feat: Windows Computer Use enhancement — PrintWindow, UI Automation, OCR

New Windows-native capabilities: - windowCapture.ts: PrintWindow API for per-window screenshot (works on occluded/background windows) - windowEnum.ts: EnumWindows for precise window enumeration with HWND - uiAutomation.ts: IUIAutomation for UI tree reading, element clicking, text input, and coordinate-based element identification - ocr.ts: Windows.Media.Ocr for screen text recognition (en-US + zh-CN) Updated win32.ts backend to use EnumWindows for listRunning() and added captureWindowTarget() for window-specific screenshots. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-06-15 12:55:51 +00:00 · 2026-04-04 00:00:02 +08:00
parent e3264a1691
commit 3707c3c0ba
7 changed files with 1141 additions and 5 deletions
--- a/DEV-LOG.md
+++ b/DEV-LOG.md
@@ -1,5 +1,32 @@
 # DEV-LOG

+## Computer Use Windows 增强：窗口绑定截图 + UI Automation + OCR (2026-04-03)
+
+在三平台基础实现之上，利用 Windows 原生 API 增强 Computer Use 的 Windows 专属能力。
+
+**新增文件：**
+
+| 文件 | 行数 | 说明 |
+|------|------|------|
+| `src/utils/computerUse/win32/windowCapture.ts` | — | `PrintWindow` 窗口绑定截图，支持被遮挡/后台窗口 |
+| `src/utils/computerUse/win32/windowEnum.ts` | — | `EnumWindows` 精确窗口枚举（HWND + PID + 标题） |
+| `src/utils/computerUse/win32/uiAutomation.ts` | — | `IUIAutomation` UI 元素树读取、按钮点击、文本写入、坐标识别 |
+| `src/utils/computerUse/win32/ocr.ts` | — | `Windows.Media.Ocr` 截图+文字识别（英语+中文） |
+
+**修改文件：**
+
+| 文件 | 变更 |
+|------|------|
+| `packages/@ant/computer-use-swift/src/backends/win32.ts` | `listRunning` 改用 EnumWindows；新增 `captureWindowTarget` 窗口级截图 |
+
+**验证结果（Windows x64）：**
+- 窗口枚举：38 个可见窗口 ✅
+- 窗口截图：VS Code 2575x1415, 444KB ✅（PrintWindow, 即使被遮挡）
+- UI Automation：坐标元素识别 ✅
+- OCR：识别 VS Code 界面文字，34 行 ✅
+
+---
+
 ## Enable Computer Use — macOS + Windows + Linux (2026-04-03)

 恢复 Computer Use 屏幕操控功能。参考项目仅 macOS，本次扩展为三平台支持。
--- a/docs/features/computer-use-windows-enhancement.md
+++ b/docs/features/computer-use-windows-enhancement.md
@@ -0,0 +1,315 @@
+# Computer Use Windows 增强实施计划
+
+更新时间：2026-04-03
+依赖文档：`docs/features/windows-ai-desktop-control.md`、`docs/features/computer-use.md`
+
+## 1. 目标
+
+在已有的 PowerShell 子进程方案基础上，利用 Windows 原生 API 增强 Computer Use 的 Windows 实现，解决 3 个核心问题：
+
+1. **窗口绑定截图**：当前 `CopyFromScreen` 只能全屏截图，无法对指定窗口截图（尤其是被遮挡/最小化窗口）
+2. **UI 结构感知**：当前只能通过坐标点击，无法像 macOS Accessibility 那样理解 UI 元素树
+3. **性能**：每次 PowerShell 启动约 273ms，剪贴板/窗口枚举等高频操作需要更快的方式
+
+## 2. 已验证的 Windows API 能力
+
+以下 API 全部通过 PowerShell P/Invoke 实测通过：
+
+| 能力 | API | 验证结果 |
+|------|-----|---------|
+| 窗口绑定截图 | `PrintWindow(hwnd, hdc, PW_RENDERFULLCONTENT)` | ✅ VS Code 342KB, Chrome 273KB |
+| 枚举窗口+HWND | `EnumWindows` + `GetWindowText` + `GetWindowThreadProcessId` | ✅ 38 个窗口，含 HWND/PID/标题 |
+| UI 元素树 | `System.Windows.Automation.AutomationElement` | ✅ 记事本 39 个元素 |
+| UI 写值 | `ValuePattern.SetValue()` | ✅ 成功写入记事本文本 |
+| UI 点击 | `InvokePattern.Invoke()` | ✅ 按钮可程序化点击 |
+| 坐标元素识别 | `AutomationElement.FromPoint(x, y)` | ✅ 返回元素类型+名称 |
+| OCR | `Windows.Media.Ocr.OcrEngine` | ✅ 英语+中文引擎可用 |
+| 全局热键 | `RegisterHotKey` | ✅ API 可调 |
+| 剪贴板直接操作 | `System.Windows.Forms.Clipboard` | ✅ 读/写/图片检测 |
+| Shell 启动 | `ShellExecute` | ✅ 打开文件/URL/应用 |
+
+## 3. 架构设计
+
+### 3.1 文件结构
+
+在现有 `backends/win32.ts` 基础上新增 Windows 专属模块：
+
+```
+packages/@ant/computer-use-input/src/
+├── backends/
+│   ├── darwin.ts          ← 不动
+│   ├── win32.ts           ← 增强：直接 Win32 API 替代部分 PowerShell
+│   └── linux.ts           ← 不动
+
+packages/@ant/computer-use-swift/src/
+├── backends/
+│   ├── darwin.ts          ← 不动
+│   ├── win32.ts           ← 增强：PrintWindow 窗口截图 + EnumWindows
+│   └── linux.ts           ← 不动
+
+packages/@ant/computer-use-mcp/src/
+│   └── tools.ts           ← 增加 Windows 专属工具定义（UI Automation、OCR）
+
+src/utils/computerUse/
+│   └── win32/              ← 新增目录：Windows 专属能力
+│       ├── uiAutomation.ts  ← UI 元素树、点击、写值
+│       ├── ocr.ts           ← 截图 + OCR 文字识别
+│       ├── windowCapture.ts ← PrintWindow 窗口绑定截图
+│       └── windowEnum.ts    ← EnumWindows 窗口枚举
+```
+
+### 3.2 分层
+
+```
+┌──────────────────────────────────────────────┐
+│           Computer Use MCP Tools             │
+│  screenshot / click / type / request_access  │
+│  + Windows 专属: ui_tree / ocr / window_cap  │
+├──────────────────────────────────────────────┤
+│           src/utils/computerUse/             │
+│  executor.ts → 按平台 dispatch               │
+│  win32/ → Windows 专属能力模块               │
+├──────────────────────────────────────────────┤
+│     packages/@ant/computer-use-{input,swift}  │
+│  backends/win32.ts → PowerShell + Win32 API  │
+├──────────────────────────────────────────────┤
+│           Windows Native API                 │
+│  PrintWindow / EnumWindows / UI Automation   │
+│  SendInput / Clipboard / OCR / ShellExecute  │
+└──────────────────────────────────────────────┘
+```
+
+## 4. 实施计划
+
+### Phase A：窗口绑定截图（解决核心问题）
+
+**问题**：当前 `CopyFromScreen` 只能全屏截图，无法对指定窗口截图。
+**方案**：用 `PrintWindow` + `FindWindow` 实现窗口级截图。
+
+| 步骤 | 文件 | 改动 |
+|------|------|------|
+| A.1 | `src/utils/computerUse/win32/windowCapture.ts` | 新建：`captureWindow(title)` 用 PrintWindow 截取指定窗口 |
+| A.2 | `src/utils/computerUse/win32/windowEnum.ts` | 新建：`listWindows()` 用 EnumWindows 返回 {hwnd, pid, title}[] |
+| A.3 | `packages/@ant/computer-use-swift/src/backends/win32.ts` | `screenshot.captureExcluding` 增加按窗口截图能力 |
+| A.4 | `packages/@ant/computer-use-swift/src/backends/win32.ts` | `apps.listRunning` 用 EnumWindows 替代 Get-Process（返回 HWND） |
+
+**PowerShell 脚本核心**：
+
+```powershell
+# PrintWindow 截取指定窗口
+Add-Type -AssemblyName System.Drawing
+Add-Type -ReferencedAssemblies System.Drawing @'
+using System; using System.Runtime.InteropServices; using System.Drawing; using System.Drawing.Imaging;
+public class WinCap {
+    [DllImport("user32.dll", CharSet=CharSet.Unicode)]
+    public static extern IntPtr FindWindow(string c, string t);
+    [DllImport("user32.dll")]
+    public static extern bool GetWindowRect(IntPtr h, out RECT r);
+    [DllImport("user32.dll")]
+    public static extern bool PrintWindow(IntPtr h, IntPtr hdc, uint f);
+    [StructLayout(LayoutKind.Sequential)]
+    public struct RECT { public int L, T, R, B; }
+    // ... CaptureByTitle(string title) → base64
+}
+'@
+```
+
+**验证标准**：
+- 能按窗口标题截图
+- 被遮挡的窗口也能截图
+- 返回 base64 + width + height
+
+### Phase B：UI Automation（Windows 专属新能力）
+
+**问题**：macOS 有 Accessibility API 可以读取/操作 UI 元素，Windows 当前只能坐标点击。
+**方案**：用 `System.Windows.Automation` 实现 UI 树读取和元素操作。
+
+| 步骤 | 文件 | 改动 |
+|------|------|------|
+| B.1 | `src/utils/computerUse/win32/uiAutomation.ts` | 新建：核心 UIA 操作封装 |
+| B.2 | `packages/@ant/computer-use-mcp/src/tools.ts` | 增加 Windows 专属工具定义 |
+
+**uiAutomation.ts 导出函数**：
+
+```typescript
+// 获取窗口的 UI 元素树
+getUITree(windowTitle: string, depth: number): UIElement[]
+
+// 按名称/类型/AutomationId 查找元素
+findElement(windowTitle: string, query: {name?, controlType?, automationId?}): UIElement | null
+
+// 点击元素（InvokePattern）
+clickElement(windowTitle: string, automationId: string): boolean
+
+// 设置元素值（ValuePattern）
+setValue(windowTitle: string, automationId: string, value: string): boolean
+
+// 获取坐标处的元素
+elementAtPoint(x: number, y: number): UIElement | null
+```
+
+**UIElement 类型**：
+```typescript
+interface UIElement {
+  name: string
+  controlType: string    // Button, Edit, Text, List, etc.
+  automationId: string
+  boundingRect: { x: number, y: number, w: number, h: number }
+  isEnabled: boolean
+  value?: string         // ValuePattern 可用时
+  children?: UIElement[]
+}
+```
+
+**PowerShell 脚本核心**：
+```powershell
+Add-Type -AssemblyName UIAutomationClient
+Add-Type -AssemblyName UIAutomationTypes
+
+# 读取 UI 树
+$root = [AutomationElement]::RootElement
+$window = $root.FindFirst([TreeScope]::Children, 
+  [PropertyCondition]::new([AutomationElement]::NameProperty, $title))
+$elements = $window.FindAll([TreeScope]::Descendants, [Condition]::TrueCondition)
+
+# 写入文本
+$element.GetCurrentPattern([ValuePattern]::Pattern).SetValue($text)
+
+# 点击按钮
+$element.GetCurrentPattern([InvokePattern]::Pattern).Invoke()
+```
+
+**验证标准**：
+- 能读取记事本的 UI 树（按钮、文本框、菜单）
+- 能向文本框写入内容
+- 能点击按钮
+- 能识别坐标处的元素
+
+### Phase C：OCR 屏幕文字识别
+
+**问题**：截图后 AI 只能看到图片，无法直接读取文字。
+**方案**：用 `Windows.Media.Ocr` 对截图进行文字识别。
+
+| 步骤 | 文件 | 改动 |
+|------|------|------|
+| C.1 | `src/utils/computerUse/win32/ocr.ts` | 新建：截图 + OCR 识别 |
+| C.2 | `packages/@ant/computer-use-mcp/src/tools.ts` | 增加 `screen_ocr` 工具定义 |
+
+**ocr.ts 导出函数**：
+```typescript
+// 对屏幕区域 OCR
+ocrRegion(x: number, y: number, w: number, h: number, lang?: string): OcrResult
+
+// 对指定窗口 OCR
+ocrWindow(windowTitle: string, lang?: string): OcrResult
+
+interface OcrResult {
+  text: string
+  lines: { text: string, bounds: {x,y,w,h} }[]
+  language: string
+}
+```
+
+**已确认可用语言**：英语 (en-US) + 中文 (zh-Hans-CN)
+
+**验证标准**：
+- 能识别屏幕区域中的英文和中文
+- 返回文字内容 + 每行的位置信息
+
+### Phase D：高频操作性能优化
+
+**问题**：每次 PowerShell 启动 273ms，鼠标移动等高频操作太慢。
+**方案**：用 .NET `System.Windows.Forms.Clipboard` 等直接 API 替代 PowerShell 子进程。
+
+| 步骤 | 文件 | 改动 |
+|------|------|------|
+| D.1 | `src/utils/computerUse/executor.ts` | 剪贴板操作用直接 API 替代 PowerShell |
+| D.2 | 考虑驻留 PowerShell 进程 | 通过 stdin/stdout 交互，摊平启动成本 |
+
+**剪贴板直接 API**（不需要 PowerShell 子进程）：
+```powershell
+# 读：50ms → <1ms
+[System.Windows.Forms.Clipboard]::GetText()
+
+# 写：50ms → <1ms  
+[System.Windows.Forms.Clipboard]::SetText($text)
+
+# 图片检测
+[System.Windows.Forms.Clipboard]::ContainsImage()
+```
+
+### Phase E：`request_access` Windows 适配
+
+**问题**：`request_access` 依赖 macOS bundleId 识别应用，Windows 没有这个概念。
+**方案**：在 Windows 上用 exe 路径 + 窗口标题替代 bundleId。
+
+| 步骤 | 文件 | 改动 |
+|------|------|------|
+| E.1 | `packages/@ant/computer-use-mcp/src/toolCalls.ts` | `resolveRequestedApps` 在 Windows 上用 exe 路径匹配 |
+| E.2 | `packages/@ant/computer-use-mcp/src/sentinelApps.ts` | 增加 Windows 危险应用列表（cmd.exe, powershell.exe 等） |
+| E.3 | `packages/@ant/computer-use-mcp/src/deniedApps.ts` | 增加 Windows 浏览器/终端识别规则 |
+| E.4 | `src/utils/computerUse/hostAdapter.ts` | `ensureOsPermissions` Windows 上检查 UAC 状态 |
+
+**Windows 应用标识映射**：
+```
+macOS bundleId          →  Windows 等价
+com.apple.Safari        →  C:\Program Files\...\msedge.exe（或窗口标题匹配）
+com.google.Chrome       →  chrome.exe
+com.apple.Terminal      →  WindowsTerminal.exe / cmd.exe
+```
+
+### Phase F：全局热键（ESC 拦截）
+
+**问题**：当前非 darwin 直接跳过 ESC 热键，用 Ctrl+C 替代。
+**方案**：用 `RegisterHotKey` 或 `SetWindowsHookEx(WH_KEYBOARD_LL)` 实现。
+
+| 步骤 | 文件 | 改动 |
+|------|------|------|
+| F.1 | `src/utils/computerUse/escHotkey.ts` | Windows 分支：RegisterHotKey 注册 ESC |
+
+**优先级低**——当前 Ctrl+C fallback 可用，ESC 热键是体验优化。
+
+## 5. 执行优先级
+
+```
+Phase A: 窗口绑定截图          ← P0 核心需求，解决"操作其他界面"
+Phase B: UI Automation         ← P0 核心能力，AI 理解 UI 结构
+Phase C: OCR                   ← P1 增值能力，AI 读屏幕文字
+Phase D: 性能优化              ← P1 体验优化，高频操作提速
+Phase E: request_access 适配   ← P1 功能完整性，权限模型适配
+Phase F: ESC 热键              ← P2 体验优化，可后做
+```
+
+## 6. 每个 Phase 的改动量估算
+
+| Phase | 新增文件 | 修改文件 | 新增代码行 | 风险 |
+|-------|---------|---------|-----------|------|
+| A 窗口截图 | 2 | 1 | ~200 | 低 |
+| B UI Automation | 1 | 1 | ~300 | 中 |
+| C OCR | 1 | 1 | ~150 | 低 |
+| D 性能优化 | 0 | 2 | ~50 | 低 |
+| E request_access | 0 | 3 | ~100 | 中 |
+| F ESC 热键 | 0 | 1 | ~50 | 低 |
+| **总计** | **4** | **9** | **~850** | — |
+
+## 7. 不动的文件
+
+- `backends/darwin.ts`（两个包都不动）
+- `backends/linux.ts`（两个包都不动）
+- `src/utils/computerUse/` 中 macOS 相关代码路径不动
+- `packages/@ant/computer-use-mcp/src/` 中已复制的参考项目代码不动（只追加 Windows 工具）
+
+## 8. 与 macOS/Linux 方案的对比
+
+| 能力 | macOS | Windows (增强后) | Linux |
+|------|-------|-----------------|-------|
+| 截图方式 | SCContentFilter (per-app) | **PrintWindow (per-window)** | scrot (全屏/区域) |
+| UI 结构 | Accessibility API | **UI Automation** | 无 |
+| OCR | 无内置 | **Windows.Media.Ocr** | 无内置 |
+| 键鼠 | CGEvent + enigo | SendInput + keybd_event | xdotool |
+| 窗口管理 | NSWorkspace | **EnumWindows + Win32** | wmctrl |
+| 剪贴板 | pbcopy/pbpaste | **Clipboard 直接 API** | xclip |
+| ESC 热键 | CGEventTap | RegisterHotKey | 无 |
+| 应用标识 | bundleId | exe 路径 + 窗口标题 | /proc + wmctrl |
+
+**Windows 增强后将在 UI Automation 和 OCR 方面超过 macOS 方案**——这两项 macOS 原始实现也没有（Anthropic 用的是截图 + Claude 视觉理解，没有结构化 UI 数据）。
--- a/packages/@ant/computer-use-swift/src/backends/win32.ts
+++ b/packages/@ant/computer-use-swift/src/backends/win32.ts
@@ -11,6 +11,9 @@ import type {
  SwiftBackend, WindowDisplayInfo,
 } from '../types.js'

+import { listWindows } from 'src/utils/computerUse/win32/windowEnum.js'
+import { captureWindow, captureWindowByHwnd } from 'src/utils/computerUse/win32/windowCapture.js'
+
 // ---------------------------------------------------------------------------
 // PowerShell helper
 // ---------------------------------------------------------------------------
@@ -155,11 +158,11 @@ $apps | Select-Object -Unique | Select-Object -First 200

  listRunning() {
    try {
-      const raw = ps(`Get-Process | Where-Object { $_.MainWindowTitle -ne '' } | Select-Object -First 50 | ForEach-Object { "$($_.MainModule.FileName)|$($_.ProcessName)" }`)
-      return raw.split('\n').filter(Boolean).map(line => {
-        const [exePath, name] = line.split('|', 2)
-        return { bundleId: exePath ?? '', displayName: name ?? '' }
-      })
+      const windows = listWindows()
+      return windows.map(w => ({
+        bundleId: String(w.hwnd),
+        displayName: w.title,
+      }))
    } catch {
      return []
    }
@@ -246,4 +249,15 @@ $ms.Dispose()
    const base64 = raw.slice(secondComma + 1)
    return { base64, width: w, height: h }
  },
+
+  /**
+   * Capture a specific window by title or HWND using PrintWindow.
+   * Works even for occluded or background windows.
+   */
+  captureWindowTarget(titleOrHwnd: string | number): ScreenshotResult | null {
+    if (typeof titleOrHwnd === 'number') {
+      return captureWindowByHwnd(titleOrHwnd)
+    }
+    return captureWindow(titleOrHwnd)
+  },
 }
--- a/src/utils/computerUse/win32/ocr.ts
+++ b/src/utils/computerUse/win32/ocr.ts
@@ -0,0 +1,257 @@
+/**
+ * OCR module using Windows.Media.Ocr.OcrEngine via PowerShell.
+ * Captures a screen region or window, then runs WinRT OCR to extract text.
+ */
+
+export interface OcrLine {
+  text: string
+  bounds: { x: number; y: number; w: number; h: number }
+}
+
+export interface OcrResult {
+  text: string
+  lines: OcrLine[]
+  language: string
+}
+
+function emptyResult(language: string): OcrResult {
+  return { text: '', lines: [], language }
+}
+
+function runPs(script: string): string {
+  const result = Bun.spawnSync({
+    cmd: ['powershell', '-NoProfile', '-NonInteractive', '-Command', script],
+    stdout: 'pipe',
+    stderr: 'pipe',
+  })
+  return new TextDecoder().decode(result.stdout).trim()
+}
+
+/**
+ * PowerShell script that:
+ * 1. Screenshots a screen region using CopyFromScreen
+ * 2. Saves to temp PNG
+ * 3. Loads via WinRT BitmapDecoder -> SoftwareBitmap
+ * 4. Runs OcrEngine.RecognizeAsync
+ * 5. Outputs JSON with text, lines, and bounding rects
+ */
+function buildOcrRegionScript(
+  x: number,
+  y: number,
+  w: number,
+  h: number,
+  lang: string,
+): string {
+  return `
+Add-Type -AssemblyName System.Drawing
+Add-Type -AssemblyName System.Runtime.WindowsRuntime
+
+# Load WinRT types
+$null = [Windows.Media.Ocr.OcrEngine, Windows.Foundation, ContentType = WindowsRuntime]
+$null = [Windows.Graphics.Imaging.SoftwareBitmap, Windows.Foundation, ContentType = WindowsRuntime]
+$null = [Windows.Graphics.Imaging.BitmapDecoder, Windows.Foundation, ContentType = WindowsRuntime]
+$null = [Windows.Storage.StorageFile, Windows.Foundation, ContentType = WindowsRuntime]
+$null = [Windows.Storage.Streams.RandomAccessStream, Windows.Foundation, ContentType = WindowsRuntime]
+$null = [Windows.Globalization.Language, Windows.Foundation, ContentType = WindowsRuntime]
+
+# Await helper for WinRT async operations
+$asTaskGeneric = ([System.WindowsRuntimeSystemExtensions].GetMethods() | Where-Object {
+    $_.Name -eq 'AsTask' -and $_.GetParameters().Count -eq 1 -and
+    $_.GetParameters()[0].ParameterType.Name -eq 'IAsyncOperation\`1'
+})[0]
+Function Await($WinRtTask, $ResultType) {
+    $asTask = $asTaskGeneric.MakeGenericMethod($ResultType)
+    $netTask = $asTask.Invoke($null, @($WinRtTask))
+    $netTask.Wait(-1) | Out-Null
+    $netTask.Result
+}
+
+try {
+    # Step 1: Screenshot region
+    $bmp = New-Object System.Drawing.Bitmap(${w}, ${h})
+    $g = [System.Drawing.Graphics]::FromImage($bmp)
+    $g.CopyFromScreen(${x}, ${y}, 0, 0, (New-Object System.Drawing.Size(${w}, ${h})))
+    $g.Dispose()
+
+    # Step 2: Save to temp file
+    $tmpFile = [System.IO.Path]::Combine([System.IO.Path]::GetTempPath(), "ocrtemp_$([guid]::NewGuid().ToString('N')).png")
+    $bmp.Save($tmpFile, [System.Drawing.Imaging.ImageFormat]::Png)
+    $bmp.Dispose()
+
+    # Step 3: Open as StorageFile -> BitmapDecoder -> SoftwareBitmap
+    $storageFile = Await ([Windows.Storage.StorageFile]::GetFileFromPathAsync($tmpFile)) ([Windows.Storage.StorageFile])
+    $stream = Await ($storageFile.OpenAsync([Windows.Storage.FileAccessMode]::Read)) ([Windows.Storage.Streams.IRandomAccessStream])
+    $decoder = Await ([Windows.Graphics.Imaging.BitmapDecoder]::CreateAsync($stream)) ([Windows.Graphics.Imaging.BitmapDecoder])
+    $softwareBmp = Await ($decoder.GetSoftwareBitmapAsync()) ([Windows.Graphics.Imaging.SoftwareBitmap])
+
+    # Step 4: Create OCR engine
+    $ocrLang = New-Object Windows.Globalization.Language('${lang}')
+    $engine = [Windows.Media.Ocr.OcrEngine]::TryCreateFromLanguage($ocrLang)
+    if ($engine -eq $null) {
+        # Fallback to en-US
+        $ocrLang = New-Object Windows.Globalization.Language('en-US')
+        $engine = [Windows.Media.Ocr.OcrEngine]::TryCreateFromLanguage($ocrLang)
+    }
+    if ($engine -eq $null) {
+        Write-Output '{"text":"","lines":[],"language":"${lang}"}'
+        return
+    }
+
+    # Step 5: Run OCR
+    $ocrResult = Await ($engine.RecognizeAsync($softwareBmp)) ([Windows.Media.Ocr.OcrResult])
+
+    # Step 6: Extract lines with bounding rects
+    $lines = @()
+    foreach ($line in $ocrResult.Lines) {
+        $minX = [double]::MaxValue; $minY = [double]::MaxValue
+        $maxX = 0.0; $maxY = 0.0
+        foreach ($word in $line.Words) {
+            $r = $word.BoundingRect
+            if ($r.X -lt $minX) { $minX = $r.X }
+            if ($r.Y -lt $minY) { $minY = $r.Y }
+            if (($r.X + $r.Width) -gt $maxX) { $maxX = $r.X + $r.Width }
+            if (($r.Y + $r.Height) -gt $maxY) { $maxY = $r.Y + $r.Height }
+        }
+        $lines += @{
+            text = $line.Text
+            bounds = @{
+                x = [int]$minX
+                y = [int]$minY
+                w = [int]($maxX - $minX)
+                h = [int]($maxY - $minY)
+            }
+        }
+    }
+
+    $output = @{
+        text = $ocrResult.Text
+        lines = $lines
+        language = $ocrLang.LanguageTag
+    }
+    Write-Output (ConvertTo-Json $output -Depth 4 -Compress)
+
+    # Cleanup
+    $stream.Dispose()
+    Remove-Item $tmpFile -ErrorAction SilentlyContinue
+} catch {
+    Write-Output '{"text":"","lines":[],"language":"${lang}"}'
+}
+`
+}
+
+/**
+ * PowerShell script to get a window's bounding rect by title.
+ */
+function buildGetWindowRectScript(windowTitle: string): string {
+  const escaped = windowTitle.replace(/'/g, "''")
+  return `
+Add-Type @'
+using System;
+using System.Runtime.InteropServices;
+public class WinRect {
+    [DllImport("user32.dll", CharSet=CharSet.Unicode)]
+    public static extern IntPtr FindWindow(string c, string t);
+    [DllImport("user32.dll")]
+    public static extern bool GetWindowRect(IntPtr h, out RECT r);
+    [StructLayout(LayoutKind.Sequential)]
+    public struct RECT { public int L, T, R, B; }
+    public static string Get(string title) {
+        IntPtr hwnd = FindWindow(null, title);
+        if (hwnd == IntPtr.Zero) return "NOT_FOUND";
+        RECT r; GetWindowRect(hwnd, out r);
+        int w = r.R - r.L; int h = r.B - r.T;
+        if (w <= 0 || h <= 0) return "INVALID_SIZE";
+        return r.L + "," + r.T + "," + w + "," + h;
+    }
+}
+'@
+[WinRect]::Get('${escaped}')
+`
+}
+
+function parseOcrOutput(raw: string, lang: string): OcrResult {
+  if (!raw) return emptyResult(lang)
+  try {
+    const parsed = JSON.parse(raw)
+    return {
+      text: parsed.text ?? '',
+      lines: Array.isArray(parsed.lines)
+        ? parsed.lines.map((l: any) => ({
+            text: l.text ?? '',
+            bounds: {
+              x: l.bounds?.x ?? 0,
+              y: l.bounds?.y ?? 0,
+              w: l.bounds?.w ?? 0,
+              h: l.bounds?.h ?? 0,
+            },
+          }))
+        : [],
+      language: parsed.language ?? lang,
+    }
+  } catch {
+    return emptyResult(lang)
+  }
+}
+
+/**
+ * Perform OCR on a screen region.
+ * Screenshots the specified rectangle, then runs WinRT OcrEngine.
+ *
+ * @param x - Left coordinate
+ * @param y - Top coordinate
+ * @param w - Width in pixels
+ * @param h - Height in pixels
+ * @param lang - BCP-47 language tag (default 'en-US'). Confirmed: 'en-US', 'zh-Hans-CN'
+ */
+export async function ocrRegion(
+  x: number,
+  y: number,
+  w: number,
+  h: number,
+  lang?: string,
+): Promise<OcrResult> {
+  const language = lang ?? 'en-US'
+  if (w <= 0 || h <= 0) return emptyResult(language)
+
+  try {
+    const script = buildOcrRegionScript(x, y, w, h, language)
+    const raw = runPs(script)
+    return parseOcrOutput(raw, language)
+  } catch {
+    return emptyResult(language)
+  }
+}
+
+/**
+ * Perform OCR on a specific window by its title.
+ * Gets the window rect, then delegates to ocrRegion.
+ *
+ * @param windowTitle - Exact window title to find via FindWindow
+ * @param lang - BCP-47 language tag (default 'en-US')
+ */
+export async function ocrWindow(
+  windowTitle: string,
+  lang?: string,
+): Promise<OcrResult> {
+  const language = lang ?? 'en-US'
+
+  try {
+    const rectScript = buildGetWindowRectScript(windowTitle)
+    const raw = runPs(rectScript)
+    const trimmed = raw.trim()
+
+    if (!trimmed || trimmed === 'NOT_FOUND' || trimmed === 'INVALID_SIZE') {
+      return emptyResult(language)
+    }
+
+    const parts = trimmed.split(',')
+    if (parts.length !== 4) return emptyResult(language)
+
+    const [x, y, w, h] = parts.map(Number)
+    if (!w || !h) return emptyResult(language)
+
+    return ocrRegion(x, y, w, h, lang)
+  } catch {
+    return emptyResult(language)
+  }
+}
--- a/src/utils/computerUse/win32/uiAutomation.ts
+++ b/src/utils/computerUse/win32/uiAutomation.ts
@@ -0,0 +1,308 @@
+/**
+ * Windows UI Automation module
+ *
+ * Provides UI element tree inspection, element lookup, programmatic click,
+ * value setting, and hit-testing via PowerShell + System.Windows.Automation.
+ */
+
+export interface UIElement {
+  name: string
+  controlType: string // Button, Edit, Text, List, Window, etc.
+  automationId: string
+  boundingRect: { x: number; y: number; w: number; h: number }
+  isEnabled: boolean
+  value?: string
+  children?: UIElement[]
+}
+
+// ---------------------------------------------------------------------------
+// Helper
+// ---------------------------------------------------------------------------
+
+const UIA_ASSEMBLIES = `
+Add-Type -AssemblyName UIAutomationClient
+Add-Type -AssemblyName UIAutomationTypes
+Add-Type -AssemblyName WindowsBase
+`
+
+function ps(script: string): string {
+  const result = Bun.spawnSync({
+    cmd: ['powershell', '-NoProfile', '-NonInteractive', '-Command', script],
+    stdout: 'pipe',
+    stderr: 'pipe',
+  })
+  return new TextDecoder().decode(result.stdout).trim()
+}
+
+function parseJsonSafe<T>(raw: string, fallback: T): T {
+  try {
+    if (!raw) return fallback
+    return JSON.parse(raw) as T
+  } catch {
+    return fallback
+  }
+}
+
+// PowerShell snippet that finds a window by exact or partial title match.
+// Assumes $title is already set in the calling script.
+const PS_FIND_WINDOW = `
+$root = [System.Windows.Automation.AutomationElement]::RootElement
+$window = $root.FindFirst(
+  [System.Windows.Automation.TreeScope]::Children,
+  [System.Windows.Automation.PropertyCondition]::new(
+    [System.Windows.Automation.AutomationElement]::NameProperty, $title))
+if ($window -eq $null) {
+  $all = $root.FindAll(
+    [System.Windows.Automation.TreeScope]::Children,
+    [System.Windows.Automation.Condition]::TrueCondition)
+  foreach ($el in $all) {
+    if ($el.Current.Name -and $el.Current.Name.Contains($title)) {
+      $window = $el
+      break
+    }
+  }
+}
+`
+
+// ---------------------------------------------------------------------------
+// Public API
+// ---------------------------------------------------------------------------
+
+/**
+ * Get the UI element tree of a window, up to `depth` levels deep (default 3).
+ */
+export function getUITree(windowTitle: string, depth: number = 3): UIElement[] {
+  const escapedTitle = windowTitle.replace(/'/g, "''")
+  const script = `
+${UIA_ASSEMBLIES}
+$title = '${escapedTitle}'
+${PS_FIND_WINDOW}
+if ($window -eq $null) {
+  Write-Output '[]'
+  exit
+}
+
+function Get-UIChildren($parent, $currentDepth, $maxDepth) {
+  if ($currentDepth -ge $maxDepth) { return @() }
+  $children = $parent.FindAll(
+    [System.Windows.Automation.TreeScope]::Children,
+    [System.Windows.Automation.Condition]::TrueCondition)
+  $result = @()
+  foreach ($el in $children) {
+    $rect = $el.Current.BoundingRectangle
+    $obj = @{
+      name = [string]$el.Current.Name
+      controlType = $el.Current.ControlType.ProgrammaticName -replace 'ControlType\\.', ''
+      automationId = [string]$el.Current.AutomationId
+      boundingRect = @{
+        x = [int]$rect.X
+        y = [int]$rect.Y
+        w = [int]$rect.Width
+        h = [int]$rect.Height
+      }
+      isEnabled = $el.Current.IsEnabled
+    }
+    try {
+      $vp = $el.GetCurrentPattern([System.Windows.Automation.ValuePattern]::Pattern)
+      if ($vp -ne $null) { $obj['value'] = $vp.Current.Value }
+    } catch {}
+    $sub = Get-UIChildren $el ($currentDepth + 1) $maxDepth
+    if ($sub.Count -gt 0) { $obj['children'] = $sub }
+    $result += $obj
+  }
+  return $result
+}
+
+$tree = Get-UIChildren $window 0 ${depth}
+if ($tree -eq $null -or $tree.Count -eq 0) {
+  Write-Output '[]'
+} else {
+  $tree | ConvertTo-Json -Depth 20 -Compress
+}
+`
+  const raw = ps(script)
+  const parsed = parseJsonSafe<UIElement | UIElement[]>(raw, [])
+  return Array.isArray(parsed) ? parsed : [parsed]
+}
+
+/**
+ * Find a single element inside a window matching the given query fields.
+ */
+export function findElement(
+  windowTitle: string,
+  query: { name?: string; controlType?: string; automationId?: string },
+): UIElement | null {
+  const escapedTitle = windowTitle.replace(/'/g, "''")
+
+  // Build conditions array
+  const conditions: string[] = []
+  if (query.name) {
+    const v = query.name.replace(/'/g, "''")
+    conditions.push(
+      `[System.Windows.Automation.PropertyCondition]::new([System.Windows.Automation.AutomationElement]::NameProperty, '${v}')`,
+    )
+  }
+  if (query.controlType) {
+    const v = query.controlType.replace(/'/g, "''")
+    conditions.push(
+      `[System.Windows.Automation.PropertyCondition]::new([System.Windows.Automation.AutomationElement]::ControlTypeProperty, [System.Windows.Automation.ControlType]::${v})`,
+    )
+  }
+  if (query.automationId) {
+    const v = query.automationId.replace(/'/g, "''")
+    conditions.push(
+      `[System.Windows.Automation.PropertyCondition]::new([System.Windows.Automation.AutomationElement]::AutomationIdProperty, '${v}')`,
+    )
+  }
+
+  if (conditions.length === 0) return null
+
+  let conditionExpr: string
+  if (conditions.length === 1) {
+    conditionExpr = conditions[0]
+  } else {
+    conditionExpr = `[System.Windows.Automation.AndCondition]::new(@(${conditions.join(', ')}))`
+  }
+
+  const script = `
+${UIA_ASSEMBLIES}
+$title = '${escapedTitle}'
+${PS_FIND_WINDOW}
+if ($window -eq $null) {
+  Write-Output 'null'
+  exit
+}
+$cond = ${conditionExpr}
+$el = $window.FindFirst([System.Windows.Automation.TreeScope]::Descendants, $cond)
+if ($el -eq $null) {
+  Write-Output 'null'
+  exit
+}
+$rect = $el.Current.BoundingRectangle
+$obj = @{
+  name = [string]$el.Current.Name
+  controlType = $el.Current.ControlType.ProgrammaticName -replace 'ControlType\\.', ''
+  automationId = [string]$el.Current.AutomationId
+  boundingRect = @{
+    x = [int]$rect.X
+    y = [int]$rect.Y
+    w = [int]$rect.Width
+    h = [int]$rect.Height
+  }
+  isEnabled = $el.Current.IsEnabled
+}
+try {
+  $vp = $el.GetCurrentPattern([System.Windows.Automation.ValuePattern]::Pattern)
+  if ($vp -ne $null) { $obj['value'] = $vp.Current.Value }
+} catch {}
+$obj | ConvertTo-Json -Compress
+`
+  const raw = ps(script)
+  return parseJsonSafe<UIElement | null>(raw, null)
+}
+
+/**
+ * Click an element by its automationId using InvokePattern.
+ */
+export function clickElement(windowTitle: string, automationId: string): boolean {
+  const escapedTitle = windowTitle.replace(/'/g, "''")
+  const escapedId = automationId.replace(/'/g, "''")
+
+  const script = `
+${UIA_ASSEMBLIES}
+$title = '${escapedTitle}'
+${PS_FIND_WINDOW}
+if ($window -eq $null) {
+  Write-Output 'false'
+  exit
+}
+$cond = [System.Windows.Automation.PropertyCondition]::new(
+  [System.Windows.Automation.AutomationElement]::AutomationIdProperty, '${escapedId}')
+$el = $window.FindFirst([System.Windows.Automation.TreeScope]::Descendants, $cond)
+if ($el -eq $null) {
+  Write-Output 'false'
+  exit
+}
+try {
+  $ip = $el.GetCurrentPattern([System.Windows.Automation.InvokePattern]::Pattern)
+  $ip.Invoke()
+  Write-Output 'true'
+} catch {
+  Write-Output 'false'
+}
+`
+  return ps(script) === 'true'
+}
+
+/**
+ * Set the value of an element by its automationId using ValuePattern.
+ */
+export function setValue(windowTitle: string, automationId: string, value: string): boolean {
+  const escapedTitle = windowTitle.replace(/'/g, "''")
+  const escapedId = automationId.replace(/'/g, "''")
+  const escapedValue = value.replace(/'/g, "''")
+
+  const script = `
+${UIA_ASSEMBLIES}
+$title = '${escapedTitle}'
+${PS_FIND_WINDOW}
+if ($window -eq $null) {
+  Write-Output 'false'
+  exit
+}
+$cond = [System.Windows.Automation.PropertyCondition]::new(
+  [System.Windows.Automation.AutomationElement]::AutomationIdProperty, '${escapedId}')
+$el = $window.FindFirst([System.Windows.Automation.TreeScope]::Descendants, $cond)
+if ($el -eq $null) {
+  Write-Output 'false'
+  exit
+}
+try {
+  $vp = $el.GetCurrentPattern([System.Windows.Automation.ValuePattern]::Pattern)
+  $vp.SetValue('${escapedValue}')
+  Write-Output 'true'
+} catch {
+  Write-Output 'false'
+}
+`
+  return ps(script) === 'true'
+}
+
+/**
+ * Get the UI element at a specific screen coordinate.
+ */
+export function elementAtPoint(x: number, y: number): UIElement | null {
+  const script = `
+${UIA_ASSEMBLIES}
+try {
+  $point = [System.Windows.Point]::new(${x}, ${y})
+  $el = [System.Windows.Automation.AutomationElement]::FromPoint($point)
+  if ($el -eq $null) {
+    Write-Output 'null'
+    exit
+  }
+  $rect = $el.Current.BoundingRectangle
+  $obj = @{
+    name = [string]$el.Current.Name
+    controlType = $el.Current.ControlType.ProgrammaticName -replace 'ControlType\\.', ''
+    automationId = [string]$el.Current.AutomationId
+    boundingRect = @{
+      x = [int]$rect.X
+      y = [int]$rect.Y
+      w = [int]$rect.Width
+      h = [int]$rect.Height
+    }
+    isEnabled = $el.Current.IsEnabled
+  }
+  try {
+    $vp = $el.GetCurrentPattern([System.Windows.Automation.ValuePattern]::Pattern)
+    if ($vp -ne $null) { $obj['value'] = $vp.Current.Value }
+  } catch {}
+  $obj | ConvertTo-Json -Compress
+} catch {
+  Write-Output 'null'
+}
+`
+  const raw = ps(script)
+  return parseJsonSafe<UIElement | null>(raw, null)
+}
--- a/src/utils/computerUse/win32/windowCapture.ts
+++ b/src/utils/computerUse/win32/windowCapture.ts
@@ -0,0 +1,129 @@
+/**
+ * Window-level screenshot capture using Win32 PrintWindow API.
+ * Captures windows even when occluded or minimized.
+ */
+
+interface CaptureResult {
+  base64: string
+  width: number
+  height: number
+}
+
+const CAPTURE_BY_TITLE_PS = `
+Add-Type -AssemblyName System.Drawing
+Add-Type -ReferencedAssemblies System.Drawing @'
+using System;
+using System.Runtime.InteropServices;
+using System.Drawing;
+using System.Drawing.Imaging;
+public class WinCap {
+    [DllImport("user32.dll", CharSet=CharSet.Unicode)]
+    public static extern IntPtr FindWindow(string c, string t);
+    [DllImport("user32.dll")]
+    public static extern bool GetWindowRect(IntPtr h, out RECT r);
+    [DllImport("user32.dll")]
+    public static extern bool PrintWindow(IntPtr h, IntPtr hdc, uint f);
+    [StructLayout(LayoutKind.Sequential)]
+    public struct RECT { public int L, T, R, B; }
+
+    public static string Capture(string title) {
+        IntPtr hwnd = FindWindow(null, title);
+        if (hwnd == IntPtr.Zero) return "NOT_FOUND";
+        RECT r; GetWindowRect(hwnd, out r);
+        int w = r.R - r.L; int h = r.B - r.T;
+        if (w <= 0 || h <= 0) return "INVALID_SIZE";
+        Bitmap bmp = new Bitmap(w, h);
+        Graphics g = Graphics.FromImage(bmp);
+        IntPtr hdc = g.GetHdc();
+        PrintWindow(hwnd, hdc, 2);
+        g.ReleaseHdc(hdc); g.Dispose();
+        var ms = new System.IO.MemoryStream();
+        bmp.Save(ms, ImageFormat.Png);
+        bmp.Dispose();
+        return w + "," + h + "," + Convert.ToBase64String(ms.ToArray());
+    }
+}
+'@
+`
+
+const CAPTURE_BY_HWND_PS = `
+Add-Type -AssemblyName System.Drawing
+Add-Type -ReferencedAssemblies System.Drawing @'
+using System;
+using System.Runtime.InteropServices;
+using System.Drawing;
+using System.Drawing.Imaging;
+public class WinCapH {
+    [DllImport("user32.dll")]
+    public static extern bool GetWindowRect(IntPtr h, out RECT r);
+    [DllImport("user32.dll")]
+    public static extern bool PrintWindow(IntPtr h, IntPtr hdc, uint f);
+    [DllImport("user32.dll")]
+    public static extern bool IsWindow(IntPtr hWnd);
+    [StructLayout(LayoutKind.Sequential)]
+    public struct RECT { public int L, T, R, B; }
+
+    public static string Capture(IntPtr hwnd) {
+        if (!IsWindow(hwnd)) return "NOT_FOUND";
+        RECT r; GetWindowRect(hwnd, out r);
+        int w = r.R - r.L; int h = r.B - r.T;
+        if (w <= 0 || h <= 0) return "INVALID_SIZE";
+        Bitmap bmp = new Bitmap(w, h);
+        Graphics g = Graphics.FromImage(bmp);
+        IntPtr hdc = g.GetHdc();
+        PrintWindow(hwnd, hdc, 2);
+        g.ReleaseHdc(hdc); g.Dispose();
+        var ms = new System.IO.MemoryStream();
+        bmp.Save(ms, ImageFormat.Png);
+        bmp.Dispose();
+        return w + "," + h + "," + Convert.ToBase64String(ms.ToArray());
+    }
+}
+'@
+`
+
+function parseCaptureOutput(raw: string): CaptureResult | null {
+  const trimmed = raw.trim()
+  if (!trimmed || trimmed === 'NOT_FOUND' || trimmed === 'INVALID_SIZE') {
+    return null
+  }
+  const firstComma = trimmed.indexOf(',')
+  const secondComma = trimmed.indexOf(',', firstComma + 1)
+  if (firstComma === -1 || secondComma === -1) return null
+
+  const width = Number(trimmed.slice(0, firstComma))
+  const height = Number(trimmed.slice(firstComma + 1, secondComma))
+  const base64 = trimmed.slice(secondComma + 1)
+
+  if (!width || !height || !base64) return null
+  return { base64, width, height }
+}
+
+function runPs(script: string): string {
+  const result = Bun.spawnSync({
+    cmd: ['powershell', '-NoProfile', '-NonInteractive', '-Command', script],
+    stdout: 'pipe',
+    stderr: 'pipe',
+  })
+  return new TextDecoder().decode(result.stdout).trim()
+}
+
+/**
+ * Capture a window screenshot by its exact title.
+ * Uses PrintWindow which works even for occluded/background windows.
+ */
+export function captureWindow(title: string): CaptureResult | null {
+  const escaped = title.replace(/'/g, "''")
+  const script = `${CAPTURE_BY_TITLE_PS}\n[WinCap]::Capture('${escaped}')`
+  const raw = runPs(script)
+  return parseCaptureOutput(raw)
+}
+
+/**
+ * Capture a window screenshot by its HWND handle.
+ */
+export function captureWindowByHwnd(hwnd: number): CaptureResult | null {
+  const script = `${CAPTURE_BY_HWND_PS}\n[WinCapH]::Capture([IntPtr]::new(${hwnd}))`
+  const raw = runPs(script)
+  return parseCaptureOutput(raw)
+}
--- a/src/utils/computerUse/win32/windowEnum.ts
+++ b/src/utils/computerUse/win32/windowEnum.ts
@@ -0,0 +1,86 @@
+/**
+ * Window enumeration using Win32 EnumWindows API.
+ * Returns visible windows with their HWND, PID, and title.
+ */
+
+export interface WindowInfo {
+  hwnd: number
+  pid: number
+  title: string
+}
+
+const ENUM_WINDOWS_PS = `
+Add-Type @'
+using System;
+using System.Collections.Generic;
+using System.Runtime.InteropServices;
+using System.Text;
+public class WinEnum {
+    public delegate bool EnumWindowsProc(IntPtr hWnd, IntPtr lParam);
+
+    [DllImport("user32.dll")]
+    public static extern bool EnumWindows(EnumWindowsProc lpEnumFunc, IntPtr lParam);
+    [DllImport("user32.dll")]
+    public static extern bool IsWindowVisible(IntPtr hWnd);
+    [DllImport("user32.dll", CharSet=CharSet.Unicode)]
+    public static extern int GetWindowText(IntPtr hWnd, StringBuilder lpString, int nMaxCount);
+    [DllImport("user32.dll")]
+    public static extern int GetWindowTextLength(IntPtr hWnd);
+    [DllImport("user32.dll")]
+    public static extern uint GetWindowThreadProcessId(IntPtr hWnd, out uint processId);
+
+    public static List<string> results = new List<string>();
+
+    public static void Run() {
+        results.Clear();
+        EnumWindows(delegate(IntPtr hWnd, IntPtr lParam) {
+            if (!IsWindowVisible(hWnd)) return true;
+            int len = GetWindowTextLength(hWnd);
+            if (len == 0) return true;
+            StringBuilder sb = new StringBuilder(len + 1);
+            GetWindowText(hWnd, sb, sb.Capacity);
+            string title = sb.ToString();
+            if (string.IsNullOrWhiteSpace(title)) return true;
+            uint pid = 0;
+            GetWindowThreadProcessId(hWnd, out pid);
+            results.Add(hWnd.ToInt64() + "|" + pid + "|" + title);
+            return true;
+        }, IntPtr.Zero);
+    }
+}
+'@
+[WinEnum]::Run()
+[WinEnum]::results | ForEach-Object { $_ }
+`
+
+/**
+ * List all visible windows with non-empty titles.
+ * Returns HWND, PID, and window title for each.
+ */
+export function listWindows(): WindowInfo[] {
+  const result = Bun.spawnSync({
+    cmd: ['powershell', '-NoProfile', '-NonInteractive', '-Command', ENUM_WINDOWS_PS],
+    stdout: 'pipe',
+    stderr: 'pipe',
+  })
+  const raw = new TextDecoder().decode(result.stdout).trim()
+  if (!raw) return []
+
+  return raw
+    .split('\n')
+    .filter(Boolean)
+    .map(line => {
+      const trimmed = line.trim()
+      const firstPipe = trimmed.indexOf('|')
+      const secondPipe = trimmed.indexOf('|', firstPipe + 1)
+      if (firstPipe === -1 || secondPipe === -1) return null
+
+      const hwnd = Number(trimmed.slice(0, firstPipe))
+      const pid = Number(trimmed.slice(firstPipe + 1, secondPipe))
+      const title = trimmed.slice(secondPipe + 1)
+
+      if (isNaN(hwnd) || isNaN(pid) || !title) return null
+      return { hwnd, pid, title }
+    })
+    .filter((item): item is WindowInfo => item !== null)
+}