ScreenTools

Documentation: English · Français · 简体中文

ScreenTools is an experimental Windows desktop utility that brings screenshots, clipboard history, screen recording, and instant replay into one lightweight application.

Project status: The application is functional but is not yet production-ready. The documented 15/60-minute physical recording matrix, Windows 10 hardware verification, multi-monitor behavior, and release automation still need real-world validation.

Features

• Screenshots with global shortcuts, format and output-folder settings, and recent-output history.
• Clipboard history management.
• Screen recording with start, pause, resume, quality profiles, microphone capture, and MP4 export through FFmpeg.
• Instant replay with a rolling frame buffer capped at 160 MiB, time-based trimming, and clip export.
• Lower-allocation capture encoding: captured bitmaps are reused when possible, and resized frames use a 24-bit pixel format before JPEG encoding.
• Settings, output history, and clipboard history are written through temporary files and replaced atomically. Corrupted JSON settings and history files are preserved separately for recovery.
• System-audio loopback capture through the Windows default playback endpoint using NAudio/WASAPI; Stereo Mix is not required. Microphone and system audio are timestamped and mixed during MP4 export.
• Recording diagnostics include captured/skipped/failed frame counts, JPEG bytes written, and peak process working-set/private memory in the recording manifest.
• Per-Monitor V2 DPI awareness with per-monitor region-selection overlays and monitor-aware floating-window placement.
• MSIX packaging project for Store and sideload distribution.
• 97 xUnit tests covering persistence, replay limits, localization, recording lifecycle/pause-resume, MP4 duration and audio continuity, FFmpeg mixing, audio failure isolation, real Windows microphone/WASAPI capture when endpoints are available, and physical capture probes on every connected display.
• A standalone benchmark harness for real desktop capture and synthetic replay-buffer workloads.

Known gaps

• Execute the documented 15/60-minute recording matrix on physical Windows 11 and Windows 10 hardware.
• Measure replay duration, frame drops, memory use, and concurrent recording/replay workloads under sustained load.
• Complete the physical multi-monitor/high-DPI acceptance matrix (125%, 150%, mixed DPI, vertical layouts, and Windows 10).
• Expand integration tests and automate packaging and releases.

See AUDIO_TESTING.md for the WASAPI capture design and hardware checks; RECORDING_VALIDATION.md for the recording stability matrix; MULTI_MONITOR_VALIDATION.md for issue #4's display matrix and diagnostic script; REMAINING_WORK.md for the current backlog; and DESIGN.md for the UI standards.

Technology

• C# and WPF on .NET 8 for Windows.
• FFmpeg for MP4 encoding.
• NAudio for WASAPI system-audio capture.
• xUnit for automated tests.

Build, test, and run

Requirements:

• Windows 10 or Windows 11.
• .NET 8 SDK.
• FFmpeg for MP4 export. The application checks its application directory, common installation locations (including Chocolatey), and trusted entries on PATH.

Run the solution commands from the ScreenTools solution directory so that project-relative paths resolve correctly:

Push-Location .\ScreenTools
try {
    dotnet restore .\ScreenTools.sln
    dotnet build .\ScreenTools.sln -c Release
    dotnet test .\ScreenTools.sln -c Release
    dotnet run --project .\ScreenTools\ScreenTools.csproj -c Release
} finally {
    Pop-Location
}

Performance benchmarks

Run the benchmark harness from the solution directory:

Push-Location .\ScreenTools
try {
    dotnet run --project ..\ScreenTools.Benchmarks\ScreenTools.Benchmarks.csproj -c Release -- capture --iterations 60 --csv ..\BenchmarkResults\capture.csv
    dotnet run --project ..\ScreenTools.Benchmarks\ScreenTools.Benchmarks.csproj -c Release -- buffer --iterations 340 --csv ..\BenchmarkResults\buffer.csv
} finally {
    Pop-Location
}

capture measures real desktop capture, resizing, and JPEG encoding. buffer measures replay-buffer insertion and verifies the configured memory ceiling with synthetic 512 KiB frame payloads. all runs both scenarios. CSV output is optional; benchmark results are intentionally ignored by Git.

Without FFmpeg, captured frames can still be retained as an image sequence, but MP4 export is unavailable.

Contributing

Reliability improvements are more valuable than adding features at this stage. Please read CONTRIBUTING.md, review the open issues, and discuss substantial architectural changes before submitting a large patch.

License

MIT. See LICENSE.