Download Shadow Claude for Windows

The only AI interview assistant powered by Claude. Invisible to screen share, GPU-accelerated Whisper transcription, resume-aware answers. Now with model choice and NVIDIA CUDA acceleration.

Download for Windows -Free

sh_claw_0.2.0_x64-setup.exe · ~367 MB · Windows 10+ · Includes CUDA

100% Invisible to Screen Share

The overlay is completely hidden from all screen capture and sharing software. Invisible to Zoom, Teams, Meet, Discord, OBS, and the Snipping Tool.

Local Whisper Transcription

Audio is transcribed on-device using OpenAI's Whisper model. Your interview audio never leaves your machine -zero cloud dependency for speech-to-text.

Claude-Powered Answers

Anthropic's Claude generates natural, context-aware answers using your resume and job description. Responses stream in real-time -first token in under 1 second.

Screenshot Analysis

Capture LeetCode problems, system design diagrams, or any on-screen content. Vision AI reads constraints, examples, and code, then generates solutions instantly.

<2s
Transcription
<4s
End-to-end
99.9%
Invisibility
150 MB
Install size

Windows SmartScreen Warning

Windows SmartScreen is a reputation-based security feature that flags software with low download volume. Because Shadow Claude is new, you'll see a "Windows protected your PC" popup. This is standard for any new application and is not a security concern.

1
Click "More info"
When you see the blue SmartScreen popup, click the small 'More info' text link at the bottom.
2
Click "Run anyway"
A 'Run anyway' button will appear. Click it to proceed with the installation.
3
Follow the installer
The installer will guide you through setup. Shadow Claude will launch automatically after installation.

Technical Specifications

File namesh_claw_0.2.0_x64-setup.exe
File size~367 MB (app + CUDA libraries)
Install size~400 MB + model size (142MB–1.5GB)
OSWindows 10 or later, 64-bit
RAM usage~150 MB + model (Base 142MB, Small 466MB, Medium 1.5GB)
GPUNVIDIA GPU supported (CUDA). Falls back to CPU automatically.
ModelsDownloaded on first launch (Base ~142MB, Small ~466MB, Medium ~1.5GB)
Invisibility99.9%* -excluded from all screen capture via WDA_EXCLUDEFROMCAPTURE
NetworkRequired for AI answers and model download. Not for transcription.
Version0.2.0

*99.9% invisible to all screen capture and sharing software including Zoom, Teams, Meet, Discord, and OBS. The only methods that can detect it -window enumeration and hardware capture -are never used in interview platforms.

What's Included in Every Plan

  • Invisible overlay -excluded from all screen sharing and recording tools using OS-level APIs
  • Local Whisper AI -the speech recognition model (~148 MB) ships bundled with the app, no download required
  • Claude-powered answers -real-time AI responses streamed via SSE from Anthropic's Claude
  • Resume integration -upload your resume and job description for personalized, context-aware answers
  • Voice Activity Detection -automatic speech detection triggers transcription without manual input
  • Global hotkeys -Ctrl+Shift+G (toggle overlay), Ctrl+Shift+L (listen), Ctrl+Shift+A (force answer), Ctrl+Shift+C (copy)

Frequently Asked Questions

Currently available for Windows 10 or later (64-bit)

macOS support coming soon