A modern, feature-rich Windows Task Manager alternative with real-time performance monitoring, intelligent system insights, and a premium dark UI.
This application is not code-signed. Windows SmartScreen may show a warning when you first run it. See Installation for how to bypass this safely.
Download Latest Release (Windows x64)
Download the TaskManagerPlus_x.x.x_x64-setup.exe installer from the latest release.
Because this app is not code-signed (code signing certificates cost $200-400/year), Windows will show a SmartScreen warning the first time you run the installer or the app.
How to bypass:
- When you see "Windows protected your PC", click "More info"
- Click "Run anyway"
This is standard behavior for any unsigned application. The full source code is available in this repository for review.
- Download
TaskManagerPlus_x.x.x_x64-setup.exefrom Releases - Run the installer (bypass SmartScreen as described above)
- The app installs and creates a Start Menu shortcut
- Updates are checked automatically on launch
- Real-time process list grouped by application with expand/collapse
- CPU, memory, disk I/O, network, GPU, and power usage per process
- Process icons, status badges (running/suspended), and group counts
- Right-click context menu with End Task and priority controls
- Search/filter processes by name
- Configurable column visibility
- CPU - Live usage graph, per-core grid with P-core/E-core labels, frequency tracking
- Memory - Usage graph with stacked breakdown, composition bar (in use/cached/available), memory pressure indicator, page file usage
- Disk - Read/write throughput graph, active %, queue depth, top consumers
- Network - Send/receive throughput graph, link speed, top bandwidth consumers
- GPU - Usage and VRAM graph, temperature monitoring (via D3DKMT API), top GPU consumers
- Battery - Power draw graph, charge status, battery health, power flow visualization, per-process power estimates, cycle count
- Performance Score (0-100) with real-time system health gauge
- Memory leak detection using linear regression on per-process memory trends
- Workload detection - Automatically identifies gaming, editing, development, browsing, streaming, office, or idle workloads based on running processes
- Fan profile suggestions - Recommends Silent/Balanced/Performance/Turbo based on detected workload
- Optimization suggestions - Context-aware recommendations to close unneeded apps (e.g., "Close browser to free 1.2 GB for gaming")
- Bottleneck detection - Sustained high CPU, disk saturation, network saturation alerts
- GPU temperature warnings with thermal throttling alerts
- Battery health assessment and high power drain warnings
- Handle/thread leak detection via trend analysis
- Analysis runs continuously in background, persists across tab switches
- Bundled leak classifier — a ~4 KB on-device decision tree, compiled into the app, sharpens memory-leak detection by distinguishing a genuine leak from benign cache warmup or a one-off startup spike. It runs for everyone; there is nothing to enable.
- AI tier setting in Settings — choose Off, Standard, or Enhanced. This governs only the optional semantic embedding model used by the file organizer. Defaults to Off.
- All models run entirely on your device. Nothing is uploaded; the only network call AI ever makes is the one-time Enhanced model download from GitHub release assets.
- AI is strictly a refinement layer: every feature works without the embedding model using the rules-based engine. No feature depends on a tier above Off.
- Compact popup showing CPU, memory, disk, network, GPU at a glance
- Smart positioning that stays within monitor bounds
- Click to open full app
- Dark/Light theme with accent color presets (8 colors)
- Graph size: Small / Medium / Large
- Display mode: Percent / Values
- Temperature unit: Celsius / Fahrenheit
- Column visibility toggles
- Update interval (500ms - 5s)
- Confirm before ending tasks toggle
- Minimize to tray toggle
- Checks for updates from GitHub Releases on launch
- One-click update and restart
- Cryptographically signed update packages
| Layer | Technology |
|---|---|
| Frontend | React 19, TypeScript 6, Vite 8 |
| Backend | Tauri v2 (Rust) as IPC bridge |
| Native | C++ DLL for Windows performance counters, process enumeration, GPU telemetry |
| UI | Custom CSS with glassmorphism, neumorphic controls, system-native fonts |
| Charts | Canvas-based real-time graphs with requestAnimationFrame |
| State | TanStack React Query for data fetching, global listener pattern for settings |
| Distribution | NSIS installer, Tauri auto-updater with signature verification |
React UI (TypeScript)
|
| Tauri IPC (invoke)
v
Rust Backend (src-tauri/)
|
| FFI (libloading)
v
C++ DLL (native/)
|
| Windows APIs
v
PDH Counters, DXGI, D3DKMT, NtQuerySystemInformation
- PDH (Performance Data Helper) - CPU, disk, network counters
- PROCESS_MEMORY_COUNTERS_EX - Per-process memory (private, shared, working set)
- DXGI - GPU adapter enumeration and VRAM info
- D3DKMTQueryAdapterInfo - Cross-vendor GPU temperature (WDDM 2.5+)
- NtQuerySystemInformation - CPU set topology (P-core/E-core detection)
- GetSystemPowerStatus / IOCTL_BATTERY - Battery and power telemetry
- Node.js (v18+)
- Rust (latest stable)
- Visual Studio 2022 with C++ Desktop workload
- CMake (3.20+)
# Clone
git clone https://github.com/Elitelord/TaskManagerPlus.git
cd TaskManagerPlus
# Install dependencies
npm install
# Build native DLL (first time only)
cd native && cmake -B build -G "Visual Studio 17 2022" && cmake --build build --config Release && cd ..
# Dev mode (hot reload)
npx tauri dev
# Production build
npx tauri buildThe installer will be at src-tauri/target/release/bundle/nsis/TaskManagerPlus_x.x.x_x64-setup.exe.
TaskManager+ is built to be fully offline. It collects, transmits, and stores no personal data.
- No telemetry, no analytics, no accounts. The app does not phone home.
- Process and performance data never leaves your machine. It is read from Windows APIs, shown in the UI, and discarded — nothing is logged off-device.
- AI runs locally. All inference executes on your CPU. The leak classifier is compiled into the app; the Standard embedding model is bundled too. No process names, memory series, file paths, or any other data are sent anywhere for inference — ever. The single exception is the Enhanced tier, which downloads its larger model file once from GitHub release assets; that download moves a model to you and sends nothing about you.
- The only network request the app makes is the auto-updater checking GitHub Releases for a newer version on launch. This sends no data about you — it is a plain version check — and can be ignored entirely by not installing updates.
The webview is locked down with a Content-Security-Policy of default-src 'self', so the UI cannot make outbound requests even in principle. A unit test (src/lib/ai/telemetry.test.ts) asserts the AI code paths perform no network calls regardless of tier.
This software is provided as-is, without warranty of any kind. This application is not code-signed and is not verified by Microsoft. Windows SmartScreen and antivirus software may flag it as unrecognized. The complete source code is available in this repository for transparency and security review.
This application requires access to Windows performance counters and process information to function. It does not collect, transmit, or store any personal data.
ISC