agentsclimarketplace

Winui code review

Skill unliftedq/agents/winui-dev/skills/winui-code-review

A collection of portable, domain-focused AI agents managed by kman.

Install
npx -y skills add unliftedq/agents --skill winui-code-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Code quality review for WinUI 3 apps — MVVM compliance, x:Bind correctness, accessibility, theming, security, and performance. Use before committing to catch issues that the compiler and UI tests won't find.

SKILL.md

5.9 KB, ~1.4k tokens by cl100k_base, as published. Nobody here has run it

When to Use

Run a code review after the app builds and before committing. This catches quality issues that aren't build errors and aren't visible in UI tests — patterns that compile and run but are wrong, fragile, or slow.

How to Review

Read through the project's XAML and C# files and check each section below. The Microsoft.WindowsAppSDK.Analyzers Roslyn analyzer ships with the winui-dev-workflow skill and is injected into your build when you compile via the BuildAndRun.ps1 script that ships with that skill — the script drops a temporary Directory.Build.props into the project that loads the analyzer DLL and its .targets, then cleans up after the build. Plain dotnet build (or VS) does not load the analyzer automatically; if you want it to surface as build diagnostics outside the script, add the <Analyzer Include="..." /> and <Import Project="..." /> to your project's own Directory.Build.props (or wait for the planned NuGet package).

The analyzer catches a curated set of WinUI 3 / Windows App SDK issues with categorized 4-digit IDs:

  • WUI0xxx — UWP → WinUI 3 API compatibility (UwpXamlNamespace, Window.Current, CoreDispatcher, GetForCurrentView)
  • WUI1xxx — Migration-table data-driven hints (UWP API has WinAppSDK equivalent, no equivalent, feature-area hint)
  • WUI2xxx — Runtime / layout / XAML pitfalls (raw TabView content, nested x:Bind without fallback, x:Bind without Mode, null Converter, missing AutomationId, attached-property syntax)
  • WUI3xxx — MVVM patterns (old [ObservableProperty] field syntax)
  • WUI4xxx — Interop (WebView2 not initialized, removed ONNX Runtime GenAI APIs WUI4101-WUI4103)

Every diagnostic ships at Warning severity (no rule is Error) and includes a helpLinkUri. Suppress noise with #pragma warning disable WUIxxxx or <NoWarn> as usual — the analyzer's SuppressionTests verify that pragma suppression round-trips correctly.

MVVM Compliance

  • ViewModels extend ObservableObject, use [ObservableProperty] partial properties (not fields)
  • Commands use [RelayCommand] attribute, not manual ICommand implementations
  • No UI types in ViewModels (SolidColorBrush, Visibility, BitmapImage) — these belong in converters or XAML
  • No business logic in code-behind — only navigation, dialog coordination, and event wiring
  • async Task for async methods, async void only for event handlers
  • Never replace ObservableCollection<T> — use .Clear() + re-add

x:Bind and Data Binding

  • All bindings use {x:Bind}, not {Binding}
  • Mode=OneWay or TwoWay set explicitly — OneTime default causes blank UI for dynamic data
  • x:DataType set on every DataTemplate — required for compiled x:Bind
  • No nested nullable paths (e.g., ViewModel.Selected.Name) without FallbackValue
  • Command bindings can use OneTime (commands don't change) — don't add Mode=OneWay to Command="{x:Bind}"

Accessibility

  • AutomationProperties.AutomationId on every interactive control (Button, TextBox, ComboBox, ToggleSwitch, ListView, NavigationViewItem)
  • AutomationProperties.Name on icon-only buttons and controls without visible text
  • Semantic controls (Button, HyperlinkButton) — not clickable Border/TextBlock
  • No information conveyed by color alone

Theming

  • All colors use {ThemeResource} brushes — no hardcoded #FF0000 or Color="Blue"
  • Typography uses built-in styles (TitleTextBlockStyle, SubtitleTextBlockStyle, BodyTextBlockStyle, CaptionTextBlockStyle) — no raw FontSize
  • Spacing uses 4px grid multiples (4, 8, 12, 16, 24, 32, 48)
  • Corner radius uses ControlCornerRadius / OverlayCornerRadius — not hardcoded values
  • Styles referenced with {StaticResource} not {ThemeResource} (except for brush usage sites)

Security

  • No secrets, API keys, or tokens in source code
  • No Process.Start with unsanitized user input
  • External input validated and sanitized before use
  • File paths from user input not used directly in File.Delete / File.WriteAllText without validation

Performance

  • Long or dynamic lists use ListView/GridView (virtualized), not StackPanel with foreach
  • x:Load for content that's not always visible (e.g., dialogs, secondary panels)
  • Heavy work off UI thread via Task.Run or async/await — never block UI
  • No .Result / .Wait() / .GetAwaiter().GetResult() — these deadlock the UI thread
  • using statements on all disposable objects (Model, Tokenizer, InferenceSession, Generator)

Globalization

  • User-facing strings use x:Uid in XAML and ResourceLoader in C# — not hardcoded
  • String resources in Strings/en-us/Resources.resw (not .resx)
  • Date/number formatting uses CultureInfo.CurrentCulture — not hardcoded formats
  • Layout supports RTL (FlowDirection inherited from root, no absolute positioning that breaks in RTL)
  • No string concatenation for user-facing messages — use string.Format or interpolation with resource strings

Review Report

After reviewing, summarize:

  1. Issues found: List each with file, line, and what's wrong
  2. Severity: Error (must fix), Warning (should fix), or Note (could improve)
  3. Suggested fixes: Specific code changes for each issue

References

For detailed rules with code examples, see references/quality-rules.md — covers performance deep dives (x:Phase, layout optimization), security (PasswordVault, DPAPI, WebView2 hardening), accessibility (keyboard nav, screen readers), code quality (.editorconfig, naming), and globalization (x:Uid patterns, RTL, pluralization).

Gives 0 of the 12 instructions most code review skills give in ~1.4k tokens

Counted across 610 of the 674 authors here whose files we hold, read 2026-08-06

  • push back with technical reasoning if wrongin 60 of 610, across 24 files
  • ask for clarification on unclear itemsin 51 of 610, across 16 files
  • fix critical issues immediatelyin 45 of 610, across 29 files
  • implement one item at a timein 45 of 610, across 11 files
  • group findings by severityin 44 of 610, across 43 files
  • verify feedback against the codebasein 42 of 610, across 8 files
  • dispatch a code reviewer subagentin 39 of 610, across 23 files
  • fix important issues before proceedingin 37 of 610, across 22 files
  • test each fix individuallyin 35 of 610, across 7 files
  • reply in github comment threadsin 33 of 610, across 5 files
  • check for security vulnerabilitiesin 31 of 610, across 27 files
  • factualize corrections without over-explainingin 30 of 610, across 2 files

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.