindow` after the first call. SendInput keys silently leak to your terminal. **Always `Assert-PtForegroundOrAbort` before SendInput.**
5.**CmdPal AppX enters TextChanged-broken state** every ~30 probes — `Test-CmdPalDegraded` + `Reset-CmdPalAppX` to recover.
6.**CmdPal alias detection (`=`, `<`, `>`, `:`, `$`, `??`, `)`) requires real keystrokes** — `winapp ui set-value` bypasses TextChanged and the alias never fires. Use Send-PtChord + `Assert-PtForegroundOrAbort` for aliases; use set-value for plain queries.
7.**CmdPal Esc handler is filtered** by WinUI 3 raw-input hook — use `winapp ui invoke BackButton` instead (see `Reset-CmdPalToHome`).
8.**GPO HKLM vs HKCU**: HKLM wins when both are set with conflicting values.
9.**HKLM `Software\Policies\PowerToys` writes require admin** — verify with `Test-PtAdmin`.
10.**`Stop-Process` is policy-blocked in this session unless you pass `-Id <int>` literally**. Always inline the PID.
11.**WinUI 3 islands are largely invisible to UIA** (QuickAccess flyout, RegistryPreview Monaco editor, Peek WebView2). For these, fall back to screenshot + OCR or settings.json diff.
12.**OS-reserved chords (Win+L, Win+Tab)** are consumed by Windows before any hook and cannot be injected via SendInput at all.
13.**RDP minimized = `SendInput` denied.** Even though `quser` shows the remote session State=Active, minimizing the mstsc client detaches the session's input desktop. `GetForegroundWindow()` returns 0; `SendInput` returns `ACCESS_DENIED (5)`; tests that need synthetic input fail. **Same applies to: closed mstsc with X (Disconnected), local PC sleep (RDP TCP drops), remote screensaver/workstation lock, remote machine sleep.** Run `scripts/pt-session-diagnose.ps1` in pre-flight to detect, and see `references/environment-setup.md` for the full per-scenario table + `powercfg` setup commands the user should run before starting the agent. The agent should call `Test-PtForeground` mid-run before each input-injection-dependent item; if it returns False, mark `BLK-ENV` with mitigation citation (an environment block — not a product FAIL).
14.**`winapp ui` arg-order quirk**: `winapp ui inspect --depth N -w $hwnd` may intermittently fail to parse `--depth` as Int64 if `-w` precedes it. **Put `-w $hwnd` AFTER `--depth N`** or as the first arg before any flag. If you see "Cannot bind argument" or numeric parse errors, swap the order and retry.
15.**`winapp ui list-windows` line wrapping**: when window titles or process names are long, output may wrap a single window's `HWND <id>: "<title>" ... (proc, PID N)` across multiple lines, breaking single-line regexes. Either pipe through `Out-String` and use a multi-line regex, or use `--json` (when supported) and parse structured output.
16.**De-elevation: launching a NON-elevated (Medium IL) child from an elevated agent shell.** The drive-stack only covers gaining *more* privilege; some items need the opposite. From a High-IL shell you cannot `Start-Process` a Medium-IL child directly. Use `scripts/pt-nonelevated.ps1` (`Start-PtNonElevated` / `Invoke-PtNonElevatedCapture`) — a one-shot `RunLevel Limited` + `LogonType Interactive` scheduled task that lands on the user's desktop at their filtered token. Confirm with `Test-ProcessElevated`. Needed for elevation-visibility pairs (File Locksmith L649/L650: non-elevated FL must not see the elevated runner; elevated FL must).
17.**Win11 packaged context menus are not observable without real Explorer.** Modern PT context-menu entries are packaged `IExplorerCommand`s (sparse MSIX, e.g. File Locksmith CLSID `{AAF1E27D-…}`). They are **NOT** enumerable via classic `Shell.Application … FolderItem.Verbs()` and **NOT**`CoCreate`-able from a non-Explorer host (`REGDB_E_CLASSNOTREGISTERED`). So "verify the entry appears / no longer appears" cannot be pixel-verified by API. Verify instead via the gate flag the entry's `GetState` reads (e.g. general `enabled.<Module>`) + a source citation that maps it to `ECS_HIDDEN`; treat the literal render as `BLK-VISUAL-RENDER` and recommend a 5-second manual right-click. (Disabling does NOT unregister the package — it stays `Status Ok`; the entry is hidden dynamically.)
18.**Shell-extension modules read a module-OWNED settings file, NOT the PT-store `<Module>\settings.json`.** PowerRename, File Locksmith, Image Resizer, and New+ context-menu handlers and exes run *outside* the runner (hosted by Explorer / launched on demand) and cannot use the PT-Settings IPC. Each reads its **own** json in `%LOCALAPPDATA%\Microsoft\PowerToys\<Module>\`*at process/handler launch* (registry-migrated `CSettings`/`Settings` classes — `lib/Settings.cpp``Load→ParseJson`). The PT-Settings UI writes the *PT-store*`settings.json` (the `bool_*`/`int_*` file `Get-PtModuleSettings` reads); the runner's module DLL syncs PT-store→module-store **only on a Settings-UI change event** — so the PT-store file can be **stale for days** and editing it has **no effect** on the running shell handler. **To drive a settings item on these modules, edit the module-owned file directly (drive-stack §2.A) and relaunch the module (or restart runner+Explorer for the menu handlers), then restore.**
**Pitfall #18 — module-owned files + their key style** (verified 2026-06-10 against `<PT-repo>\src`):
Confirm which file actually drives behavior with a quick A/B: edit the module-owned file → relaunch → observe; if behavior follows, that's the source of truth (PowerRename L394/L395/L396/L397/L409 were all driven this way).
If you find another gap during verification, update this skill (add a recipe) AND consider proposing the addition to references/winapp-ui-testing.md if it's generic enough.
**Audience**: human user preparing a test machine before running a verification agent.
**One-time** (per test session) — restore afterward.
## Why this matters
PowerToys release checklists test real user interactions: pressing hotkeys, dragging files, switching windows. Many tests use `SendInput` to inject keystrokes. Windows refuses `SendInput` when the calling session has **no attached input desktop** — and several common Windows states cause exactly that to happen:
- RDP client minimized
- Workstation locked (screensaver kicked in, idle timeout)
- Remote machine asleep
- Local machine asleep (RDP TCP drops)
If any of these happens mid-verification, items that need synthetic input fail with `BLK-ENV` even though the feature itself works fine. This guide eliminates the env causes so the only BLOCKED verdicts you see are real test/framework limitations.
## Per-scenario reference table
| Scenario | Remote session State | `GetForegroundWindow()` | `SendInput` | Verdict for input-injection tests |
|---|---|---|---|---|
| mstsc window focused | Active | Real HWND | Works | ✅ Drivable |
| mstsc visible but not focused (covered or alt-tabbed) | Active | Real HWND | Works | ✅ Drivable |
| Remote screensaver / auto-lock kicks in | Active but desktop locked | 0 | ACCESS_DENIED | ❌ BLK-ENV |
| **2nd RDP login as the SAME user** (you reconnect from another client) | the OLD session flips to **Disconnected** | 0 (in the old session) | ACCESS_DENIED | ❌ BLK-ENV — your running test's session got taken over |
**Key insight**: "Active" in `quser` ≠ "can inject input". Always check `GetForegroundWindow()` first (the diagnostic script `scripts/pt-session-diagnose.ps1` does this).
## Can I verify two modules at once in two RDP sessions?
Short answer on a **client edition of Windows (Windows 10/11, ProductType=1)**: **no — not as the same user, and effectively not at all.** This was investigated live on this machine (Windows 11 Enterprise, build 26200, `fSingleSessionPerUser=1` default):
- **Two monitors ≠ two sessions.** A multi-monitor setup is **one** session spanning both screens — it shares a single input desktop, foreground window, and `SendInput` queue across the monitors. Monitor count has nothing to do with session count, so "I have two monitors" does not give you two sessions to run two modules in.
- **Sessions are isolated** — each Windows session has its own input desktop, its own foreground window, and its own `SendInput` queue. So *typing in session B genuinely does NOT disturb session A's foreground or input.* Cross-session interference is **not** the problem (so if you somehow DID have two live sessions — Server/RDS — they could run in parallel without colliding).
- **The real blocker is session takeover.** Client Windows allows only **one interactive (console/owning) session at a time**, and `fSingleSessionPerUser=1` (the default) means one user gets **one** session. When you open the *second* RDP connection (as the same user), Windows **disconnects the first session** — it flips to `Disconnected`, its input desktop detaches, `GetForegroundWindow()` → 0, and any in-flight UI test there fails with `ACCESS_DENIED` → BLK-ENV. It's not your *typing* that breaks the test; it's the act of logging in the second session that evicts the first.
- A different *user* account doesn't rescue it either: client Windows still permits only one connected interactive session, so the second login still disconnects the first.
- Therefore, on client Windows, **run modules serially in one session.** True concurrent multi-session needs Windows Server + the RDS (Remote Desktop Session Host) role; unofficial multi-session patches exist but are out of scope here.
> **Verdict on the common assumption "I can run two modules in two RDP sessions because I have two monitors":** the *conclusion* (can't run two at once on client Windows) is correct, but the *reasoning* is wrong on two counts — two monitors is still one session, and you can't get two simultaneously-Active sessions on client Windows at all (the 2nd login disconnects the 1st). The limit is "can't open a 2nd Active session", not "the two sessions fight each other".
**Practical guidance:** keep a single RDP session for the whole run; don't reconnect/relogin mid-run; if you must check something elsewhere, alt-tab inside the *same* session rather than opening a new RDP connection. To detect a takeover after the fact, `qwinsta` will show your former session as `Disconnected`.
## Pre-run setup checklist
Run these BEFORE starting the verification agent.
### On the test machine (the one being verified)
```powershell
# Snapshot current power settings so you can restore after
powercfg/querySCHEME_CURRENTSUB_SLEEP|Select-String'Power Setting GUID|Current AC Power Setting Index'
```
### On the local machine (the one with the RDP client)
```powershell
# Disable local sleep so RDP TCP stays alive
powercfg/changestandby-timeout-ac0
# Practical habit: put mstsc on a monitor you're NOT actively working on.
# Don't minimize. Alt-tab is fine; minimize is not.
```
## Mid-run discipline
While the agent is running:
- **Don't minimize mstsc.** Visible-but-unfocused is OK; minimized is not.
- **Don't close mstsc with the X.** If you have to step away, fine — leave it open.
- **Don't disconnect or reconnect RDP.** Stay continuously connected for the duration of the run.
- **Don't sign out** on either end.
- If you do step away and the screen locks (despite the setup above), reconnect/unlock and the agent's `Test-PtSessionStillInteractive` guard (if used) will resume; otherwise items mid-execution will be BLK-ENV.
## Post-run cleanup (restore)
```powershell
# Restore the values you captured to $bk before starting
(Values above are typical; adjust to your environment policy.)
## Diagnostic before you start
Run `scripts/pt-session-diagnose.ps1` from the agent shell. Expected output for a GO:
```
PASS - this shell can drive interactive PowerToys tests.
```
If it prints FAIL with a `psexec -i <consoleSession> -s pwsh.exe` hint, you're in a non-console session — relaunch the agent shell as suggested before starting verification.
## Why this isn't in the global SKILL.md
These are **human prep steps**, not agent instructions. The agent needs to *detect* a bad environment (via `Test-PtInteractiveDesktop` in pre-flight + `Test-PtSessionStillInteractive` mid-run); the user needs to *prevent* one. Different audiences, different docs.
# Explorer context-menu flow — driving PowerToys shell-menu modules end-to-end
**Audience**: agents verifying any PowerToys module whose entry point is the **Windows Explorer right-click context menu** — i.e. **File Locksmith, Image Resizer, PowerRename, New+ (NewPlus)**, and similar.
This is the *true user flow*: open Explorer → select file(s) → right-click → click the module's menu item. Use it when an item's assertion is specifically about the **context menu** (e.g. "the entry appears / no longer appears", "right-click → X launches the module on the selection"). For the module's *internal* behavior you can still prefer a faster back-door (CLI / `last-run.log` / Named Event) — see each module profile — but the menu presence/launch itself can only be observed this way.
## Which approach first? (CLI / back-door vs synthetic menu)
**Pick the tool by what the item ASSERTS — not "always synthetic" or "always CLI".**
| The item asserts… | First approach | Why |
|---|---|---|
| **The menu itself** — entry *appears / no longer appears*, "right-click → select X", caption / localization of the entry | **Synthetic Explorer menu (this doc)** — the *only* valid observer | The CLI/back-door is **blind to the menu**: it runs even when the entry is correctly hidden, so it gives a false PASS (the L652 trap). If the desktop is locked → `BLK-ENV`; do **not** substitute the CLI. |
| **Module behavior** — engine finds the lockers, images get resized, files get renamed (the menu is just the trigger) | **CLI / back-door** (`FileLocksmithCLI.exe`, `last-run.log`, Named Event, DSC) | Instant, deterministic, foreground-free, works on a locked desktop. Synthetic adds ~10s + foreground/retry fragility without changing the assertion. |
**Golden-path rule (do once per module):** run **one** full synthetic right-click → invoke-the-item → confirm-launch. That proves the menu→launch wiring is actually registered *and* validates that the fast back-door is behaviorally equivalent to the real menu (e.g. File Locksmith L641 `step-04/05` did exactly this). After that one golden run, trust the back-door for the remaining behavior items.
Net: for a context-menu module, **most items are behavior → CLI-first**; the **menu-presence/absence/launch/localization items → synthetic-first**; plus one golden-path synthetic launch.
## Is it stable?
**Yes — with the robust variant below.** Verified repeatedly on Win11 (2026-06-08) launching File Locksmith via a genuine right-click + menu click. Two rules make it reliable; ignore them and it gets flaky:
1.**Invoke the menu item by UIA InvokePattern, not a coordinate left-click.** The menu item exposes `InvokePattern` (`isInvokable=True`). `winapp ui invoke <selector> -w <menuHwnd>` is robust and needs no foreground/coordinates for the *click*. A synthetic left-click at the item's pixel center also works but is the fragile part (DPI, menu repositioning near screen edges, scrolled menus).
2.**The right-click that OPENS the menu still needs synthetic input on a foregrounded window — and occasionally a retry.** The first right-click right after Explorer opens sometimes misses (foreground not settled). `Open-PtExplorerContextMenu` retries up to 3×; that removed the flakiness in testing.
**Hard prerequisite — unlocked interactive desktop.** Synthetic right-click injects into the session input stream, so it requires foreground. If the workstation is locked / RDP minimized (`GetForegroundWindow()=0`), this flow is `BLK-ENV` — there is no foreground-free way to open a context menu. `Open-PtExplorerContextMenu` throws a clear BLK-ENV error in that case. (A 4-hour idle auto-lock is the common culprit — see `references/environment-setup.md`.)
**Other constraints:**
- **Settings for these modules live in a module-OWNED file, not the PT-store `settings.json`** — see `SKILL.md` pitfall #18. The context-menu handler reads e.g. `power-rename-settings.json` / `file-locksmith-settings.json` / `image-resizer-settings.json` / `New\settings.json` at launch; editing the PT-store `<Module>\settings.json` (what `Get-PtModuleSettings` reads) often has **no effect** on the live handler. Drive icon/extended-menu/feature toggles via the module-owned file + relaunch (restart runner+Explorer for the menu handlers), then restore.
- This is the **Win11 packaged** context menu (`Microsoft.UI.Content.PopupWindowSiteBridge` / "PopupHost"). The packaged module commands appear **only** here — not in classic `Shell.Application.Verbs()` and not via `CoCreate` of the command CLSID (`REGDB_E_CLASSNOTREGISTERED`). On Win10, or under "Show more options", you'd get the classic menu instead (different structure).
- The menu exists in the UIA tree **only while open** — you must open it with real input first; you can't enumerate it cold.
- A menu-launched module UI runs **non-elevated** (Explorer's integrity), even if your agent shell is elevated. Mind elevation-visibility (e.g. a non-elevated File Locksmith can't see higher-IL processes — match locker integrity with `scripts/pt-nonelevated.ps1`).
## Recipe (robust)
```powershell
."$skill\scripts\pt-explorer-contextmenu.ps1"
# 0) Guard: must be an unlocked desktop
if(-not(Test-PtDesktopInteractive)){<# mark BLK-ENV, cite references/environment-setup.md #>}
# 1) Open Explorer on the target folder and grab its CabinetWClass HWND
$ui=Get-ProcessPowerToys.FileLocksmithUI-EASilentlyContinue# or PowerToys.ImageResizer, PowerToys.PowerRename
```
To **assert absence** after disabling a module: re-open the menu and check `Get-PtContextMenuItems` no longer contains the caption (the packaged `GetState` re-reads the enabled flag live, so no Explorer restart is needed between toggles).
These operate on a **selection** of files. Select first (Shell COM is reliable and foreground-free), then right-click one of the selected items:
- Use `scripts/pt-explorer-com.ps1` → `Open-PtExplorerAtPath` + `Select-PtExplorerFiles` to establish the multi-select.
- Then `Open-PtExplorerContextMenu` on one selected file and `Invoke-PtContextMenuItem` — the module receives the whole selection (the shell handler enumerates all selected `IShellItem`s).
## Module captions (match by NAME)
Match the **visible caption**, not the AutomationId (Explorer assigns per-session numeric IDs like `32012` whose value/order varies). Discover the exact caption at runtime with `Get-PtContextMenuItems`. Verified captions:
| Module | Launched process | Menu caption (verified ✓ / expected) |
|---|---|---|
| File Locksmith | `PowerToys.FileLocksmithUI.exe` | ✓ `Unlock with File Locksmith` (NB: **not** the checklist's "What's using this file?") |
| PowerRename | `PowerToys.PowerRename.exe` | ✓ `Rename with PowerRename` |
| Image Resizer | `PowerToys.ImageResizer.exe` | `Resize images` (verify via `Get-PtContextMenuItems` — caption shifted across versions) |
| New+ | (creates from template) | `New+` (submenu) |
> Tip: if a module's caption is unknown, enable the module, open the menu on an applicable file, and run `Get-PtContextMenuItems` to read the exact string — then hard-match it for present/absent assertions.
## Common failure modes → fixes
| Symptom | Cause | Fix |
|---|---|---|
| `BLK-ENV: ... GetForegroundWindow()=0` | desktop locked / RDP minimized | unlock & keep mstsc un-minimized (`references/environment-setup.md`); mark `BLK-ENV`, not a test failure |
| "popup not found after N attempts" | foreground not settled (esp. first right-click after Explorer opens) | the helper already retries 3×; raise `-MaxTries`, or pre-foreground the window once before calling |
| menu item `invoke` returns but nothing launches | matched the wrong node / item disabled | match `type -eq 'MenuItem'` by exact Name; confirm the module is enabled |
| caption not found though module enabled | wrong/old caption string, or it's under "Show more options" (classic menu) | enumerate with `Get-PtContextMenuItems`; for classic menu invoke `expandtoclassic` first |
| launched UI shows nothing | menu-launched UI is non-elevated and can't see higher-IL targets | match target integrity (`scripts/pt-nonelevated.ps1`) |
## Referenced by
-`references/modules/file-locksmith.md` (L641/L652 — real right-click launch + menu present/absent)
- *(future)* `references/modules/image-resizer.md`, `references/modules/power-rename.md`, `references/modules/new-plus.md` — reference this doc for their context-menu items.
This folder holds **one short profile per PowerToys module**. Each profile is self-contained guidance specific to that module — paths, entry-paths, capability/control recipes, common BLOCKED traps, fixture lists, source citations.
## When to read
When this skill runs for a specific module, check whether `references/modules/<module>.md` exists here. If yes: **read it BEFORE walking the SKILL.md drive-stack** — it tells you which entry-paths actually work for this module's quirks and which BLOCKED traps to avoid.
If no profile exists, fall back to SKILL.md + the helper scripts.
## Shared cross-module flows
Some flows are common to several modules and live in their own top-level docs (not per-module):
- **`../references/explorer-context-menu-flow.md`** — driving the real Win11 Explorer right-click context menu end-to-end (open + assert present/absent + launch). Referenced by File Locksmith and any future **Image Resizer / PowerRename / New+** profiles.
## Why per-module (not just one big SKILL.md)
- Each module has its own quirks (Peek's `_isFromCli` guard, CmdPal's TextChanged-broken state, PT Run's mini-popup HWND, Workspaces' snapshot-elevation rules). Bundling all of them into the global SKILL.md bloats context and forces every verification to load 25+ KB of mostly-irrelevant text.
- A profile lets a focused verification run with only the relevant 5-10 KB.
- New gotchas discovered during a module verification round get added to that module's profile, not the global one — keeps the global doc stable.
## Profile catalog
| Module | Profile | Status |
|---|---|---|
| Peek | `peek.md` | ✅ written 2026-06-08 |
| File Locksmith | `file-locksmith.md` | ✅ written 2026-06-08 |
| Image Resizer | `image-resizer.md` | ✅ written 2026-06-09 |
| PowerRename | `power-rename.md` | ✅ written 2026-06-10 (first to cite `../context-menu-cookbook.md` for shared mechanics) |
| New+ | `new-plus.md` | ✅ written 2026-06-18 (registration-gate for menu presence; Settings-UI toggle drives template auto-copy) |
| (other modules to be added as we encounter sign-off needs) | — | — |
## For Explorer-context-menu modules: read the canonical flow doc first
If you're writing a profile for a module that registers an entry in Explorer's Win11 right-click menu (PowerRename, File Locksmith, Image Resizer, New+, Preview Pane, RegistryPreview), **read `../references/explorer-context-menu-flow.md` first**. It has the canonical synthetic-right-click + UIA-invoke recipe with:
- Which-approach-first decision rule (CLI back-door vs synthetic menu, with the false-positive trap warning)
- Stability rules (UIA InvokePattern, retry on first right-click)
- The unlocked-desktop requirement (BLK-ENV gating)
The shared helper is `scripts/pt-explorer-contextmenu.ps1` (`Test-PtDesktopInteractive`, `Open-PtExplorerContextMenu`, `Invoke-PtContextMenuItem`, `Get-PtContextMenuItems`).
Your module profile then only documents the **module-specific** quirks: settings.json schema keys, expected verb caption regex, capability/control recipes, source citations, ceiling.
`power-rename.md` is the model — ~9 KB despite covering 18 items because the generic mechanics live in the canonical flow doc.
## Recipes — a control/observation map, NOT a per-test-case answer key
| # | Capability | Drive (control / settings key) | Observe (where the result shows) |
|---|---|---|---|
| 1 | <a module capability, e.g. "context-menu entry present when enabled"> | <which AutomationId / control / settings key drives it> | <where the result is visible: preview column, settings.json, disk, log, menu> |
| 2 | <next capability> | <...> | <...> |
> **Mapping process** (agent at runtime): read the actual checklist item → identify the capability → find its row → drive the named control and **design your own inputs + assertions for that item**. If no row matches, it's a NEW capability — drive ad-hoc and add a row (capability + control + observation point; no canned inputs).
> **Why a map, not an answer key**: the table must carry only **durable module knowledge** — which control drives a capability and where to observe the result. Concrete Search/Replace inputs and expected-output assertions are *per-test-case answers*; baking them in turns the profile into a cheat sheet that (a) lets the agent copy answers without understanding and (b) goes stale the moment a checklist item changes its wording or values. Keep inputs + assertions OUT. Only a real UI redesign (a renamed/moved/removed control) should force an edit to this table.
## Common BLOCKED traps
<list of mistakes prior agents made + how to avoid them>
## Fixture files needed
<list of pre-canned files the verification expects>
## Source citations
<paths in PT repo that explain module behavior or guards>
## Ceiling
<observed PASS rate / total>
## Don'ts
<list of common mistakes>
```
## Hygiene
- **Keep each profile under ~10 KB.** If it grows beyond that, the module has too many quirks — escalate to maintainer review of the upstream checklist.
- **The recipe table is a control/observation MAP, not an answer key.** Columns are *Capability → Drive (control/key) → Observe*. **Do NOT bake in concrete Search/Replace inputs or expected-output assertions** — those are per-test-case answers that go stale when a checklist item changes and let the agent copy without understanding. The agent designs inputs + assertions at runtime from the actual checklist item.
- **Tables are capability-keyed, NOT line-keyed.** Upstream checklist line numbers (`L<n>`) **must not appear** in the profile — they drift between releases (items added/removed/reordered) and turn the table into a silent mismatch trap. PT-source-code file:line citations (e.g. `dllmain.cpp:73`) ARE allowed; they're version-pinned and serve a different purpose.
- **Cite source-code line numbers** where module behavior surprises (e.g. CLI guards, debounce timings, fallback chains). Reviewers can verify your claims by reading those lines.
- **Update the profile after every verification round**; promote any new technique into the right helper script if it generalizes beyond this module.
**Context menu**: Win11 packaged `IExplorerCommand` CLSID `{AAF1E27D-4976-49C2-8895-AAFA743C0A7E}` (sparse pkg `Microsoft.PowerToys.FileLocksmithContextMenu`); legacy `FileLocksmithExt.dll`. Caption resource = "What's using this file?".
**Named Event / DSC**: no Named Event. DSC resource `microsoft.powertoys.FileLocksmith.settings` exists (controls module settings, not the master enable flag).
**No global hotkey** — entry is the Explorer context menu only.
## The two back-doors that make this module fully drivable (no Explorer needed)
### 1. `FileLocksmithCLI.exe` — deterministic engine ground-truth (PREFER for assertions)
Accepts paths as args; `--json`, `--kill`, `--wait`, `--timeout`. Uses the **same**`find_processes_recursive` engine as the UI.
&$cli"<path>"--kill# terminate lockers (== End task)
```
Detection = open **File handles** + **loaded modules** under the path (exact file, exact dir, dir-prefix recursive). `FileLocksmith.cpp:18-113`.
### 2. `last-run.log` IPC + launch UI — exercises the REAL UI code path
The context-menu handler writes selected paths to `…\File Locksmith\last-run.log` then launches the UI; the UI reads them in `MainViewModel()` ctor. Reproduce it:
```powershell
# UTF-16LE, each path + WCHAR \n, trailing empty-line terminator, NO BOM
- Per-row End task button = `btn-…` (parent of `lbl-endtask-…`); invoke it (`InvokePattern`).
-`RestartAsAdminBtn` = shield icon, **visible only when non-elevated** (`MainPage.xaml:72-82`).
-`ProcessesListView` is virtualized; use `winapp ui scroll ProcessesListView --direction down/up`.
## Recipes — a control/observation map, NOT a per-test-case answer key
> Maps each capability to **how to drive it** and **where the result shows**. No canned process counts / paths / assertions — design those at runtime from the actual checklist item.
| # | Capability | Drive (entry-path / control) | Observe (where the result shows) |
|---|---|---|---|
| 1 | A locked file lists all its locking processes | CLI + UI on one locked file (with multiple lockers) | each locker shows as a ListItem |
| 2 | "End task" kills the locker and de-lists it | `winapp ui invoke` the End-task button | locker PID dies + row removed |
| 3 | Reload rediscovers a locker started after the UI opened | start a new locker → invoke Reload | the new locker appears |
| 4 | Closing a locker externally auto-removes it | external `Stop-Process` on a locker | auto-delisted via `WatchProcess`; empty state shown |
| 5 | Directory path finds lockers recursively | CLI/UI with a directory path | lockers inside the tree are listed |
| 6 | Drive root lists many lockers without crashing | CLI/UI with a drive root | large list renders; no crash |
| 7 | Non-elevated FL does NOT see the elevated runner | run CLI/UI non-elevated via scheduled task `RunLevel Limited` | `PowerToys.exe` absent (medium-IL FL can't see elevated procs) |
| 8 | "Restart as administrator" surfaces elevated-only lockers | non-elev UI shows the button; elevated run shows them | elevated run lists `PowerToys.exe` (UAC consent click NOT automatable) |
| 9 | Scrolling a large list doesn't crash | UI on a drive root + `winapp ui scroll` | process alive + responsive after scroll |
| 10 | Disabling FL removes the Explorer context-menu entry | Settings toggle Off (winapp ui) | `enabled."File Locksmith"`→false; `GetState→ECS_HIDDEN` (source) |
> **Mapping process**: read the actual checklist item → identify the capability → find its row → drive the named control and design your own inputs + assertions. If no row matches, drive ad-hoc and add a row (capability + control + observation point; no canned inputs).
## Common BLOCKED traps (avoid)
- **Don't BLOCK the lock-detection / End-task / scroll-doesn't-crash items as "needs a real installer / right-click"** — both the CLI and the `last-run.log`+UI back-door fully drive them with any locked file. The context menu literally just writes `last-run.log` + launches the same exe.
- **Launching the UI from an elevated shell makes it elevated** (title "Administrator: …") and hides the `RestartAsAdminBtn`. To test the non-elevated case (Recipes 7-8), launch via a **scheduled task with `-RunLevel Limited` / LogonType Interactive** — it lands on the user's desktop at medium IL.
- **`set-value` does fire the Reload** here (TogglePattern/Invoke work); no TextChanged gotchas.
A medium-IL File Locksmith can't `DuplicateHandle`/read modules of higher-integrity processes, so the **elevated PowerToys.exe runner is invisible** to a non-elevated FL and **visible** to an elevated FL (which also calls `SetDebugPrivilege`, `App.xaml.cs:53-61`). Per-user installs put PowerToys under `%LOCALAPPDATA%\PowerToys`, not "Program Files" — use the PT install dir as the stand-in and note the caveat.
## Context-menu disable gate (Recipe 10)
`GetEnabled()` reads general `enabled."File Locksmith"` (`Settings.cpp:53-77`). When false: Win11 `GetState→ECS_HIDDEN` (`dllmain.cpp:81-84`); legacy `QueryContextMenu→E_FAIL` (`ExplorerCommand.cpp:116-119`). Disabling does NOT unregister the sparse package (stays `Status Ok`) — the entry is hidden dynamically. The packaged `IExplorerCommand` is **not** enumerable via Shell `Verbs()` and **not**`CoCreate`-able from a non-Explorer host (`REGDB_E_CLASSNOTREGISTERED`), so the pixel-level render is the only un-automatable bit (`BLK-VISUAL-RENDER` if you need it).
## Fixture files needed
None pre-canned. Create a temp file and lock it with a helper process (pwsh holding `File.Open(path, OpenOrCreate, Read, ReadWrite)` so multiple lockers coexist).
10/10 PASS observed (2026-06-08). The lock-detection / End-task / refresh / drive-scroll items cleanly driven; the Restart-as-admin item PASS-with-caveat (UAC consent click not automatable; outcome verified). **The disable-removes-menu item PASS — behaviorally verified** by a real Explorer right-click: with FL enabled the Win11 menu shows `MenuItem "Unlock with File Locksmith"`; after disabling in Settings the same right-click menu no longer shows it (no Explorer restart needed — `GetState` re-reads `enabled."File Locksmith"` live). NB the **shipped caption is "Unlock with File Locksmith"**, not the checklist's "What's using this file?". The right-click test needs an **unlocked interactive desktop** (a 4-hour idle auto-lock makes `GetForegroundWindow()=0` → `BLK-ENV`).
## Real right-click verification (Recipe 10) — works on an unlocked desktop
**Use the shared flow: `references/explorer-context-menu-flow.md` + `scripts/pt-explorer-contextmenu.ps1`.** FL's caption is **"Unlock with File Locksmith"**. Quick version:
**No global hotkey / no Named Event / no DSC for the engine** — entry is the Explorer menu (or direct exe launch).
## The back-door that makes this module ~fully drivable (no Explorer needed)
### `PowerToys.ImageResizerCLI.exe` — the deterministic engine (PREFER for all resize-behavior items)
Shares the exact `ResizeBatch.FromCliOptions` → `ResizeBatch.ProcessAsync` → `ResizeOperation.ExecuteAsync` engine as the GUI (`ui/Cli/ImageResizerCliExecutor.cs:76-108`, `App.xaml.cs:102`). Reads live `Image Resizer\settings.json` then applies CLI overrides (`CliSettingsApplier.cs`).
--show-config --help <files…> (also accepts \\.\pipe\<name> and stdin file list)
```
`--show-config` dumps the effective settings (great pre/post check). `-d <dir>` keeps outputs isolated. Assert output dimensions with `[System.Drawing.Image]::FromFile(p)`.
**Caveat**: `--ignore-orientation`/`--shrink-only`/`--replace`/`--keep-date-modified`/`--remove-metadata` are *flags* — they can only set the value **true**; to test the **false** case, temporarily edit `settings.json` (back up + restore).
### Direct GUI launch — for the two UI-only items (gif warning, size-list populated)
`Start-Process PowerToys.ImageResizer.exe "<file>"` opens the window pre-loaded (argv/stdin via `ResizeBatch.FromCommandLine`). Behaviorally identical to the context-menu launch (`dllmain.cpp:219-245` just writes a pipe + launches the same exe). Then drive with `winapp ui`:
- Size selector: `SizeComboBox` → `winapp ui invoke SizeComboBox -w <hwnd>` to expand, then `inspect` shows `itm-<name>-XXXX` ListItems.
- Gif warning: `Message Text "Gif files with animations may not be correctly resized."` InfoBar, bound to `ViewModel.HasGifFiles` (set when any file ends `.gif`).
## Engine facts (verified from source — cite these for the resize items)
- Replace: `File.Replace(out, src, backup)` then recycle backup — no copy left (`ResizeOperation.cs:151-156`).
- IgnoreOrientation swap: gated by `IgnoreOrientation && !HasAuto && Unit != Percent` (`ResizeOperation.cs:419-444`).
## Recipes — a control/observation map, NOT a per-test-case answer key
> Maps each capability to **which control/CLI flag drives it** and **where the result shows**. CLI flag *names* and fit-mode/unit/field enumerations are stable IR knowledge and stay; concrete flag *values*, fixtures, and expected outputs are per-test-case — design those at runtime.
| # | Capability | Drive (control / CLI flag) | Observe (where the result shows) |
| 2 | Module enabled → entry present (modern + classic), click launches the GUI | synthetic menu + invoke | `Get-PtContextMenuItems` shows "Resize with Image Resizer"; classic "Show more options" too; invoke → `PowerToys.ImageResizer.exe` launches |
| 3 | Remove a built-in size / add a custom size | edit `imageresizer_sizes` (INTEGER Ids!) + launch GUI | `SizeComboBox` reflects the edit (removed gone, custom present) |
| 4 | Resize one / multiple files end-to-end | CLI `--size <id> [files…]` | outputs at the size's Fit dimensions |
| 5 | GIF animation warning on `.gif` input | GUI on a `.gif` | warning InfoBar present (`winapp ui inspect`) |
| 6 | Fit modes (Fill / Fit / Stretch) | CLI `--width --height --fit <mode>` | output shape matches the mode (crop / letterbox / exact) |
| 7 | Unit conversion (cm / inch / percent / pixel) | CLI `--unit <u>` | output px = unit converted at the image's DPI |
| 8 | Custom filename format (`%1`..`%6` fields) | CLI `--filename <fmt>` | output filename follows the format fields |
| 9 | "Keep date modified" | CLI `--keep-date-modified` | output mtime == source mtime (control: without the flag, differs) |
| 10 | "Shrink only" | CLI `--shrink-only` | an already-small image is untouched (control: a large one still shrinks) |
| 11 | "Replace original" | CLI `--replace` | original replaced in place; no `(name) (1)` copy |
| 12 | "Ignore orientation" | settings (false) vs flag (true) | on a portrait target over a landscape image: false→no W/H swap, true→swap |
> **Mapping process**: read the actual checklist item → identify the capability → find its row → drive the named control/flag and design your own inputs + assertions. If no row matches, drive ad-hoc and add a row (capability + control + observation point; no canned inputs).
## Common BLOCKED traps (avoid)
- **Don't mark the resize-behavior items BLOCKED for "needs a real right-click".** The CLI fully drives them with the identical engine; the menu is just the trigger (prove the menu→launch wiring once with the golden path in Recipe 2).
- **PowerShell `ConvertTo-Json` writes computed numbers as doubles (`"Id": 3.0`)** → `System.Text.Json` rejects `imageresizer_sizes` and the app silently falls back to the 4 built-in default presets (Small/Medium/Large/Phone). Cast Ids to `[int]` or regex-strip `\.0`. This bit the remove-size-add-custom item (Recipe 3) the first time.
- **cm/inch outputs depend on the fixture's DPI, not 96.** System.Drawing saves at the session display DPI (here 120). Compute expectations from the actual DPI.
- **Caption is "Resize with Image Resizer", not the checklist's "Resize images"** (both menus). Hard-match the real caption.
- **Idle auto-lock = BLK-ENV for the disabled-absent + enabled-present items (Recipes 1-2)** (synthetic right-click needs foreground). Disable lock/sleep before the run (`references/environment-setup.md`).
## Fixture files needed
None pre-canned. Generate with `System.Drawing`: a landscape (e.g. 1200×800) and portrait (800×1200) JPEG, a small (100×100) PNG, a square (400×400) PNG, a `.gif` (single frame is fine — the warning is extension-based), and 3 identical images for the multi-file batch.
**18/18 PASS** observed (2026-06-09). All 14 resize-behavior items + the enabled-entry-present-in-both-menus + the remove-size + the gif-warning items cleanly driven via CLI/GUI. The disabled-entry-absent case (in both modern + classic menus, with sibling entries remaining and the entry returning on re-enable) verified live once the desktop was unlocked. NB: an idle auto-lock will turn the menu-presence Recipes 1-2 into BLK-ENV — disable lock/sleep up front (`references/environment-setup.md`).
## Don'ts
- Don't expect `Shell.Application.Verbs()` to list the entry — it's a Win11 packaged command (classic verbs are blind; `CoCreate` → `REGDB_E_CLASSNOTREGISTERED`).
- Don't hardcode 96 DPI for cm/inch math.
- Don't write preset Ids as JSON doubles.
- Don't kill processes by name; use `Stop-Process -Id <pid>`.
- Don't forget to restore `enabled."Image Resizer"=true` + restart runner, and revert any `settings.json`/`sizes.json` edits.
> Read **`../references/explorer-context-menu-flow.md` first** — New+ is a Win11 packaged-IExplorerCommand context-menu module; the menu can only be eyeballed via a real synthetic right-click on an **unlocked interactive desktop**. On a locked/RDP-minimized desktop (`Test-PtDesktopInteractive=False`) all "menu appears / template appears / hidden-caption" assertions are BLK-ENV / BLK-VISUAL-RENDER, not product FAILs.
Flip `enabled.NewPlus` in master `settings.json` + `Restart-PtRunner`, **or** toggle the Settings switch (below). Observe the gate, no foreground needed:
-`btn-hidethefileexte-24a0` (Hide file extension), `btn-hideleadingdigi-24a8` (Hide leading digits). AutomationIds carry a per-session suffix — re-`inspect` to get the live id.
### 3. Synthetic right-click on the folder **BACKGROUND** (the menu-render observer) — needs unlocked desktop
New+ lives in the folder-background ("New") menu, **not** a file's context menu — so `pt-explorer-contextmenu.ps1`'s `Open-PtExplorerContextMenu` (which right-clicks a *file item*) is the wrong entry. Right-click an **empty area of the file list** instead, then expand the `New+` submenu (a separate popup window one level deeper):
```powershell
# force-foreground the CabinetWClass window, GetWindowRect, RightClick at ~45% width / 68% height (empty list area)
# -> main bg menu popup (PopupWindowSiteBridge). Then:
winappuiinvoke$np.selector-w$mainMenuHwnd# expands the New+ submenu
# the submenu is the PopupWindowSiteBridge popup that contains 'Open templates' but NOT 'Sort by'
# enumerate its MenuItems (templates are 1:1 with the Templates-folder entries) / invoke one by name
```
Template items render with **caption transforms applied** (HideFileExtension strips `.txt`; HideStartingDigits strips `01. `). Selecting a template creates it in the current folder + enters rename mode. BLK-ENV only if `Test-PtDesktopInteractive` is False. **No Explorer restart needed** for setting A/B — the handler re-reads `NewPlus\settings.json` on each menu build.
## Recipes — a control/observation map, NOT an answer key
| # | Capability | Drive (control / settings key) | Observe (where the result shows) |
|---|---|---|---|
| 1 | Menu entry present when enabled | enable (master flag + restart, or `btn-new-248c`) | CLSID registered in `HKCU\…\CLSID`, log `context menu registered`; *visible submenu* = synthetic menu only (BLK-VISUAL-RENDER if locked) |
| 2 | Menu entry absent when disabled | disable | CLSID absent, log `context menu unregistered`; package still `Status Ok` |
| 3 | Templates folder created empty | shell ext `create_folder_if_not_exist(root)` on menu build (delete folder → right-click) | folder recreated **empty** — needs synthetic menu (BLK-ENV if locked) |
| 4 | Default templates copied when empty | `CopyTemplateExamples` on Settings-UI **enable** transition (`btn-new-248c` off→on) while folder empty | Templates folder repopulated from install Assets (filesystem — headless-safe) |
| 5 | A template (file/folder) shows + creates on select | put item in Templates folder; select it in the New+ submenu | submenu item (1:1 with dir entries) + `SHFileOperation FO_COPY` to target — synthetic menu only |
| 6 | Hide file extension | `HideFileExtension` / `btn-hidethefileexte-24a0` | strips ext from **menu caption only** (`get_menu_title`, `show_extension=false`); created file keeps ext — caption is BLK-VISUAL-RENDER if locked |
| 7 | Hide starting digits/spaces/dots | `HideStartingDigits` / `btn-hideleadingdigi-24a8` | strips leading digits+separator from **both** menu caption and **created filename** (`remove_starting_digits_from_filename` via `get_menu_title` + `copy_template`); needs a digit-prefixed template + render |
> Verify a setting actually drives behavior by editing the **module-owned** `NewPlus\settings.json` (not the PT-store mirror) and relaunching; the Settings toggles round-trip into this same file.
## Common BLOCKED traps
- **Master-flip + runner restart does not copy default templates** — that's a Settings-UI action (`NewPlusViewModel.IsEnabled`). Use the UIA toggle for any template-auto-copy item.
- **Menu render is invisible without a real right-click** — packaged command is not `CoCreate`-able (`REGDB_E_CLASSNOTREGISTERED`) and not in classic `Shell.Application.Verbs()`. Locked desktop ⇒ BLK-ENV; do not substitute a CLI/back-door (there isn't one, and it'd be a false PASS).
- **No template-count observable** — `saved_number_of_templates` is an in-memory static (`new_utilities.cpp`), not registry/log.
## Fixture files needed
- A plain file (e.g. `test.txt`) and a folder-with-files to drop into Templates (template-appears items).
- A digit-prefixed template (e.g. `01. Test.txt`) to exercise Hide-starting-digits.
-`src/settings-ui/Settings.UI/ViewModels/NewPlusViewModel.cs` — `CopyTemplateExamples` (creates dir; copies examples only when files==0 && dirs==0), called from `IsEnabled` setter / `OpenNewTemplateFolder` / `DashboardViewModel`.
- Locked/non-interactive desktop: ~3/9 (registration-gate present/absent + template auto-copy); menu-render/select items fall to BLK-ENV/BLK-VISUAL-RENDER — re-run after unlocking.
## Don'ts
- Don't edit `…\PowerToys\New\settings.json` — wrong path; the file is under `NewPlus\`.
- Don't use `Open-PtExplorerContextMenu` (file-item right-click) for New+ — it's the folder **background** ("New") menu; right-click empty list space instead.
- Don't forget to **expand the `New+` submenu** (invoke it) before enumerating templates — they live one popup deeper than the main menu.
- Don't mark menu-render items as product FAIL on a locked desktop — it's BLK-ENV.
- Don't restart Explorer to apply a setting change — the handler re-reads `NewPlus\settings.json` per menu build.
**Source**: `Peek.UI\PeekXAML\App.xaml.cs:106-134` — when last arg is not int (=runner PID) and is an existing file, it sets `_launchedFromCli=true`, builds `SelectedItemByPath`, calls `OnShowPeek()`. Bypasses hotkey + Explorer foreground.
**Use for**: single-file previewer rendering tests (Recipes 1-2) and the CLI-accepts-path assertion (Recipe 8).
**Cannot use for**: navigation tests (Recipes 4-7, 10-11) — source has `if (_isFromCli) return;` guard that disables arrow navigation, and CLI mode spawns a fresh process every call (no pin-state-across-reopen).
### 2. Shell.Application COM + Ctrl+Space — Explorer-driven, supports navigation
This is the canonical "do what a real user would do" path that drives all the navigation/pin tests.
```powershell
# Dot-source helpers first
."$skill\scripts\pt-explorer-com.ps1"
."$skill\scripts\pt-sendinput-chord.ps1"
# Set up multi-file selection in Explorer + trigger Peek in one call:
**Requires**: interactive desktop session (`Test-PtInteractiveDesktop` must show both `ForegroundOk=True` and `ShellComOk=True`).
### 3. Named Event signal — quick smoke
```powershell
Invoke-PtSharedEvent-Name'Peek.Show'
```
Wakes the resident Peek process (different from CLI back-door — respects current Explorer foreground selection). Used by some framework tests for the "Peek is enabled and listening" assertion.
## Recipes — a control/observation map, NOT a per-test-case answer key
> Maps each Peek *capability* to **how to drive it** and **where the result shows**. It does NOT prescribe concrete fixtures/coords/inputs or expected values — design those at runtime from the actual checklist item. Only a real UI/behavior change should force an edit here.
| # | Capability | Drive (control / command) | Observe (where the result shows) |
|---|---|---|---|
| 1 | File-type previewer renders (image / text+code / markdown / PDF / HTML / archive / unsupported) | `Peek.UI.exe <fixture>` (entry-path 1) → `winapp ui inspect -w <hwnd> --depth 7` | the type's previewer node present (`ImagePreview Image`; `PreviewBrowser Pane` for dev/text/md/HTML; archive tree for zip; File-Type/Size/Date view for unsupported). Prefer `winapp ui search` for an in-fixture marker over OCR |
| 2 | "Open with default app" via button | `winapp ui invoke LaunchAppButton` | a new editor process/window for `<file>` appears (PID diff) |
| 3 | "Open with default app" via Enter | `Assert-PtForegroundOrAbort` → `Send-PtChord -Key <Enter>` | same as #2 |
| 4 | Pin keeps window position when switching files | Shell COM + Ctrl+Space (entry-path 2) → `winapp ui invoke PinButton` → move window → navigate to next file | window stays at the pinned coordinates |
| 5 | Pin position persists across close + reopen | pinned → Esc to close (graceful — **don't `Stop-Process`**, it bypasses the pin-save handler) → reopen via Shell COM + Ctrl+Space | new window opens at the same pinned coordinates |
| 6 | Unpin releases the lock; switching file reverts to default | `winapp ui invoke PinButton` again (unpin) → navigate | window moves to the default position |
| 7 | Unpinned reopen uses default position | unpinned → Esc-close → reopen | new window at default, not the stale pinned coords |
| 8 | `Peek.UI.exe <file>` CLI opens Peek | entry-path 1 | covered by #1 across file types |
| 9 | Concurrent Peek sessions don't crash/interfere | launch `Peek.UI.exe` several times on different files, leaving windows open | each spawns its own process/window; no error in `Peek\Logs` |
| 10 | Arrow keys cycle between selected files | Shell COM multi-file selection → Ctrl+Space → `Send-PtChord` Right/Left | window title updates to each file in sequence, wraps at the ends |
| 11 | Multi-file selection scopes navigation | select a subset of a folder → navigate | only the selected files cycle, not the rest |
| 12 | Activation-hotkey reassignment takes effect | edit `Peek\settings.json``properties.ActivationShortcut` → `Restart-PtRunner` (**not hot-reloaded** — see Gotchas) → press the new chord, then the old chord | new chord opens Peek; old chord does nothing |
> **Mapping process**: read the actual checklist item → identify the capability → find its row → drive the named control and design your own inputs + assertions for *that* item. If no row matches, it's a NEW capability — drive ad-hoc and add a row (capability + control + observation point; no canned inputs).
## BLOCKED triage (single source of truth)
If the agent only tried the CLI back-door and marked the pin / navigation tests BLOCKED → **misdiagnosis**, try entry-path #2 (Shell.Application COM + Ctrl+Space).
If the agent tried Shell COM + Ctrl+Space and got `GetForegroundWindow()=0` + `SendInput → ACCESS_DENIED (5)` → **environment**, not framework. The session has no attached input desktop (RDP minimized, screen locked, screensaver, etc.). See `SKILL.md` pitfall #13 and `references/environment-setup.md` for the per-scenario table + powercfg setup commands. Mark BLK-ENV with mitigation citation.
Both traps were observed in 2026-06-08 sign-off runs; preventing both is now the agent's pre-flight job (`pt-session-diagnose.ps1`).
## Fixture files needed
Put these in a workspace `fixtures/` folder before starting:
-`small-image.png` (any 200x150 PNG)
-`Program.cs` (any C# file)
-`readme.md` (markdown with H1 + bold + bullet list)
-`test-pdf.pdf` (PDF with embedded text "PDF_FIXTURE_OK" + "PDF_MARKER_42")
-`page.html` (HTML with `<h1>` containing "HTMLPEEKMARKER")
-`archive.zip` (zip containing 1 small text file)
-`unsupported.xyz` (any small binary)
- 3 differently-sized images for the pin-position tests (e.g. 320x240, 800x600, 1920x1080)
**18/18 = 100%** achievable from a normal interactive admin console session (verified 2026-06-08). The change-shortcut item is PASS-able via the settings.json + runner-restart path — see Recipe 12.
## Peek-specific gotchas
- **Activation-shortcut is NOT hot-reloaded.** Editing `Peek\settings.json``ActivationShortcut` and waiting for the file-watcher debounce does nothing — the centralized keyboard hook only re-registers the chord after `Restart-PtRunner`. Restart after the change AND again after restoring.
- **PinButton spawns a `PopupHost` teaching-tip.** Invoking `PinButton` pops a small confirmation flyout (≈192x63) titled `PopupHost` that surfaces *first* in `winapp ui list-windows`. A naive "first HWND" regex grabs the popup, not Peek. Match by title suffix `- Peek` (regex like `HWND (\d+): "([^"]*- Peek)"`) and/or cache the original Peek HWND before invoking PinButton.
- **Win11 Notepad tabs/session-restore** muddy the "open-with-default-app" tests (Recipes 2-3): the spawned Notepad restores prior tabs, so the foreground Notepad's title may not show your file. Enumerate all Notepad windows and match `"<file> - Notepad"` explicitly.
## Don'ts
- **Don't `Stop-Process PowerToys.Peek.UI -Force`** to close Peek between iterations — bypasses the save handler, breaks the pin-state-persistence tests (Recipes 5, 7). Use Esc / `winapp ui invoke CloseButton`.
- **Don't assume CLI back-door supports navigation** — it doesn't (`_isFromCli` guard). For nav tests use Shell COM + Ctrl+Space.
- **Don't OCR the previewer surface** when UIA already exposes the correct nodes (`ImagePreview`, `PreviewBrowser`, `LaunchAppButton`, `PinButton`). UIA is more reliable than OCR.
**Activation**: Explorer right-click → "Rename with PowerRename" (Win11 Tier-1 menu; **no classic HKCR verb on Win11**); optional global hotkey if user-configured
For the synthetic-right-click + context-menu-invoke flow that ALL Explorer-context-menu modules use, see **`references/explorer-context-menu-flow.md`** + **`scripts/pt-explorer-contextmenu.ps1`** (`Test-PtDesktopInteractive`, `Open-PtExplorerContextMenu`, `Invoke-PtContextMenuItem`, `Get-PtContextMenuItems`). That doc covers stability rules, multi-file selection, BLK-ENV handling, and module-caption table. Don't duplicate; cite by section.
For the Win11 IExplorerCommand vs classic HKCR distinction, see `scripts/pt-shell-verbs.ps1` header — PR is **modern-menu-only on Win11**, so classic-verb enumeration via Shell.Application **will not find it**.
## Entry-paths (try in order)
### 1. Direct CLI launch with file args — PREFERRED for UI-driven tests (verified 2026-06-10)
Bypasses the context menu entirely; same code path inside the exe (it parses argv as the file list). **Use for every UI-driven option/regex/preview test** (Recipes 4-12 below).
Use the canonical flow from `references/explorer-context-menu-flow.md` Recipe. The menu-presence assertion is the ONE thing the CLI back-door cannot prove (it works even if the menu entry is correctly hidden — the false-positive trap described in that doc).
Returns False on Win11 because PT registers PR only via IExplorerCommand, not as a classic HKCR shell verb. **Use only for negative checks** (and prefer the synthetic-menu enumeration above, which observes the actual Tier-1 menu).
## Recipes — a control/observation map, NOT a per-test-case answer key
> **What this table is (and isn't):** it maps each PowerRename *capability* to **which control drives it** (AutomationId / settings key) and **where the result shows up**. It deliberately does **NOT** prescribe specific Search/Replace inputs or expected-output assertions — those are the agent's job to design from the actual checklist item at runtime. Keeping it input/assertion-free means the table survives checklist-wording changes; only a real UI redesign (renamed/moved control) should force an edit here (as happened to rows 5 & 12 in build 0.100.0).
| # | Capability | Drive (control / settings key) | Observe (where the result shows) |
|---|---|---|---|
| 1 | Context-menu entry present when enabled, gone when disabled | master `enabled.PowerRename` flip + `Restart-PtRunner`; synthetic menu (entry-path 2) | `Get-PtContextMenuItems` includes / excludes "Rename with PowerRename" |
| 2 | "Show icon on context menu" | `ShowIcon` in `power-rename-settings.json` + relaunch | menu entry shows icon vs text-only (screenshot); or HKCR `Icon` |
| 3 | "Appear only in extended menu" | `ExtendedContextMenuOnly` + relaunch | Tier-1 menu hides PR; classic "Show more options" still lists it |
| 9 | Datetime tokens | Replace accepts `$DD``$MMMM``$YYYY``$hh``$mm``$ss``$fff` | preview value matches `(Get-Item <file>).CreationTime` formatted the same way |
| 10 | Boost library (Perl regex beyond .NET, e.g. lookbehind) | `UseBoostLib` — **read at process start; relaunch PR after toggling** | the Perl-only pattern matches in the preview without error |
| 11 | Per-row include/exclude in the preview | invoke a row checkbox to uncheck | the unchecked file is unchanged on disk after Rename |
| 12 | Filter preview / select-all (NOT a column-header click — headers `TxtBlock_Original`/`TxtBlock_Renamed` are non-interactive labels) | `btn-filter-XXXX` → `button_showAll` / `button_showRenamed`; `checkBox_selectAll` | visible row set shrinks/grows; all rows toggle on/off |
> **Mapping process**: read the actual checklist item → identify the capability → find its row → drive the named control and design your own inputs + assertions for *that* item. If no row matches, it's a NEW capability — drive it ad-hoc and add a row (capability + control + observation point, no canned inputs).
- subfolder `subdir/` with 2 inner files — folder/subfolder exclusion
-`Foo_A_A_A.txt` — match-all
-`MIXED.txt` — case-sensitive
Always copy fixtures to a disposable temp folder before running actual rename operations.
## Gotchas
- **TWO settings files — PR reads `power-rename-settings.json`, NOT `settings.json`** (verified 2026-06-10). `%LOCALAPPDATA%\Microsoft\PowerToys\PowerRename\` holds both: (1) `settings.json` = PT-store, keys `bool_mru_enabled`/`bool_persist_input`/`bool_show_icon_on_menu`/`bool_show_extended_menu`/`bool_use_boost_lib`/`int_max_mru_size` (what `Get-PtModuleSettings` + the Settings UI bind to); (2) `power-rename-settings.json` = the module's own store, keys `ShowIcon`/`ExtendedContextMenuOnly`/`PersistState`/`MRUEnabled`/`MaxMRUSize`/`UseBoostLib` — **this is the file the PR UI exe and the context-menu COM handlers actually read at launch** (`lib/Settings.cpp``CSettings::Load→ParseJson`). The runner (`dll/dllmain.cpp:301-307`) syncs PT-store→module-store only on a Settings-UI *change event*; the PT-store file can sit stale for days. **To drive ShowIcon / ExtendedContextMenuOnly / MRUEnabled / PersistState / UseBoostLib deterministically, edit `power-rename-settings.json` directly + relaunch PR (or restart runner+Explorer for the menu handlers), then restore.** Map (settings.json key → user-facing toggle): ShowIcon→"Show icon on context menu", ExtendedContextMenuOnly→"Appear only in extended menu", MRUEnabled→autocomplete, PersistState→"Show values from last use", UseBoostLib→"Use Boost library". MRU values live in `search-mru.json`/`replace-mru.json`; last-used (persist) in `power-rename-last-run-data.json`.
- **"Show icon on context menu" has no Settings-UI toggle in current builds** — drive it via `power-rename-settings.json``ShowIcon`. Behavior is observable on the synthetic menu (icon vs text-only); source `PowerRenameContextMenu/dllmain.cpp:73` (`GetIcon→null`).
- **The "Appear only in extended menu" classic `#32768` popup is not winapp-enumerable** — assert the Tier-1 *hide* (observed; `dllmain.cpp:108``ECS_HIDDEN`) and cite `PowerRenameExt.cpp:84` (`E_FAIL` unless `CMF_EXTENDEDVERBS`) for the "still in extended menu" half.
- **PR registers on the directory *background* menu too** — the synthetic right-click often lands on background (View/Sort by/Group by/...) yet still shows/hides `Rename with PowerRename`, which is a valid, stable surface for menu-entry / icon-visibility / extended-menu-only present-absent comparisons.
- **`set-value` on search/replace DOES fire the preview** (TextChanged works, unlike CmdPal) — Apply button enabling/disabling is a reliable match/no-match signal. The search/replace Edit AutomationIds are random per launch (`txt-textbox-XXXX`); discover them each launch by name (`Edit "Search for"` / `Edit "Replace with"`).
- **Preview-row uncheck + column-header invokes need the Preview populated first** — set Search/Replace and wait ~500 ms for the regex engine; otherwise the invokes hit an empty list.
- **Boost library is read at PR process start** — close + relaunch PR after toggling.
- **Icon-on-menu and extended-only checks prefer registry over screenshot** — read HKCR `Extended` / `Icon` REG_SZ; more reliable + locale-independent.
- **Disk mutation is real** — run renames against `$env:TEMP\pr-test-<random>`, not real fixtures.
- **COM cache staleness** when re-checking verbs after enable/disable — call `Reset-PtShellComCache` from `scripts/pt-shell-verbs.ps1`.
## Source citations
-`<PT-repo>\src\modules\PowerRename\dllmain.cpp` — IExplorerCommand registration (no classic HKCR shadow on Win11).
-`<PT-repo>\src\modules\PowerRename\PowerRenameUILib\` — XAML for main PR window (toggle/checkbox AutomationIds).
Expected **18/18 = 100%** from an interactive admin console session. Direct-CLI (#1) covers UI-driven items; synthetic-menu (#2) covers menu-presence assertions.
## Don'ts
- **Don't** try `Invoke-PtShellVerb 'PowerRename'` — returns False on Win11 (no classic registration). Use synthetic menu via `Invoke-PtContextMenuItem` or direct-CLI.
- **Don't** run rename operations against reusable fixtures — copy to a disposable temp folder.
- **Don't** trust screenshot-only for icon-on-menu or extended-only checks — registry inspection is faster + locale-independent.
- **Don't** skip the synthetic-menu test for the menu-presence assertion — CLI back-door PASSes even when the menu entry is correctly hidden (false-positive trap described in `references/explorer-context-menu-flow.md`).
This doc covers the **agent-runtime** environment probing and lifecycle hooks. Read alongside `SKILL.md` (the playbook) and `references/environment-setup.md` (one-time user env prep).
## Pre-flight checks (do these first; abort if any fails)
1.**Admin check** — `Test-PtAdmin` must return the elevation level matching `[ADMIN: YES]` items in the module's checklist. If the module contains `[ADMIN: YES]` items and `Test-PtAdmin` returns `False`, **STOP** and tell the user "this module requires an elevated session". Do NOT silently mark those items BLOCKED-LACK-ADMIN — that hides a fixable env issue.
2.**PT runner present** — `Test-PtRunnerAdmin` should show the runner exists. If it doesn't exist, start PowerToys (`Start-Process "$env:LOCALAPPDATA\PowerToys\PowerToys.exe"`).
4.**Interactive-desktop availability + session attachment** — the single most common cause of false-BLOCKED reports is a session mismatch where the agent runs in an elevated **non-console session** (e.g. RDP that's been disconnected/minimized, fast user switching, run-as-different-user, or scheduled-task-with-highest-privilege). In that scenario `Test-PtAdmin=True` but `GetForegroundWindow()=0` and `SendInput` returns `ERROR_ACCESS_DENIED (5)` — input injection cannot reach the active desktop.
$consoleSession = (Get-Process explorer -EA SilentlyContinue | Select-Object -First 1).SessionId
"Agent session=$agentSession Console explorer session=$consoleSession"
# Foreground + Shell COM probe (use scripts/pt-session-diagnose.ps1 for the full version)
Add-Type 'using System; using System.Runtime.InteropServices; public class FG4 { [DllImport("user32.dll")] public static extern IntPtr GetForegroundWindow(); }'
- **ShellComOk only (ForegroundOk false)** → Non-interactive (e.g. Session ≠ console, RDP minimized, screen locked, screensaver). Only schema / UIA-invoke / CLI / Named-Event tests work. Mark input-injection items as `BLK-ENV` and **cite `references/environment-setup.md` in the report** so the user can fix env and re-run.
- **Neither (ShellComOk false)** → Session 0 / service context — even Shell COM fails. Very few tests possible.
5. **Discipline: try AT LEAST 2 distinct entry-paths before marking BLOCKED.** For Peek/FZ/Workspaces/Image Resizer/PowerRename/File Locksmith specifically, the obvious entry-path is the global hotkey but Shell.Application COM driving Explorer also works — see per-module profiles under `references/modules/`. Marking BLOCKED after trying only the CLI launch (a common trap) hides easily-PASS-able items in an interactive session.
## Bootstrap (paste at start of your verification script)
```powershell
$skill = '<this skill folder>' # the folder containing SKILL.md
## Final wrap-up (run AFTER all per-item tables are written)
1. **Run state-hygiene cleanup** above for everything that wasn't restored per-item.
2. **Write the top-of-report summary** per `references/reporting-format.md` §B.
3. **Write the §G Retrospective** — reflect on the run itself: every friction (classified by source + severity + minutes/attempts cost + suggested fix), or `Everything was smooth — no friction encountered.` See `references/reporting-format.md` §G. Don't skip it; it's how the skill improves.
4. **Verify every screenshot referenced in the report actually exists on disk** (before the move, while paths still resolve under `$workspace`):
The report uses **relative** `artifacts/…` paths, so the whole tree moves intact.
6. **Print the FINAL (moved) report path** as the very last line of your response — the `…\Module-Signoff\verify-<Module>-<timestamp>\verify-<Module>.md` path, NOT the temp path.
## Hard rules
- **Never silently send keys via SendInput** to a target window without first calling `Assert-PtForegroundOrAbort -AppId <id>`. Keys silently leak to your terminal if the target isn't foreground.
- **Never mark BLOCKED without trying at least 2 distinct entry-paths from the drive-stack** (SKILL.md §2). If you can't drive the item, name the specific obstacle (not "I can't").
- **Never assume any external repo is cloned locally.** The helpers under `scripts/` are self-contained. Use `Test-Path` guards before referencing any external path.
- **Never invent test steps for a `[CLARITY: VAGUE-*]` item** — mark it **FAIL (cause: checklist-ambiguous)** and quote the original wording so the user can fix the checklist. The checklist is test code; an undefinable test is a broken test.
- **Always restore state** before exiting (even on error). State hygiene wraps every mutation in try/finally.
- **Separate the two FAIL causes**: *product* FAILs are bugs to file; *checklist* FAILs (stale feature or ambiguous spec) are items to rewrite/prune. If a large share of a module's items are checklist-FAILs, the checklist needs an overhaul before re-verifying — don't punt drivable items into a FAIL.
- **Never continue past 3 consecutive errors against the same item** — mark it BLOCKED with the concrete symptom/obstacle and move on. Per-item budget is ~5 minutes; if stuck longer, it's BLOCKED (name the wall).
-`[ADMIN: COND]` - conditional - the basic case is non-admin but specific sub-cases require admin (e.g. "test with elevated target app", "Restart as admin" variants)
**Clarity**:
- (no marker) - clear, has explicit assert
-`[CLARITY: VAGUE-NO-STEPS]` - original wording is just a module/feature name without procedural steps
-`[CLARITY: VAGUE-NO-ASSERT]` - original wording describes an action but does not state the expected outcome
-`[CLARITY: VAGUE-AMBIGUOUS]` - original wording uses vague verbs like "works" without a measurable outcome
-`[REWRITTEN]` - original wording was vague; this checklist has rewritten the description to be concrete. Original wording preserved in italics below the item.
---
## Environment Variables (20 items)
- [ ]**[ADMIN: YES]** (L791) Launch as administrator ON - Launch Environment Variables and confirm that SYSTEM variables ARE editable and Add variable button is enabled
- [ ]**[ADMIN: YES]** (L792) Launch as administrator OFF - Launch Environment Variables and confirm that SYSTEM variables ARE NOT editable and Add variable button is disabled
- [ ]**[ADMIN: NO]** (L795) Add new User variable. Open OS Environment variables window and confirm that added variable is there. Also, confirm that it's added to "Applied variables" list.
- [ ]**[ADMIN: NO]** (L796) Edit one User variable. Open OS Environment variables window and confirm that variable is changed. Also, confirm that change is applied to "Applied variables" list.
- [ ]**[ADMIN: NO]** (L797) Remove one User variable. Open OS Environment variables window and confirm that variable is removed. Also, confirm that variable is removed from "Applied variables" list.
- [ ]**[ADMIN: NO]** (L801) Add new profile with no variables and name it "Test_profile_1" (referenced below by name)
- [ ]**[ADMIN: NO]** (L802) Edit "Test_profile_1": Add one new variable to profile e.g. name: "profile_1_variable_1" value: "profile_1_value_1"
- [ ]**[ADMIN: NO]** (L803) Add new profile "Test_profile_2": From "Add profile dialog" add two new variables (profile_2_variable_1:profile_2_value_1 and profile_2_variable_2:profile_2_value_2). Set profile to enabled and click Save. Open OS Environment variables window and confirm that all variables from the profile are applied correctly. Also, confirm that "Applied variables" list contains all variables from the profile.
- [ ]**[ADMIN: NO]** (L804) Apply "Test_profile_1" while "Test_profile_2" is still aplpied. Open OS Environment variables window and confirm that all variables from Test_profile_2 are unapplied and that all variables from Test_profile_1 are applied. Also, confirm that state of "Applied variables" list is updated correctly.
- [ ]**[ADMIN: NO]** (L805) Unapply applied profile. Open OS Environment variables window and confirm that all variables from the profile are unapplied correctly. Also, confirm that "Applied variables" list does not contain variables from the profile.
- [ ]**[ADMIN: NO]** (L808) To "Test_profile_1" add one existing variable from USER variables, e.g. TMP. After adding, change it's value to e.g "test_TMP" (or manually add variable named TMP with value test_TMP).
- [ ]**[ADMIN: NO]** (L809) Apply "Test_profile_1". Open OS Environment variables window and confirm that TMP variable in USER variables has value "test_TMP". Confirm that there is backup variable "TMP_PowerToys_Test_profile_1" with original value of TMP var. Also, confirm that "Applied variables" list is updated correctly - there is TMP profile variable, and backup User variable..
- [ ]**[ADMIN: NO]** (L810) Unapply "Test_profile_1". Open OS Environment variables window and confirm that TMP variable in USER variable has original value and that there is no backup variable. Also, confirm that "Applied variables" list is updated correctly.
- [ ]**[ADMIN: NO]** (L813) In "Applied variables" list confirm that PATH variable is shown properly: value of USER Path concatenated to the end of SYSTEM Path.
- [ ]**[ADMIN: NO]** (L814) To "Test_profile_1" add variable named PATH with value "path1;path2;path3" and click Save. Confirm that PATH variable in profile is shown as list (list of 3 values and not as path1;path2;path3).
- [ ]**[ADMIN: NO]** (L815) Edit PATH variable from "Test_profile_1". Try different options from ... menu (Delete, Move up, Move down, etc...). Click Save.
- [ ]**[ADMIN: NO]** (L816) Apply "Test_profile_1". Open OS Environment variables window and confirm that profile is applied correctly - Path value and backup variable. Also, in "Applied variables" list check that Path variable has correct value: value of profile PATH concatenated to the end of SYSTEM Path.
- [ ]**[ADMIN: NO]** (L819) Close the app and reopen it. Confirm that the state of the app is the same as before closing.
- [ ]**[ADMIN: NO]** (L821) "Test_profile_1" should still be applied (if not apply it). Delete "Test_profile_1". Confirm that profile is unapplied (both in OS Environment variables window and "Applied variables" list).
- [ ]**[ADMIN: NO]** (L822) Delete "Test_profile_2". Check profiles.json file and confirm that both profiles are gone.
-`[ADMIN: COND]` - conditional - the basic case is non-admin but specific sub-cases require admin (e.g. "test with elevated target app", "Restart as admin" variants)
**Clarity**:
- (no marker) - clear, has explicit assert
-`[CLARITY: VAGUE-NO-STEPS]` - original wording is just a module/feature name without procedural steps
-`[CLARITY: VAGUE-NO-ASSERT]` - original wording describes an action but does not state the expected outcome
-`[CLARITY: VAGUE-AMBIGUOUS]` - original wording uses vague verbs like "works" without a measurable outcome
-`[REWRITTEN]` - original wording was vague; this checklist has rewritten the description to be concrete. Original wording preserved in italics below the item.
---
## File Locksmith (10 items)
- [ ]**[ADMIN: COND]** (L641) Right-click the executable file, select "Unlock with File Locksmith" and verify it shows up. (2 entries will show, since the installer starts two processes)
- [ ]**[ADMIN: COND]** (L642) End the tasks in File Locksmith UI and verify that closes the installer.
- [ ]**[ADMIN: COND]** (L643) Start the installer executable again and press the Refresh button in File Locksmith UI. It should find new processes using the files.
- [ ]**[ADMIN: COND]** (L644) Close the installer window and verify the processes are delisted from the File Locksmith UI. Close the window
- [ ]**[ADMIN: COND]** (L646) Right click the directory where the executable is located, select "Unlock with File Locksmith" and verify it shows up.
- [ ]**[ADMIN: COND]** (L647) Right click the drive where the executable is located, select "Unlock with File Locksmith" and verify it shows up. You can close the PowerToys installer now.
- [ ]**[ADMIN: COND]** (L649) Right click "Program Files", select "Unlock with File Locksmith" and verify "PowerToys.exe" doesn't show up.
- [ ]**[ADMIN: YES]** (L650) Press the File Locksmith "Restart as an administrator" button and verify "PowerToys.exe" shows up.
- [ ]**[ADMIN: YES]** (L651) Right-click the drive where Windows is installed, select "Unlock with File Locksmith" and scroll down and up, verify File Locksmith doesn't crash with all those entries being shown. Repeat after clicking the File Locksmith "Restart as an administrator" button.
- [ ]**[ADMIN: COND]** (L652) Disable File Locksmith in Settings and verify the context menu entry no longer appears.
-`[ADMIN: COND]` - conditional - the basic case is non-admin but specific sub-cases require admin (e.g. "test with elevated target app", "Restart as admin" variants)
**Clarity**:
- (no marker) - clear, has explicit assert
-`[CLARITY: VAGUE-NO-STEPS]` - original wording is just a module/feature name without procedural steps
-`[CLARITY: VAGUE-NO-ASSERT]` - original wording describes an action but does not state the expected outcome
-`[CLARITY: VAGUE-AMBIGUOUS]` - original wording uses vague verbs like "works" without a measurable outcome
-`[REWRITTEN]` - original wording was vague; this checklist has rewritten the description to be concrete. Original wording preserved in italics below the item.
---
## Image Resizer (18 items)
- [ ]**[ADMIN: NO]** (L309) Disable the Image Resizer and check that `Resize with Image Resizer` is absent in the context menu
- [ ]**[ADMIN: NO]** (L310) Enable the Image Resizer and check that `Resize with Image Resizer` is present in the context menu (both Win11 modern and old menus)
- [ ]**[ADMIN: NO]** (L311) Remove one image size and add a custom image size. Open the Image Resize window from the context menu and verify changes are populated
- [ ]**[ADMIN: NO] [CLARITY: VAGUE-NO-STEPS]** (L312) Resize one image
One file per module; a verification run loads only its module's file.
> **Scope:** only modules that have been verified end-to-end (with a sign-off report) are checked in here so far. The remaining modules' checklists will be added as each is verified.
-`[ADMIN: COND]` - conditional - the basic case is non-admin but specific sub-cases require admin (e.g. "test with elevated target app", "Restart as admin" variants)
**Clarity**:
- (no marker) - clear, has explicit assert
-`[CLARITY: VAGUE-NO-STEPS]` - original wording is just a module/feature name without procedural steps
-`[CLARITY: VAGUE-NO-ASSERT]` - original wording describes an action but does not state the expected outcome
-`[CLARITY: VAGUE-AMBIGUOUS]` - original wording uses vague verbs like "works" without a measurable outcome
-`[REWRITTEN]` - original wording was vague; this checklist has rewritten the description to be concrete. Original wording preserved in italics below the item.
---
## New+ (9 items)
- [ ]**[ADMIN: NO]** (L969) Verify NewPlus menu is in Explorer context menu. (Windows 11 tier 1 context menu only. May need Explorer restart.)
- [ ]**[ADMIN: NO]** (L971) Verify NewPlus menu is not in Explorer context menu.
- [ ]**[ADMIN: NO]** (L973) Verify the folder is created and empty.
- [ ]**[ADMIN: NO]** (L974) Copy a file to the templates folder, verify it's added to the New+ context menu and that if you select it the file is created.
- [ ]**[ADMIN: NO]** (L975) Copy a folder with files inside to the templates folder, verify it's added to the New+ context menu and that if you select it the folder and files inside are created.
- [ ]**[ADMIN: NO]** (L976) Delete all files and folders from inside the templates folder. Verify that no templates are available in the context menu.
- [ ]**[ADMIN: NO]** (L977) Disable and re-Enable New+ while the templates folder is still empty. Verify the default templates were copied over and are available in the context menu.
- [ ]**[ADMIN: NO]** (L979) Test the "Hide template filename extension" option in Settings.
- [ ]**[ADMIN: NO]** (L980) Test the "Hide template filename starting digits, spaces and dots" option in Settings.
-`[ADMIN: COND]` - conditional - the basic case is non-admin but specific sub-cases require admin (e.g. "test with elevated target app", "Restart as admin" variants)
**Clarity**:
- (no marker) - clear, has explicit assert
-`[CLARITY: VAGUE-NO-STEPS]` - original wording is just a module/feature name without procedural steps
-`[CLARITY: VAGUE-NO-ASSERT]` - original wording describes an action but does not state the expected outcome
-`[CLARITY: VAGUE-AMBIGUOUS]` - original wording uses vague verbs like "works" without a measurable outcome
-`[REWRITTEN]` - original wording was vague; this checklist has rewritten the description to be concrete. Original wording preserved in italics below the item.
- [ ]**[ADMIN: NO]** (L703) Any other not mentioned file (.exe for example) to verify the unsupported file view is shown
- [ ]**[ADMIN: NO]** (L706) Pin the window, switch between images of different size, verify the window stays at the same place and the same size.
- [ ]**[ADMIN: NO]** (L707) Pin the window, close and reopen Peek, verify the new window is opened at the same place and the same size as before.
- [ ]**[ADMIN: NO]** (L708) Unpin the window, switch to a different file, verify the window is moved to the default place.
- [ ]**[ADMIN: NO]** (L709) Unpin the window, close and reopen Peek, verify the new window is opened on the default place.
- [ ]**[ADMIN: NO]** (L712) By clicking a button.
- [ ]**[ADMIN: NO]** (L713) By pressing enter.
- [ ]**[ADMIN: NO]** (L716) Can use peek command to peek files
- [ ]**[ADMIN: NO] [CLARITY: VAGUE-NO-STEPS]** (L717) Peek can work without problem when a peek session is on
- [ ]**[ADMIN: NO]** (L719) Switch between files in the folder using `LeftArrow` and `RightArrow`, verify you can switch between all files in the folder.
- [ ]**[ADMIN: NO]** (L720) Open multiple files, verify you can switch only between selected files.
- [ ]**[ADMIN: NO] [CLARITY: VAGUE-AMBIGUOUS]** (L721) Change the shortcut, verify the new one works.
-`[ADMIN: COND]` - conditional - the basic case is non-admin but specific sub-cases require admin (e.g. "test with elevated target app", "Restart as admin" variants)
**Clarity**:
- (no marker) - clear, has explicit assert
-`[CLARITY: VAGUE-NO-STEPS]` - original wording is just a module/feature name without procedural steps
-`[CLARITY: VAGUE-NO-ASSERT]` - original wording describes an action but does not state the expected outcome
-`[CLARITY: VAGUE-AMBIGUOUS]` - original wording uses vague verbs like "works" without a measurable outcome
-`[REWRITTEN]` - original wording was vague; this checklist has rewritten the description to be concrete. Original wording preserved in italics below the item.
---
## PowerRename (18 items)
- [ ]**[ADMIN: NO] [CLARITY: VAGUE-AMBIGUOUS]** (L393) Check if disable and enable of the module works. (On Win11) Check if both old context menu and Win11 tier1 context menu items are present when module is enabled.
- [ ]**[ADMIN: NO]** (L394) Check that with the `Show icon on context menu` icon is shown and vice versa.
- [ ]**[ADMIN: NO] [CLARITY: VAGUE-AMBIGUOUS]** (L395) Check if `Appear only in extended context menu` works.
- [ ]**[ADMIN: NO] [CLARITY: VAGUE-NO-STEPS]** (L401) Item Name/Extension Only (one at the time)
- [ ]**[ADMIN: NO]** (L402) Enumerate Items. Test advanced enumeration using different values for every field ${start=10,increment=2,padding=4}.
- [ ]**[ADMIN: NO] [CLARITY: VAGUE-NO-STEPS]** (L403) Case Sensitive
- [ ]**[ADMIN: NO]** (L404) Match All Occurrences. If checked, all matches of text in the `Search` field will be replaced with the Replace text. Otherwise, only the first instance of the `Search` for text in the file name will be replaced (left to right).
- [ ]**[ADMIN: NO]** (L406) Search with an expression (e.g. `(.*).png`)
- [ ]**[ADMIN: NO]** (L407) Replace with an expression (e.g. `foo_$1.png`)
- [ ]**[ADMIN: NO] [CLARITY: VAGUE-NO-STEPS]** (L408) Replace using file creation date and time (e.g. `$hh-$mm-$ss-$fff``$DD_$MMMM_$YYYY`)
- [ ]**[ADMIN: NO]** (L409) Turn on `Use Boost library` and test with Perl Regular Expression Syntax (e.g. `(?<=t)est`)
- [ ]**[ADMIN: NO]** (L411) In the `preview` window uncheck some items to exclude them from renaming.
- [ ]**[ADMIN: NO]** (L412) Use the **Filter** (funnel) button above the file list → choose "Only show files that will be renamed" / "Show all files" to filter the preview.
- [ ]**[ADMIN: NO]** (L413) Use the **Select/deselect all** checkbox above the file list to toggle all rows checked/unchecked.
This doc defines the **required** report shape for every per-module verification run. Modeled on `PR-validation\Round1\PR-47211-validation\report.md` style — table-driven, reproducible, no prose narratives.
## §A — Per-item table (one per checklist item)
```markdown
## Item L<line_num> — <verbatim description from the module's checklist> — **<PASS|FAIL|BLOCKED>** <emoji>
-`artifacts/L<line>/step-02-<name>.txt` — full inspect dump
- ...
### Verdict reasoning
- ✅ <assertion 1 that PASSed, with reference to the line of code / settings key / log line that proves it>
- ✅ <assertion 2>
- ❌ <if BLOCKED, the specific obstacle: "BLK-HARDWARE because MWB needs 2 physical PCs; this session has 1 ([System.Windows.Forms.Screen]::AllScreens.Count = 1)">
### Caveats (optional)
- <Any deviation from the user-documented flow, e.g. "Tested via settings.json write rather than UI checkbox because SelectionItemPattern.Select clobbers other selections in ListView.">
```
## §B — Top-of-report summary (write LAST, after all per-item tables)
## Retrospective (self-reflection on the run — write LAST)
<Per §G. If the whole run was frictionless, write exactly: **Everything was smooth — no friction encountered.**>
```
## §C — Required rules for step tables
1.**Every `winapp ui ...` command goes in the "winapp / probe commands" cell, verbatim, in backticks**, including `-w <hwnd>` / `-a <appId>` arguments and full selector strings. Reviewers will paste these into their own shell to reproduce.
2.**Every screenshot path goes in the "Evidence" cell** of the step that produced it, formatted as `screenshot: artifacts/L<line>/step-NN-<name>.png`. Never embed screenshots as `` in the table body (breaks GitHub markdown rendering inside cells); just give the path.
3.**If a step has multiple commands**, separate them in the same cell with `<br>` so they render as one cell with multiple lines.
4.**PowerShell scriptlets > 3 lines**: write them to a separate `.ps1` in the artifacts folder and reference as ``script: `artifacts/L<line>/step-NN.ps1` `` in the cell. Keep the table cell to 1-3 lines.
5.**`—` (em dash) is allowed for non-CLI steps** like "Read sign-off entry + diff", "Create validation folder", "Cleanup notepad". Don't fabricate a command for steps that were purely cognitive or file-system level.
6.**Numbered steps must be contiguous** (1, 2, 3, ...). Don't skip numbers.
7.**At least one screenshot per PASS item if the item is a user-visible behavioral test**. Schema-only assertions (settings.json key check) don't need screenshots; behavioral tests (popup shown, dialog appeared, theme switched) do.
## §D — Reporting style
- Be specific. "Verified via UIA inspect returned `itm-calculator-XXXX`" beats "verified UIA".
- Include exact UIA selectors, log line text, settings.json keys, and screenshot filenames so the user can audit.
- For BLOCKED items, the 1-sentence reason should name **what specifically blocks**, e.g.:
- "BLK-HARDWARE: requires 2nd monitor; session has 1 (verified via `[System.Windows.Forms.Screen]::AllScreens.Count`)."
- "BLK-DRAG-REQUIRED: synthetic mouse drag insufficient for FZ snap-and-drag; needs real cursor motion."
- "BLK-ENV: SendInput returned ACCESS_DENIED (5) because Session $agentSession ≠ console Session $consoleSession. See `references/environment-setup.md`."
- "BLK-EXTERNAL-APP: requires real OpenAI API key; no key provisioned in test env."
## §E — Reporting anti-patterns (extra strict)
- Do NOT collapse multiple probe commands into a single English sentence like "verified via UIA". List every `winapp ui ...` command verbatim in a step row.
- Do NOT skip the step table for "trivial" items. Even a 1-step item (e.g. "Get-CmdPalSettings shows EnableDock=true") gets a 1-row table.
- Do NOT write screenshot references as `` inside table cells (GitHub renders markdown images poorly in cells). Write them as plain text path: `screenshot: artifacts/L<line>/step-NN-<name>.png`.
- Do NOT use "the test passed" as a screenshot caption — describe what's visible (e.g. "Settings page with FZ template grid showing 7 templates").
- Do NOT reference screenshots that you didn't actually capture. The final wrap-up `Test-Path` loop (see `references/pre-flight.md` §Final wrap-up step 3) will catch missing files; failing that check means the report is invalid.
- Do NOT cite source code line numbers (e.g. `CharacterMappings.cs:273`) without having actually read that line. If you cite source, the path must be real and the line number must contain what you claim.
## §F — Example item (reference: PR-47211 validation report style)
```markdown
## Item L455 — Activate Quick Accent (left Alt + arrow key) on a character, verify accents popup — **PASS** ✅
**Admin**: NO | **Clarity**: CLEAR | **Category**: drove full UIA flow + asserted accents popup
-`artifacts/L455/step-03-language-list.png` — Settings page with expanded language flyout
-`artifacts/L455/step-03-language-list.txt` — full UIA inspect dump of the list
-`artifacts/L455/step-04-popup-FR-E.png` — Popup with French only: `é è ê ë €`
### Verdict reasoning
- ✅ Popup characters match `CharacterMappings.cs` entries exactly (5/5 for FR.VK_E)
- ✅ Popup appeared within 500ms of hold-A; no crash
- ✅ Language list ordering is alphabetic by localized name
```
## §G — Retrospective (self-reflection)
After the run, reflect on the **process** (not the product) so the skill itself gets better over time. **If nothing slowed you down, write exactly one line: `Everything was smooth — no friction encountered.`** Otherwise, list each friction as a row and assign a source + severity.
```markdown
## Retrospective
| # | Friction (what slowed you / what was wrong) | Source | Severity | Cost | Suggested fix |
|---|---|---|---|---|---|
| 1 | <concrete description — what you expected vs what happened> | <one source tag below> | <HIGH/MED/LOW> | <~min wasted · N attempts> | <the doc line / helper function / tool behavior to change> |
```
**Source** — classify each friction into exactly one bucket so the right owner can fix it:
| Source tag | Meaning |
|---|---|
| `SKILL-UNCLEAR` | This skill's `SKILL.md` / `references/pre-flight.md` / module profile guidance was missing, ambiguous, or wrong. |
| `WINAPP-TOOL-BUG` | The `winapp` CLI itself misbehaved (crash, wrong output, flag not honored) — a product defect in the tool. |
| `WINAPP-DOC-UNCLEAR` | `references/winapp-ui-testing.md` was unclear/incorrect about how to use the tool (the tool worked; the docs misled you). |
| `HELPER-FLAW` | A shipped `scripts/*.ps1` had a logic bug, bad default, or wrong assumption. Name the function. |
| `PT-PRODUCT` | A PowerToys behavior/quirk made driving hard (distinct from a product **FAIL** — this is friction, not a checklist failure). |
| `CHECKLIST` | The checklist item itself was wrong/stale/ambiguous (e.g. describes a renamed or removed control). Note: this usually *also* produces a `FAIL (cause: checklist-*)` verdict on the item; log it here too so the checklist owner sees it as a process-improvement signal. |
| `ENVIRONMENT` | RDP/session/desktop/elevation friction not already covered by `references/environment-setup.md`. |
**Severity** — judge by *impact on future agents*, not just yourself:
- **HIGH** — most agents will hit it; blocks progress or wastes >10 min, or you needed a non-obvious workaround.
- **MED** — many agents may hit it; cost a few minutes or 2-3 retries; workaround exists once known.
- **LOW** — edge case or cosmetic; <1 min; noted for completeness.
**Cost** — be concrete: approximate minutes wasted **and** number of attempts (e.g. `~8 min · 3 attempts`). This is the raw signal for prioritizing skill fixes.
**Suggested fix** — point at the specific artifact to change: a doc line/section, a helper function name, or a `winapp` behavior to file. Vague reflections ("docs could be clearer") are not actionable — cite the line.
| 1 | `winapp ui inspect --depth 7 -w $hwnd` threw "Cannot bind argument" until I moved `-w` after `--depth`. | `WINAPP-TOOL-BUG` | MED | ~6 min · 3 attempts | Already noted in pitfall #14, but the tool should parse flag order — file against winapp. |
| 2 | SKILL.md §2.A says "wait 4s debounce" but PowerRename needed a full `Restart-PtRunner`; the module-owned-file note (pitfall #18) wasn't cross-linked from §2.A. | `SKILL-UNCLEAR` | HIGH | ~12 min · 4 attempts | Add an explicit "shell-ext modules → see pitfall #18" pointer inside §2.A. |
Automated UI testing for WinUI 3 apps — generate a batch test script, run all tests in one pass, read results. Covers element assertions, interactions, value checking (TextBox, ComboBox, ToggleSwitch), file pickers, flyouts, dialogs, persistence, and accessibility audits.
### Approach
The goal of this skill is to validate UI and app functionality automatically, without manual interaction, by exercising the app's UI elements, verifying their state, and asserting that the app behaves as expected under test conditions.
There are two main approaches:
1. Interactive exploration — manually run the app, use `winapp ui <command>` to explore the UI tree, find AutomationIds, verify element properties, and test functionality interactively. This is useful for discovery, but slow and expensive if repeated for every test iteration.
2. Scripted batch testing — generate a `ui-tests.ps1` script that exercises all UI elements and asserts expected behavior in one pass. This allows you to run the tests automatically, capture results, and iterate quickly without manually interacting with the app each time.
Unless the user asked for interactive exploration, or you are unfamiliar with the code/app or need to explore the UI tree to discover AutomationIds for hidden or dynamically generated elements (flyouts, dialogs, lazy-loaded content), **prefer scripted batch testing** — it is faster, repeatable, and produces a record of pass/fail results that can be reviewed and acted on.
### `winapp ui` Verbs
`status`, `inspect`, `search`, `get-property`, `get-value`, `screenshot`, `invoke`, `click`, `set-value`, `focus`, `scroll`, `scroll-into-view`, `wait-for`, `list-windows`, `get-focused`. Run `winapp ui --cli-schema` for the complete command structure as JSON, or `winapp ui <verb> --help` for any single verb.
### Step 1: Use the Running App
If the app is already running, use its PID. **Do NOT relaunch** — use the PID already captured from the build step. If the app is not running, build and launch it using the guidance in the winui-dev-workflow skill.
### Step 2: Write the Test Script
**If you wrote the code:** Skip inspect — you already know all the AutomationIds and control structure from the XAML and code-behind. Write tests directly from that knowledge. Inspect misses popups, flyouts, dialogs, and lazy-loaded content anyway.
**If you're verifying code you didn't write:** Run inspect first to discover the UI:
```powershell
winappuiinspect-a<PID>--interactive
```
Then read the XAML files to find AutomationIds that aren't currently visible (flyout items, dialog buttons, secondary pages).
Create a `ui-tests.ps1` file that tests all the app's requirements in one pass:
```powershell
# ui-tests.ps1
param([Parameter(Mandatory)][int]$AppPid)
# NOTE: Do NOT name the parameter $Pid — it's read-only in PowerShell
$ErrorActionPreference='Continue'
$pass=0;$fail=0;$results=@()
# Get main window HWND (avoids PopupHost interference with JSON parsing)
| "Data persists" | Set values, `invoke` a button (to commit bindings), verify data file on disk (`Get-Content` + `ConvertFrom-Json`) |
| "All controls accessible" | `inspect --interactive --json` + check all have AutomationId |
### Step 3: Run and Read Results
```powershell
.\ui-tests.ps1-AppPid<PID>
```
Read `test-results.json` for structured pass/fail. Only fix code if tests fail.
### Step 3.5: Look at the Screenshots
UIA assertions don't see clipping, overlap, wrong theming, or controls bleeding past their container — UIA returns `PASS` while the app is visually broken. **Capture screenshots with `winapp ui screenshot` and view each PNG.**
Capture the initial state and any state after a major interaction (the State Screenshots block in the script template above handles this).
**Visual checklist — fail the run if any item is `no`:**
- [ ] No unintended scrollbars
- [ ] No text ending in `…` that shouldn't be
- [ ] Hero elements fully visible (not sliced)
- [ ] Right-edge controls fully visible
- [ ] No overlapping rows
- [ ] Content uses the available width — no asymmetric dead zones (e.g. content pinned to one edge leaving empty space on the other)
- [ ] Spacing intentional — not cramped, not unintentionally vast
- [ ] Theming matches the user's ask (Light/Dark/HighContrast if relevant)
- [ ] Focus/hover/error states render if tested
If the checklist fails, it's a bug — fix before declaring done. Window too small → grow per `winui-design` Step 4.
### Step 4: Fix and Rerun (if the user asked for it)
If tests fail:
1. Read the failure details from `test-results.json`
2. Batch-fix all issues in one pass
3. Rebuild with `.\BuildAndRun.ps1` (blocking mode — shows crash info if the fix broke something)
4. Rerun `.\ui-tests.ps1 -AppPid <PID>` (parse PID from the `launched (PID: XXXXX)` output)
**Maximum 2 fix-and-rerun cycles.** If the same tests keep failing after 2 cycles, report them as known issues and move on — do not keep iterating.
### Assertion Reference
Use `wait-for --value` as the primary assertion — it uses a smart fallback chain that reads the right value for any control type:
| Dialog appeared | `winapp ui list-windows -a PID --json` (check window count) |
| Right-click menu | `winapp ui click "Id" -a PID --right` then `wait-for` menu item |
| Read raw property | `winapp ui get-property "Id" -a PID -p IsEnabled --json` |
| Read current value (no wait) | `(winapp ui get-value "Id" -a PID --json \| ConvertFrom-Json).text` — always pass `--json` when capturing into a variable (plain stdout can include advisory text like "Auto-selected HWND … from N windows"); otherwise prefer `wait-for --value` |
| Scroll item into view | `winapp ui scroll-into-view "Id" -a PID` — call before `wait-for` on virtualized ListView/repeater items below the fold |
| Set keyboard focus | `winapp ui focus "Id" -a PID` — cleaner than clicking another control to trigger a TextBox `LostFocus` commit |
### Testing File Pickers
File/folder pickers (FileOpenPicker, FileSavePicker, FolderPicker) run in a separate `PickerHost` process but are fully interactable. The picker appears as an owned dialog window.
```powershell
# 1. Trigger the picker
winappuiinvoke"BtnOpenFile"-a$AppPid
# 2. Find the picker window (it's a dialog owned by the app window)
**Tip:** Use `winapp ui inspect -w <pickerHwnd> --interactive` to discover the picker's controls — they include the folder tree, file list, filename textbox, and Open/Cancel buttons.
### Testing Context Menus and Flyouts
MenuFlyouts and ContextFlyouts are fully testable. They appear in the UI automation tree when open.
```powershell
# 1. Right-click to open a ContextFlyout
winappuiclick"LstItems"-a$AppPid--right
Start-Sleep0.5
# 2. The flyout MenuItems appear in the tree immediately
# Find them with inspect or search:
winappuiinspect-a$AppPid--interactive# shows MnuCopy, MnuDelete, etc.
**Tip:** ContentDialog buttons often don't have custom AutomationIds — use `inspect` to find the actual selector (slug or text match).
### Key Gotchas
- **`set-value` does NOT commit default TextBox bindings** — WinUI 3 `x:Bind TwoWay` on TextBox.Text updates the ViewModel on `LostFocus` by default. UIA `set-value` changes the text but doesn't trigger focus events. **Fix:** apps should use `UpdateSourceTrigger=PropertyChanged` on TextBox bindings (see design skill). If the app doesn't, `invoke` a button or `click` another element after `set-value` to trigger `LostFocus`.
- **Verify persistence via the data file, not UI relaunch** — killing and relaunching a packaged app from a test script is fragile (MSIX registration timing, PID issues). Instead, check the data file on disk: `Get-Content $dataFile | ConvertFrom-Json` and verify expected values.
- **Use `$AppPid` not `$Pid`** — `$Pid` is a read-only automatic variable in PowerShell
- **Use `--value` without `-p`** — it auto-detects the right UIA pattern (TextPattern → ValuePattern → TogglePattern → SelectionPattern → Name). Only use `-p PropertyName --value` when you need a specific property like `IsEnabled`
- **File pickers need `-w <HWND>`** — they run in a separate PickerHost process, so `-a PID` won't find them. Use `list-windows` to discover the picker HWND first
- **Flyouts need a short `Start-Sleep`** after triggering — the menu items appear in the tree asynchronously
### CRITICAL — `invoke` vs `click`: choose the right verb
**`winapp ui invoke <sel>`** dispatches through UIA's **`InvokePattern` via COM IPC**:
- ✅ Bypasses Windows UIPI (User Interface Privilege Isolation)
- ✅ Works even when your test runs elevated and the target is non-elevated AppX
- ✅ Does NOT steal foreground / does NOT trigger focus-loss handlers
- ✅ Works on Buttons, ListItems, ToggleSwitches, CheckBoxes — anything that exposes `InvokePattern` or `TogglePattern`
- ❌ Does NOT work on elements without an UIA action pattern (plain Grid, Text, Pane) — error message says "does not support any invoke pattern"
**`winapp ui click <sel>`** uses Win32 **`SendInput`** under the hood:
- ❌ **BLOCKED by UIPI** when source is elevated and target is non-elevated (or any AppX) — error: `SendInput failed — the target window may be elevated`
- ❌ Triggers foreground change → can dismiss popups, dialogs, AppX windows that hide on deactivation
- ✅ Only use when you genuinely need a synthetic mouse click (e.g. testing mouse hover/right-click flyouts where InvokePattern is unavailable)
- ✅ Subject to your process having interactive desktop access
**Rule of thumb**: try `invoke` first; only fall back to `click` if the target lacks InvokePattern AND you have a non-elevated test runner.
### CRITICAL — DataTemplate AutomationId vs ListItem InvokePattern
When XAML binds `AutomationProperties.AutomationId="{x:Bind <DataProperty>}"` inside a `ListView.ItemTemplate`'s `<DataTemplate>`, the AutomationId lives on the **inner Grid (Group)** the template produces — NOT on the outer ListItem the ListView wraps around it. The outer ListItem is what carries `InvokePattern`.
Concrete example (CmdPal PR #48033 binds Command.Id this way):
```powershell
# This FAILS with "does not support any invoke pattern":
winappuiinvoke$li.selector-w$hwnd# selector like 'itm-calculator-7e3f'
```
If you encounter "does not support any invoke pattern" while trying to use a data-bound AutomationId, this is almost always the cause. The fix is to search by Name and invoke the sibling ListItem.
### CRITICAL — Keystroke input that bypasses UIPI (PostMessage)
`winapp ui` has no `send-keys` verb. For keystroke input into elevated/AppX targets where SendInput fails, use **inline Win32 `PostMessage WM_KEYDOWN/WM_KEYUP`** which goes through the target's message queue without UIPI checks:
- WinUI3 apps' raw-input hooks may NOT process some keys via WM_KEYDOWN — `Esc` in particular often goes ignored (use BackButton invoke instead). Arrow keys + Enter typically work for ListView navigation.
- PostMessage returns immediately; allow 50-200 ms before reading state.
- Repeat `Send-KeyToHwnd` calls work for multi-step navigation (Down × 5 to scroll, then Enter).
`PostMessage` above targets a specific window's queue. To fire a **global hotkey** (e.g. a PowerToys activation chord like `Win+Shift+C`) you must inject into the **system input stream** with `SendInput` so the low-level keyboard hook (`WH_KEYBOARD_LL`) sees it. This **works for Win+ chords** — the common belief that "Win+ chords can't be injected" is false; it's almost always a **marshaling bug** (`SendInput` returns `0`, `GetLastError()==87`) from building the `INPUT[]` array in PowerShell. Build the array in C#:
- The injector must run at the **same or higher integrity level** as the hook owner (PowerToys runner). Default per-user installs run the runner at Medium IL, so a normal shell works; if the runner is elevated, run the injector elevated too (otherwise UIPI silently drops the injection).
- Must run in the interactive desktop session.
- OS-reserved chords (Win+L, Win+Tab) are consumed by Windows before any hook and cannot be injected this way.
- Verify the result via the runner trace log line `… hotkey is invoked from Centralized keyboard hook` (`%LOCALAPPDATA%\Microsoft\PowerToys\RunnerLogs\runner-log_<date>.log`) and/or the module's observable side-effect (overlay window, spawned editor process).
### CRITICAL — Verify foreground BEFORE every SendInput targeting a specific window
`SendInput` injects into the **session-wide** input stream — it goes to whatever IS foreground at the moment. If your target window has lost foreground (very common with AppX windows), the keys silently land in another window (often your own terminal) with no error returned.
Always check the foreground state immediately before calling `SendInput`. For winapp ui's output, the literal substring `foreground` appears in the line for the foreground window:
```powershell
functionTest-AppForeground{
param([Parameter(Mandatory)][string]$AppId)
$r=winappuilist-windows-a$AppId2>$null|Out-String
return($r-match'foreground')
}
# Force foreground (works ONCE per session reliably; subsequent attempts may be blocked by
throw'Cannot force CmdPal foreground; aborting SendInput batch'
}
}
# ... now safe to SendInput ...
```
**Tip**: when foreground cannot be reliably maintained, prefer `winapp ui set-value` (UIA-IPC, no foreground required) or `winapp ui invoke` (UIA InvokePattern, no foreground required) instead of SendInput.
### CRITICAL — `set-value` bypasses TextChanged for some apps (CmdPal alias detection)
`winapp ui set-value` writes the value through UIA's ValuePattern, which fires a programmatic value-change event. **It does NOT raise the `TextBox.TextChanged` event** the way real keystrokes do. For apps whose logic listens to `TextChanged` rather than to property changes — most notably CmdPal's alias detection (typing `=`, `<`, `>`, `:`, `$`, `??`, `)` in MainSearchBox triggers navigation to a provider sub-page) — `set-value` will set the text but the alias will NOT activate.
Workarounds:
- For plain queries: `winapp ui set-value` works fine (CmdPal still re-runs all providers on value change).
- For alias-triggered navigation: use **real keystrokes** via Force-AppForeground + SendInput, typing one character at a time with ~60-100ms delay so the alias detector sees the TextChanged sequence.
- Alternative: invoke the provider tile directly by its stable AutomationId (e.g. `winapp ui invoke 'com.microsoft.cmdpal.calculator' -w $hwnd`) when you only need the destination page, not the alias path.
### CRITICAL — Stunted UIA tree recovery
After ~30+ rapid `set-value` calls or after AppX has been interactive too long, an AppX window's UIA tree can degrade to a "stunted" state where `winapp ui inspect -w $h --depth 6` returns only ~5 elements (TitleBar / Close / Min / Max / RootPane) — even though the app looks fine visually.
Probe + recover pattern:
```powershell
# Probe: any healthy ListView-based AppX has >50 UIA nodes at depth 6
When the only realistic way to reach a needed test state is editing the app's persistent settings (e.g. multi-select that the UI's `SelectionItemPattern.Select` clobbers), wrap mutations with **byte-identical backup + restore-on-exit**:
**Important**: this should be used ONLY when the UI route is unreachable. Any setting flippable through the AppX Settings UI should be flipped that way instead (it's the documented user flow and tests real binding code).
Probe whether the current session is interactive (foreground + Shell COM both working).
Returns a PSCustomObject with .ForegroundOk and .ShellComOk.
.EXAMPLE
$env = Test-PtInteractiveDesktop
if (-not $env.ForegroundOk -or -not $env.ShellComOk) {
Write-Warning "Non-interactive session — Explorer-driven techniques will fail."
}
#>
Add-Type'using System; using System.Runtime.InteropServices; public class FG3 { [DllImport("user32.dll")] public static extern IntPtr GetForegroundWindow(); }'-EASilentlyContinue
Write-Warning"No classic shell verb matching '$NamePattern' on '$Path'. (Win11 PT modern-menu items are NOT visible here — use pt-explorer-contextmenu.ps1 instead.)"
return$false
}
$verb.Verb.DoIt()
return$true
}
functionReset-PtShellComCache{
<#
.SYNOPSIS
Release current Shell.Application COM instance + force a fresh one on next call.
Use when you've installed/registered a shell handler mid-test and the cached verb list
description: Toolkit for generating PowerToys release notes from GitHub milestone PRs or commit ranges. Use when asked to create release notes, summarize milestone PRs, generate changelog, prepare release documentation, request Copilot reviews for PRs, update README for a new release, manage PR milestones, or collect PRs between commits/tags. Supports PR collection by milestone or commit range, milestone assignment, grouping by label, summarization with external contributor attribution, and README version bumping.
description: Toolkit for generating PowerToys release notes from GitHub milestone PRs or commit ranges. Use when asked to create release notes, summarize milestone PRs, generate changelog, prepare release documentation, generate PR review summaries locally for release notes, update README for a new release, manage PR milestones, collect PRs between commits/tags, or prepare release assets (download installers and compute installer hashes).
license: Complete terms in LICENSE.txt
---
# Release Note Generation Skill
Generate professional release notes for PowerToys milestones by collecting merged PRs, requesting Copilot code reviews, grouping by label, and producing user-facing summaries.
Generate professional release notes for PowerToys milestones by collecting merged PRs, summarizing each PR with the local CLI agent, grouping by label, and producing user-facing summaries.
## Output Directory
@@ -26,16 +26,17 @@ Generated Files/ReleaseNotes/
- Generate release notes for a milestone
- Summarize PRs merged in a release
-Request Copilot reviews for milestone PRs
-Generate per-PR review summaries locally for release-notes copy
- **GitHub CLI (`gh`) installed and authenticated** — The collection script uses `gh pr view` and `gh api graphql` to fetch PR metadata and co-author information. Run `gh auth status` to verify; if not logged in, run `gh auth login` first. See [Step 1.0.0](./references/step1-collection.md) for details.
- MCP Server: github-mcp-server installed
-GitHub Copilot code review enabled for the org/repo
- MCP Server: github-mcp-server installed (used to fetch PR diffs/files for the local-agent review step)
-For [prepare-release-assets.ps1](./scripts/prepare-release-assets.ps1) only: **Azure CLI** authenticated against the Microsoft tenant (`az login`) with the `azure-devops` extension; access to the `microsoft/Dart` ADO project
## Required Variables
@@ -65,12 +66,12 @@ Generated Files/ReleaseNotes/
└────────────────────────────────┘
↓
┌────────────────────────────────┐
│ 3.1 Request Reviews (Copilot) │
│ 3.1 Local-agent PR summaries │
│ (writes CopilotSummary) │
└────────────────────────────────┘
↓
┌────────────────────────────────┐
│ 3.2 Refresh PR data │
│ (CopilotSummary) │
│ 3.2 (Optional) Refresh PR data │
└────────────────────────────────┘
↓
┌────────────────────────────────┐
@@ -93,7 +94,7 @@ Generated Files/ReleaseNotes/
| 1.1 | Collect PRs | From previous release tag on `stable` branch → `sorted_prs.csv` |
| 1.2 | Assign Milestones | Ensure all PRs have correct milestone |
| [prepare-release-assets.ps1](./scripts/prepare-release-assets.ps1) | Download installers + symbols from an ADO build, compute SHA256, emit the "Installer Hashes" markdown table for the GitHub release page |
## References
@@ -133,5 +135,6 @@ Do not read all steps at once—only read the step you are executing.
|-------|----------|
| `gh` command not found | Install GitHub CLI and add to PATH |
| No PRs returned | Verify milestone title matches exactly |
| Empty `CopilotSummary` for many PRs | Run Step 3.1 (local-agent summaries). Do **not** use `mcp_github_request_copilot_review` from a CLI/coding agent — the GitHub API rejects bot-initiated review requests, so the column will stay empty. |
| Many unlabeled PRs | Return to labeling step before grouping |
| `prepare-release-assets.ps1` fails with "Failed to acquire ADO access token" | Run `az login` and ensure you have access to the `microsoft/Dart` ADO project |
Use MCP tools to request Copilotreviews for all PRs in `Generated Files/ReleaseNotes/sorted_prs.csv`:
> ⚠️ **Do not use `mcp_github_request_copilot_review` (or any "request Copilot review" tool that calls the GitHub API).**
> When this skill is driven from a CLI / coding agent, the request is made from a bot identity and the GitHub API rejects it ("Bot reviewers cannot be requested"). The PR ends up with no Copilot review and `CopilotSummary` stays empty.
>
> Instead, **the local agent that is running this skill performs the review itself** and writes the summary directly into `sorted_prs.csv`.
- Use `mcp_github_request_copilot_review` for each PR ID
- Do NOT generate or run scripts for this step
For every PR listed in `Generated Files/ReleaseNotes/sorted_prs.csv` whose `CopilotSummary` is empty:
1. Fetch the PR diff using a tool that does **not** post anything back to GitHub. Any of these works:
-`mcp_github_pull_request_read` with `method: get_diff`
-`mcp_github_pull_request_read` with `method: get_files` (when the diff is large)
2. Read the PR title, body, and diff. Produce a 1–3 sentence, user-facing summary in the same style as a Copilot PR review (focus on observable behavior change, not implementation details).
3. Write the summary into the `CopilotSummary` column for that PR row in `Generated Files/ReleaseNotes/sorted_prs.csv`. Preserve all other columns and the existing row order.
**Batching guidance**
- Process PRs in the order they appear in `sorted_prs.csv`.
- Generate summaries for **all** PRs in one pass before continuing to Step 3.3, so the human reviewer can validate them together.
- For very large diffs, summarize from `get_files` (filenames + per-file patches) rather than the full diff.
- Skip PRs that already have a non-empty `CopilotSummary` (e.g. PRs where a human reviewer already pasted one). Do not overwrite existing summaries.
**Why not post the summary back to the PR?** Posting a comment from the agent's identity would not be picked up by `dump-prs-since-commit.ps1` (which only matches Copilot bot authors), and it adds noise to the PR. Writing straight into the CSV keeps the artifact self-contained.
---
## 3.2 Refresh PR Data
## 3.2 (Optional) Refresh PR Data
Re-run the collection script to capture Copilot review summaries into the `CopilotSummary` column:
Only re-run the collection script if PR metadata on GitHub has changed (new labels, retitled PRs, etc.) since Step 1.1. **Skip this step if you only want to preserve the locally generated `CopilotSummary` values from Step 3.1**, because re-running the dump will overwrite the CSV.
Creates `Generated Files/ReleaseNotes/grouped_csv/` with one CSV per label combination.
**Validation:** The `Unlabeled.csv` file should be minimal (ideally empty). If many PRs remain unlabeled, return to Step 2 (see [step2-labeling.md](./step2-labeling.md)).
Write-Error"Failed to acquire ADO access token. Run 'az login' first. (az: $script:LastAzError)"
exit1
}
# --- Step 5: Define the four installers to download ---
$targets=@(
[pscustomobject]@{Description="Per user - x64";Scope="perUser";Arch="x64";Artifact="build-x64-Release";FileName="PowerToysUserSetup-$versionParam-x64.exe"}
[pscustomobject]@{Description="Per user - ARM64";Scope="perUser";Arch="arm64";Artifact="build-arm64-Release";FileName="PowerToysUserSetup-$versionParam-arm64.exe"}
11.**Tests** — Adapt for WinUI 3 runtime, async patterns
### Key Principles
### Migration Contract: Prohibited Patterns
- **Do NOT overwrite `App.xaml` / `App.xaml.cs`** — WinUI 3 has different application lifecycle boilerplate. Merge your resources and initialization code into the generated WinUI 3 App class.
- **Do NOT create Exe→WinExe `ProjectReference`** — Extract shared code to a Library project. This causes phantom build artifacts.
- **Use `Lazy<T>` for resource-dependent statics** — `ResourceLoader` is not available at class-load time in all contexts.
These rules capture human judgment and must be applied consistently across every file. Do NOT deviate.
**Architecture prohibitions:**
- **Do NOT overwrite `App.xaml` / `App.xaml.cs`** — WinUI 3 has different lifecycle boilerplate. Merge resources and init code into the generated WinUI 3 App class.
- **Do NOT create Exe→WinExe `ProjectReference`** — Extract shared code to a Library project. Causes phantom build artifacts.
- **Do NOT instantiate services directly** — Use DI and CommunityToolkit.Mvvm patterns.
- **Do NOT create a `Window` subclass for every dialog or sub-page** — use `ContentDialog` for in-app dialogs and `Frame`/`Page` navigation for sub-views. Separate `Window` classes are reserved for distinct top-level surfaces (e.g., FancyZones editor, OOBE).
- **Do NOT omit `WindowsPackageType=None` and `WindowsAppSDKSelfContained=true`** — Both are mandatory in the csproj for every WinUI 3 module in PowerToys. Without them the app crashes at startup with `COMException: ClassFactory cannot supply requested class` because the WinUI 3 runtime DLLs are not found.
**XAML prohibitions:**
- **Do NOT use `{DynamicResource}`** — Replace with `{ThemeResource}` (theme-reactive) or `{StaticResource}`.
- **Do NOT use `{Binding}` in `Setter.Value`** — Not supported in WinUI 3. Use `{StaticResource}`.
- **Do NOT use `{x:Static}`** — Replace with `{x:Bind}`, `x:Uid`, or code-behind.
- **Do NOT use `{x:Type}`** — Not supported. Use `x:DataType` for DataTemplate, or code-behind.
- **Do NOT use `clr-namespace:`** — Replace with `using:` in all xmlns declarations.
- **Do NOT use `Style.Triggers` / `DataTrigger` / `EventTrigger`** — Replace with `VisualStateManager`.
- **Do NOT use `MultiBinding`** — Replace with `x:Bind` function binding or computed ViewModel property.
- **Do NOT use `Visibility="Hidden"`** — WinUI only has `Visible` and `Collapsed`. Use `Opacity="0"` if layout must be preserved.
- **Do NOT use `IsDefault` / `IsCancel`** — Use `AccentButtonStyle` for primary button; handle Enter/Escape in code-behind.
- **Do NOT omit `BasedOn` when overriding default styles** — Without it, your style replaces the entire default. Always use `BasedOn="{StaticResource DefaultButtonStyle}"` etc.
- **Do NOT omit `XamlControlsResources` as first merged dictionary** — It provides default Fluent styles. Without it, controls have no visual appearance.
**Code-behind prohibitions:**
- **Do NOT use `Application.Current.Dispatcher`** — Store `DispatcherQueue` in a static field explicitly.
- **Do NOT use `Window.Current`** — Not supported. Use a custom `App.Window` static property.
- **Do NOT put `DataContext`, `Resources`, or `VisualStateManager` on `Window`** — WinUI 3 `Window` is NOT a `DependencyObject`. Use a root `Page`/`UserControl`/`Grid`.
- **Do NOT use tunneling/preview events** (`PreviewMouseDown`, `PreviewKeyDown`) — WinUI has no tunneling. Use bubbling equivalents with `Handled` property or `AddHandler(handledEventsToo: true)`.
**Resource prohibitions:**
- **Do NOT use `Properties.Resources.MyString`** — Replace with `ResourceLoaderInstance.ResourceLoader.GetString("MyString")`.
- **Do NOT initialize `ResourceLoader`-dependent values as static fields** — Wrap in `Lazy<T>` or null-coalescing property.
- **Do NOT use `pack://` URIs** — Replace with `ms-appx:///` scheme.
These WPF controls have no direct counterpart and require a different control or third-party package:
| WPF Control | WinUI 3 Replacement | Notes |
|-------------|---------------------|-------|
| `DataGrid` | [`WinUI.TableView`](https://github.com/w-ahmad/WinUI.TableView) | Community library; the Toolkit `DataGrid` is no longer maintained. Legacy code may still pin v7 `CommunityToolkit.WinUI.UI.Controls.DataGrid` 7.1.2 |
| `Ribbon` | `CommandBar` / `NavigationView`, or [Toolkit Labs Ribbon](https://github.com/CommunityToolkit/Labs-Windows/tree/main/components/Ribbon) | No first-party Ribbon in WinUI; Labs component is experimental/partial |
| `Menu` / `MenuItem` | `MenuBar` / `MenuBarItem` / `MenuFlyout` | `MenuBar` for classic menu, `MenuFlyout` for context |
| `ContextMenu` | `MenuFlyout` | Assign to `ContextFlyout` property |
| `FlowDocument` | `RichTextBlock` | Partial replacement only |
| `RichTextBox` | `RichEditBox` | Rich text editing |
| `GroupBox` | `Expander` (built-in) or `HeaderedContentControl` (Toolkit) | See [Layout & Header Controls from CommunityToolkit.WinUI](#layout--header-controls-from-communitytoolkitwinui) below |
| `Label` | `TextBlock` | WPF `Label` is a `ContentControl`; use `TextBlock` + `AccessKey` |
| `TreeView` | `TreeView` (native) | Available natively, but data binding model differs significantly |
| `MessageBox` | `ContentDialog` | Must set `XamlRoot` before `ShowAsync()` |
| `MediaElement` | `MediaPlayerElement` | Different API |
| `AccessText` | Not available | Use `AccessKey` property on target control |
### Layout & Header Controls from CommunityToolkit.WinUI
These WPF controls have no built-in WinUI 3 equivalent — install the corresponding CommunityToolkit package. **The NuGet package id and the XAML namespace differ intentionally**: package names end in `.Primitives` / `.HeaderedControls`, but the registered XAML namespace is the shorter `CommunityToolkit.WinUI.Controls` (confirmed in the [official Microsoft Q&A](https://learn.microsoft.com/en-us/answers/questions/5746230/why-does-communitytoolkit-uwp-controls-primitive-u)).
| `RoutedUICommand` / `CommandBinding` | `ICommand` / `[RelayCommand]` from CommunityToolkit.Mvvm. WinUI also has `StandardUICommand` / `XamlUICommand` for platform commands. |
| `AdornerLayer` / `Adorner` | Depends on use case: `TeachingTip`/`InfoBar` (validation), `Popup` (overlays), `PlaceholderText` (watermarks), Canvas overlay (decorations) |
| `Visibility.Hidden` | `Opacity="0"` with `Visibility="Visible"` (preserves layout space) |
| `Window.Resources` / `Window.DataContext` | Move to root `Grid.Resources` / root `Page`/`UserControl` — WinUI `Window` is NOT a DependencyObject |
| Tunneling events (`Preview*`) | Use bubbling equivalents + `Handled` property or `AddHandler(handledEventsToo: true)` |
### Critical API Replacements
| WPF | WinUI 3 | Notes |
|-----|---------|-------|
| `Dispatcher.Invoke()` | `DispatcherQueue.TryEnqueue()` | Different return type (`bool`) |
| `Dispatcher.Invoke()` | `DispatcherQueue.TryEnqueue()` | Different return type (`bool`), async by default |
| `Dispatcher.CheckAccess()` | `DispatcherQueue.HasThreadAccess` | Property vs method |
| `Application.Current.Dispatcher` | Store `DispatcherQueue` in static field | See [Threading](./references/threading-and-windowing.md) |
| `Window.Current` | Custom `App.Window` static property | Not supported in Windows App SDK |
| Function binding | No | Yes (replaces `MultiBinding`) |
| Performance | Reflection-based | Compiled, no reflection |
| `MultiBinding` support | No (not in WinUI) | Use function binding |
## Detailed Reference Docs
@@ -151,6 +287,8 @@ Read only the section relevant to your current task:
| JPEG quality value wrong after migration | WPF: int 1-100; WinRT: float 0.0-1.0 |
| MSIX packaging fails in PreBuildEvent | Move to PostBuildEvent; artifacts not ready at PreBuild time |
| RC file icon path with forward slashes | Use double-backslash escaping: `..\\ui\\Assets\\icon.ico` |
| `COMException: ClassFactory cannot supply requested class` at startup | Missing `<WindowsPackageType>None</WindowsPackageType>` and/or `<WindowsAppSDKSelfContained>true</WindowsAppSDKSelfContained>` in csproj. Without these, the app tries to locate the Windows App SDK framework package (not installed) instead of using bundled runtime DLLs. **Both properties are mandatory for every WinUI 3 module in PowerToys.** |
| `CombinedGeometry` not available in WinUI 3 | WinUI 3 `UIElement.Clip` only accepts `RectangleGeometry`. For overlay hole effects (exclude region), use a `Path` element with `GeometryGroup FillRule="EvenOdd"` containing two `RectangleGeometry` children — the EvenOdd rule creates a transparent hole where geometries overlap. |
## Troubleshooting
@@ -163,3 +301,4 @@ Read only the section relevant to your current task:
| NuGet restore failures | Run `build-essentials.cmd` after adding `Microsoft.WindowsAppSDK` package |
| `Parallel.ForEach` compilation error | Migrate to `Parallel.ForEachAsync` for async imaging operations |
| Signing check fails on leaked artifacts | Run `generateAllFileComponents.ps1`; verify only `WinUI3Apps\\` paths in signing config |
| `COMException` / `ClassFactory` error at app launch | Ensure csproj has `<WindowsPackageType>None</WindowsPackageType>` and `<WindowsAppSDKSelfContained>true</WindowsAppSDKSelfContained>`. These are required for all unpackaged WinUI 3 apps in PowerToys — without them the WinUI 3 COM runtime cannot be found. |
| `ListBox` | `ListView` (or `ItemsView`) | **Deprecated in WinUI 3 — do not use.** Prefer `ListView`, or `ItemsView` (WinUI 1.5+) for modern collection scenarios |
| `AccessText` | Not available | Use `AccessKey` property on target control |
### WPF-UI (Lepo) to Native WinUI 3
@@ -109,6 +112,43 @@ Last parameter changes from `CultureInfo` to `string` (BCP-47 language tag). All
| `DataGrid` | [`WinUI.TableView`](https://github.com/w-ahmad/WinUI.TableView) | Community library; the Toolkit `DataGrid` is no longer maintained. PowerToys legacy modules may still pin v7 `CommunityToolkit.WinUI.UI.Controls.DataGrid` 7.1.2 — prefer `WinUI.TableView` for new work. |
| `Ribbon` | `CommandBar` / `NavigationView`, or [Toolkit Labs Ribbon](https://github.com/CommunityToolkit/Labs-Windows/tree/main/components/Ribbon) | No first-party Ribbon in WinUI; the Labs component is experimental and partial |
| `Menu` / `MenuItem` | `MenuBar` / `MenuBarItem` / `MenuFlyoutItem` | `MenuBar` for classic menu, `MenuFlyout` for context |
| `ContextMenu` | `MenuFlyout` | Assign to `ContextFlyout` property |
| `FlowDocument` | `RichTextBlock` | Partial replacement only |
| `RichTextBox` | `RichEditBox` | Rich text editing |
| `GroupBox` | `Expander` (built-in) or `HeaderedContentControl` (Toolkit) | See [Layout & Header Controls from CommunityToolkit.WinUI](#layout--header-controls-from-communitytoolkitwinui) below |
| `Label` | `TextBlock` | WPF `Label` is a `ContentControl`; use `TextBlock` + `AccessKey` |
| `TreeView` | `TreeView` (native) | Available natively, but data binding model differs significantly |
| `MediaElement` | `MediaPlayerElement` | Different API |
## Layout & Header Controls from CommunityToolkit.WinUI
These WPF layout/header controls have no built-in WinUI 3 equivalent — install the corresponding CommunityToolkit package. **The NuGet package id and the XAML namespace differ intentionally**: package names end in `.Primitives` / `.HeaderedControls`, but both packages register their controls in the shorter `CommunityToolkit.WinUI.Controls` XAML namespace (confirmed in the [official Microsoft Q&A](https://learn.microsoft.com/en-us/answers/questions/5746230/why-does-communitytoolkit-uwp-controls-primitive-u): *"all controls live under `CommunityToolkit.WinUI.Controls` … This is intentional"*).
Other primitives in `Primitives`: `ConstrainedBox`, `SwitchPresenter`, `WrapLayout`, `StaggeredPanel`. Other Headered controls in `HeaderedControls`: `HeaderedItemsControl`, `HeaderedTreeView`.
## NuGet Package Migration
@@ -124,6 +164,11 @@ Last parameter changes from `CultureInfo` to `string` (BCP-47 language tag). All
| (none) | `CommunityToolkit.WinUI.UI.Controls.DataGrid` | Legacy v7 — only if migrating existing `DataGrid` code; prefer [`WinUI.TableView`](https://github.com/w-ahmad/WinUI.TableView) for new work |
| (none) | `Microsoft.Web.WebView2` | If using WebView |
## Project File Changes
@@ -167,8 +212,9 @@ Last parameter changes from `CultureInfo` to `string` (BCP-47 language tag). All
- **CRITICAL: Add `WindowsPackageType=None`** — marks the app as unpackaged (no MSIX). Without this, the build produces an MSIX-style package that won't run as a standalone PowerToys module.
-**CRITICAL: Add `WindowsAppSDKSelfContained=true`** — bundles the WindowsAppSDK runtime DLLs (e.g. `Microsoft.UI.Xaml.dll`) into the output directory. Without this, the app throws `COMException: ClassFactory cannot supply requested class` at startup because the WinUI 3 COM classes cannot be found.
- Add `SelfContained=true` (usually via `Common.SelfContained.props`)
- Add `DISABLE_XAML_GENERATED_MAIN` if using custom `Program.cs` entry point
- Set `ProjectPriFileName` to match your module's assembly name
- Move icon from `Resources/` to `Assets/<Module>/`
@@ -127,6 +127,74 @@ If the module uses the `WPF-UI` library, replace all Lepo controls with native W
</Window>
```
#### Recommended: Use WindowEx from WinUIEx
> **Tip:** Prefer `WinUIEx.WindowEx` over bare `Window`. It restores many WPF-like window properties directly in XAML, avoiding boilerplate code-behind for common windowing tasks.
```xml
<!-- WinUI 3 with WindowEx (preferred in PowerToys) -->
<winuiex:WindowEx
xmlns:winuiex="using:WinUIEx"
x:Class="MyApp.MainWindow"
MinWidth="480"
MinHeight="320"
IsShownInSwitchers="True"
IsTitleBarVisible="True">
<Window.SystemBackdrop>
<MicaBackdrop/>
</Window.SystemBackdrop>
<Grid>
...
</Grid>
</winuiex:WindowEx>
```
Properties available on `WindowEx` that mirror WPF `Window`:
NuGet: `WinUIEx` — already referenced by most PowerToys modules.
#### Recommended: Page-in-Window Architecture
> **Tip:** WinUI 3 `Window` is NOT a `FrameworkElement` — it does not support `Resources`, `DataContext`, `x:Bind`, or `VisualStateManager` directly. Place a `Page` as the Window's root content to regain these WPF-like capabilities.
```xml
<!-- WinUI 3 — Window contains a Page for full FrameworkElement support -->
<winuiex:WindowExx:Class="MyApp.MainWindow"
xmlns:winuiex="using:WinUIEx"
xmlns:views="using:MyApp.Views">
<views:MainPagex:Name="mainPage"/>
</winuiex:WindowEx>
```
```xml
<!-- MainPage.xaml — has full FrameworkElement capabilities -->
> **Tip:** The `CommunityToolkit.WinUI` package provides many controls and helpers familiar to WPF developers that are missing from WinUI 3 out of the box. Before writing custom replacements, check whether CommunityToolkit already provides what you need.
Key packages (XAML namespace is `using:CommunityToolkit.WinUI.Controls` for the `Controls.*` family):
- **`CommunityToolkit.WinUI.UI.Controls.DataGrid`** — legacy v7 `DataGrid` (no longer maintained); prefer [`WinUI.TableView`](https://github.com/w-ahmad/WinUI.TableView) for new work
- **`CommunityToolkit.WinUI.Converters`** — Common value converters (`BoolToVisibilityConverter`, `StringFormatConverter`, etc.)
- **`CommunityToolkit.WinUI.Behaviors`** — XAML behaviors for animations and interactions
- **`CommunityToolkit.WinUI.Extensions`** — Extension methods for WinUI types
### Common Control Replacements
```xml
@@ -187,11 +269,135 @@ If the module uses the `WPF-UI` library, replace all Lepo controls with native W
<!-- Handle Enter/Escape keys in code-behind if needed -->
These WPF features demand design changes, not find-and-replace. Read this section BEFORE attempting to migrate any file that uses these patterns.
### MultiBinding → x:Bind Function Binding
WinUI does not support `MultiBinding`. Replace with `x:Bind` function binding (most direct replacement), a computed ViewModel property, or multiple simple bindings.
WinUI does not support routed commands or `CommandBinding`. Replace with standard `ICommand` pattern:
```csharp
// CommunityToolkit.Mvvm
[RelayCommand(CanExecute = nameof(CanSave))]
privatevoidSave(){/* save logic */}
privateboolCanSave()=>IsDirty;
```
WinUI 3 also provides `StandardUICommand` and `XamlUICommand` for pre-defined platform commands (Cut, Copy, Paste, Delete) with built-in icons and keyboard accelerators.
### Tunneling / Preview Events
WinUI has no tunneling event model. `PreviewMouseDown`, `PreviewKeyDown`, etc. do not exist.
- Replace with the bubbling equivalent (`PointerPressed`, `KeyDown`)
- If you relied on tunneling to intercept events before children, restructure using the `Handled` property
- For must-handle scenarios, use `AddHandler` with `handledEventsToo: true`:
> **Warning:** In WinUI 3, always use `BasedOn` when overriding default control styles. Without it, your style **replaces the entire default style** rather than extending it.
```xml
<!-- WRONG — replaces entire default style, control may lose all visual appearance -->
@@ -249,11 +452,42 @@ WPF `Triggers`, `DataTriggers`, and `EventTriggers` are not supported.
| `Disabled` | `Disabled` |
| `Pressed` | `Pressed` |
## Visibility.Hidden — No Equivalent
WinUI only has `Visible` and `Collapsed`. There is no `Hidden`.
| WPF | WinUI 3 | Behavior |
|-----|---------|----------|
| `Visibility.Visible` | `Visibility.Visible` | Rendered and occupies layout space |
| `Visibility.Hidden` | **Not available** | Use `Opacity="0"` with `Visibility="Visible"` to hide but keep layout |
| `Visibility.Collapsed` | `Visibility.Collapsed` | Not rendered, no layout space |
## Resource Dictionary Changes
### Window.Resources → Grid.Resources
### XamlControlsResources Must Be First
WinUI 3 `Window` is NOT a `DependencyObject` — no `Window.Resources`, `DataContext`, or `VisualStateManager`.
> **Warning:** `XamlControlsResources` must be the **first** merged dictionary in `App.xaml`. It provides the default Fluent styles. Omitting it gives you controls with no visual appearance. Resource paths use `ms-appx:///` instead of relative paths.
### Window.Resources → Grid.Resources (or use Page)
WinUI 3 `Window` is NOT a `FrameworkElement` — no `Window.Resources`, `DataContext`, or `VisualStateManager`.
**Preferred approach:** Use the [Page-in-Window architecture](#recommended-page-in-window-architecture) described above. A `Page` inside the `Window` gives you full `FrameworkElement` capabilities (Resources, DataContext, x:Bind, VisualStateManager).
**Fallback** (for simple windows without a Page):
```xml
<!-- WPF -->
@@ -317,19 +551,25 @@ Both are available. Prefer `{x:Bind}` for compile-time safety and performance.
| Performance | Reflection-based | Compiled |
| Function binding | No | Yes |
### WPF-Specific Binding Features to Remove
### Binding Differences from WPF
These WPF binding patterns behave differently in WinUI 3 — review on a case-by-case basis rather than mechanically removing.
```xml
<!--These WPF-only features must be removed or rewritten-->
<!--UpdateSourceTrigger: limited support in WinUI 3-->
| `{x:Static prefix:Resources.Key}` | `x:Uid="Key"` (with `.resw`) | Resource string — most common WPF case; mechanical `{x:Bind}` will NOT compile here |
| `MouseRightButtonUp` | `PointerReleased` + check `IsRightButtonPressed` | No direct WinUI event; use `RightTapped` only for context-menu-open semantics |
<!-- Fail fast with an actionable message instead of opaque C1083 spdlog/spdlog.h errors. -->
<Target Name="PowerToysEnsureVcpkgAvailable"
BeforeTargets="PrepareForBuild"
Condition="'$(MSBuildProjectExtension)' == '.vcxproj' and '$(VcpkgEnabled)' == 'true' and !Exists('$(VcpkgRoot)\scripts\buildsystems\msbuild\vcpkg.props')">
<Error Text="PowerToys requires the 'vcpkg' Visual Studio component, but it was not found.%0A%0AOpen the Visual Studio Installer, click Modify on your VS install, search for 'vcpkg', enable 'C++ vcpkg package manager', and click Modify. (Visual Studio will also prompt you to install missing .vsconfig components when you open PowerToys.slnx.)%0A%0AIf you have vcpkg installed elsewhere, set the VCPKG_ROOT environment variable to its root before building.%0A%0ASearched: '$(VcpkgRoot)\scripts\buildsystems\msbuild\vcpkg.props'" />
Condition="'$(MSBuildProjectExtension)' == '.vcxproj' and '$(VcpkgEnabled)' == 'true' and Exists('$(VcpkgRoot)\scripts\buildsystems\msbuild\vcpkg.targets')" />
@@ -242,6 +242,13 @@ Thank you for using PowerToys!
| Microsoft.PowerToys.FindMyMouse_EnableFindMyMouse | Triggered when Find My Mouse is enabled. |
| Microsoft.PowerToys.FindMyMouse_MousePointerFocused | Occurs when the mouse pointer is focused using Find My Mouse, including the activation method (double-tap left/right Ctrl, shake mouse, or shortcut). |
### Grab And Move
| Event Name | Description |
| --- | --- |
| Microsoft.PowerToys.GrabAndMove_EnableGrabAndMove | Triggered when Grab And Move is enabled or disabled. |
| Microsoft.PowerToys.GrabAndMove_ShortcutUse | Logs an attempt to move or resize a window via the Alt+Drag shortcut, including whether it succeeded, the action type (move or resize), and the reason (e.g., started, blocked by game mode). |
### Hosts File Editor
| Event Name | Description |
@@ -362,6 +369,15 @@ Thank you for using PowerToys!
| Microsoft.PowerToys.Peek_Settings | Triggered when the settings for Peek are modified. |
| Microsoft.PowerToys.Peek_SpaceModeEnabled | Triggered when the Space key activation mode is enabled or disabled in Peek. |
### Power Display
| Event Name | Description |
| --- | --- |
| Microsoft.PowerToys.PowerDisplay_EnablePowerDisplay | Triggered when Power Display is enabled or disabled. |
| Microsoft.PowerToys.PowerDisplay_Activate | Triggered when Power Display is activated via hotkey or tray toggle. |
| Microsoft.PowerToys.PowerDisplay_Start | Triggered when the Power Display application process starts. |
| Microsoft.PowerToys.PowerDisplay_Settings | Periodic snapshot of Power Display settings, including whether the hotkey and tray icon are enabled, the number of detected monitors, and the number of saved profiles. |
<!-- dotnet.exe seems to access files after builds. Temporarily putting in this entry for testing if we get further. This looks to be related to a .NET Roslyn Analyzer in .NET 10-->
This repo uses a common output directory with many projects writing duplicate outputs. Allow everything, but note this costs some performance in the form of requiring
the cache to use copies instead of hardlinks when pulling from cache.
Text="Windows long path support is not enabled. PowerToys source paths exceed the 260-character MAX_PATH limit, so the build will fail with cryptic 'path too long' errors. Fix it by running (from an elevated PowerShell): .\tools\build\setup-dev-environment.ps1 -- or set HKLM\SYSTEM\CurrentControlSet\Control\FileSystem\LongPathsEnabled = 1 (DWORD) and restart Windows. To bypass this check, build with /p:SkipLongPathsCheck=true." />
</Target>
<!-- Override ManifestTool to the x64 host tool under WindowsSdkDir for all projects once the SDK path is known. -->
<!-- Including MessagePack to force version, since it's used by StreamJsonRpc but contains vulnerabilities. After StreamJsonRpc updates the version of MessagePack, we can upgrade StreamJsonRpc instead. -->
<!-- Package Microsoft.Win32.SystemEvents added as a hack for being able to exclude the runtime assets so they don't conflict with 8.0.1. This is a dependency of System.Drawing.Common but the 8.0.1 version wasn't published to nuget. -->
<!-- Package System.CodeDom added as a hack for being able to exclude the runtime assets so they don't conflict with 8.0.1. This is a dependency of System.Management but the 8.0.1 version wasn't published to nuget. -->
<!-- Package System.Data.SqlClient added to force it as a dependency of Microsoft.Windows.Compatibility to the latest version available at this time. -->
<!-- Package System.Diagnostics.EventLog added as a hack for being able to exclude the runtime assets so they don't conflict with 8.0.1. This is a dependency of System.Data.OleDb but the 8.0.1 version wasn't published to nuget. -->
<!-- Package System.Diagnostics.PerformanceCounter added as a hack for being able to exclude the runtime assets so they don't conflict with 8.0.11. -->
@@ -47,21 +47,7 @@ But to get started quickly, choose one of the installation methods below:
<summary><strong>Download the .exe file from GitHub</strong></summary>
<br/>
Go to the [PowerToys GitHub releases](https://aka.ms/installPowerToys), select **Assets** to reveal the installation files, and choose the one that matches your architecture and install scope. For most devices, that would be _x64 per-user_.
<!-- items that need to be updated release to release -->
Go to the [PowerToys GitHub releases](https://aka.ms/installPowerToys), scroll down and select **Assets** to reveal the installation files, and choose the one that matches your architecture and install scope. For most devices, that would be _x64 per-user_.
</details>
@@ -106,7 +92,7 @@ There are [community driven install methods](https://learn.microsoft.com/windows
[](https://github.com/microsoft/PowerToys/releases)
To see what's new, check out the [release notes](https://github.com/microsoft/PowerToys/releases/tag/v0.99.0).
To see what's new, check out the [release notes](https://github.com/microsoft/PowerToys/releases/).
## 🛣️ Roadmap
@@ -131,3 +117,4 @@ The application logs basic diagnostic data (telemetry). For more privacy informa
"description":"Very fast, header-only/compiled, C++ logging library. PowerToys overlay pinned to gabime/spdlog@616866fc (the exact submodule commit before this migration), with a single-hunk patch that works around MSVC 14.51 removing stdext::checked_array_iterator (STL4043).",
@@ -97,6 +97,10 @@ The Shell Process Debugging Tool is a Visual Studio extension that helps debug m
- Check Event Viewer for application crashes related to `PowerToys.Settings.exe`
- Crash dumps can be obtained from Event Viewer
### Debugging Command Palette
Command Palette can be easily debugged using the solution filter in `src/modules/cmdpal/Command Palette.slnf`. This will open Command Palette as its own Visual Studio solution that can be run and debugged directly in Visual Studio without the need for the Shell Process Debugging Tool.
- [ ] The plugin is a project under `modules\launcher\Plugins`
- [ ] Microsoft plugin project name pattern: `Microsoft.PowerToys.Run.Plugin.{PluginName}`
- [ ] Community plugin project name pattern: `Community.PowerToys.Run.Plugin.{PluginName}`
- [ ] The plugin target framework should be `net9.0-windows10.0.22621.0`
- [ ] The plugin target framework should be `net10.0-windows10.0.22621.0`
- [ ] If the plugin uses any 3rd party dependencies the project file should import `DynamicPlugin.props`
- [ ] 3rd party dependencies must be compatible with .NET 9
- [ ] 3rd party dependencies must be compatible with .NET 10
- [ ] The plugin has to contain a `plugin.json` file of the following format in its root folder:
```json
@@ -42,4 +42,4 @@ In the PR that adds a new plugin, reference a new issue to track the work for fu
- [ ] Add the resource folder to https://github.com/microsoft/PowerToys/blob/21247c0bb09a1bee3d14d6efa53d0c247f7236af/installer/PowerToysSetup/Product.wxs#L825
- [ ] Add the resource files under the section https://github.com/microsoft/PowerToys/blob/21247c0bb09a1bee3d14d6efa53d0c247f7236af/installer/PowerToysSetup/Product.wxs#L882
- [ ] Your plugin's executable file (DLL) has to have correct version informations after building it. (This version information will be shown on the settings page.)
- [ ] Your plugin's executable file (DLL) has to have correct version information after building it. (This version information will be shown on the settings page.)
Shortcut Guide is a PowerToy that displays an overlay of available keyboard shortcuts when the Windows key is pressed and held. It provides a visual reference for Windows key combinations, helping users discover and utilize built-in Windows shortcuts.
Shortcut Guide is a PowerToy that displays an overlay of available keyboard shortcuts when a user-set keyboard shortcut is pressed. It helps users discover and remember keyboard shortcuts for Windows and apps.
> [!NOTE]
> The spec for the manifest files is in development and will be linked here once available.
## Usage
- Press and hold the Windows key to display the overlay of available shortcuts
- Press the hotkey again to dismiss the overlay
- The overlay displays Windows shortcuts with their corresponding actions
- Press the user-defined hotkey to display the overlay
- Press the hotkey again or press ESC to dismiss the overlay
## Build and Debug Instructions
@@ -25,67 +27,89 @@ Shortcut Guide is a PowerToy that displays an overlay of available keyboard shor
4. The executable is named PowerToys.ShortcutGuide.exe
### Debug
1. Right-click the ShortcutGuide project and select 'Set as Startup Project'
1. Right-click the ShortcutGuide.Ui project and select 'Set as Startup Project'
2. Right-click the project again and select 'Debug'
## Code Structure
> [!NOTE]
> When run in debug mode, the window behaves differently than in release mode. It will not automatically close when loosing focus, it will be displayed on top of all other windows, and it is not hidden from the taskbar.
## Project Structure
### Core Files
The Shortcut Guide module consists of the following 4 projects:
Contains DLL boilerplate code. Implements the PowertoyModuleIface, including enable/disable functionality and GPO policy handling. Captures hotkey events and starts the PowerToys.ShortcutGuide.exe process to display the shortcut guide window.
Contains the module interface code. It initializes the settings values and the keyboard event listener. Defines the OverlayWindow class, which manages the overall logic and event handling for the PowerToys Shortcut Guide.
This is the main UI project for the Shortcut Guide module. Upon startup it does the following tasks:
Contains the code for loading the SVGs, creating and rendering of the overlay window. Manages and displays overlay windows with SVG graphics through two main classes:
- D2DOverlaySVG: Handles loading, resizing, and manipulation of SVG graphics
- D2DOverlayWindow: Manages the display and behavior of the overlay window
1. Copies the built-in manifest files to the users manifest directory (overwriting existing files).
2. Generate the `index.yml` manifest file.
3. Populate the PowerToys shortcut manifest with the user-defined shortcuts.
State machine that handles the keyboard events. It's responsible for deciding when to show the overlay, when to suppress the Start menu (if the overlay is displayed long enough), etc. Handles state transitions and synchronization to ensure the overlay is shown or hidden appropriately based on user interactions.
Handles the timing and interpolation of animations. Calculates the current value of an animation based on elapsed time and a specified easing function.
This function checks if the current window is excluded from the Shortcut Guide overlay. It returns `true` if the current window is excluded otherwise it returns `false`.
Waits for a named event and executes a specified action when the event is triggered. Uses a separate thread to handle event waiting and action execution.
This function retrieves the positions of the taskbar buttons for a given monitor. It returns an array of `TasklistButton` structures (max 10), which contain the position and size of each button.
The entry point for the PowerToys Shortcut Guide application. Handles initialization, ensures single instance execution, manages parent process termination, creates and displays the overlay window, and runs the main event loop.
`size` will contain the resulting array size.
It determines the positions through Windows `FindWindowEx` function.
For the primary taskbar it searches for:
* A window called "Shell_TrayWnd"
* that contains a window called "ReBarWindow32"
* that contains a window called "MSTaskSwWClass"
* that contains a window called "MSTaskListWClass"
For any secondary taskbar it searches for:
* A window called "Shell_SecondaryTrayWnd"
* that contains a window called "WorkerW"
* that contains a window called "MSTaskListWClass"
It then enumerates all the button elements inside "MSTaskListWClass" while skipping such with a same name (which implies the user does not use combining taskbar buttons)
If this method fails, which it will for newer versions of Windows, it falls back to searching for:
* A window called "Shell_TrayWnd" or "Shell_SecondaryTrayWnd"
* that contains a window called "Windows.UI.Composition.DesktopWindowContentBridge"
* that contains a window called "Windows.UI.Input.InputSite.WindowClass"
* the first child element
It then enumerates all the button elements inside the selected while skipping such with a same name (which implies the user does not use combining taskbar buttons) and such that do not start with "Appid:" (which are not actual taskbar buttons related to apps, but others like the widgets or the search button).
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
