Table of Contents
Installation
Prerequisites
- OS: Windows 10+ (64-bit), macOS 12+, or Ubuntu 22+
- RAM: 4 GB minimum, 8 GB recommended
- Disk: ~200 MB free space
Steps
- Download the installer for your platform from the Download section
- Run the installer (e.g.,
ServiceHost-Setup.exeon Windows) - Follow the Next → Next → Finish wizard
- The application launches automatically after installation
First-Time Setup
- Welcome Screen: The app opens with a guided setup wizard. Follow the steps or click "Skip setup."
- Choose AI Provider: Open Settings and select your preferred provider:
- OpenAI
- Anthropic
- Google Gemini
- Best Model
- Mistral
- Moonshot AI
- Configure Models: Assign specific models for extraction, solution generation, and debugging
- Login: Create an account or log in to unlock full features
Daily Usage Guide
Basic Workflow — Screenshot to Solution
- Press
Ctrl+Bto show/hide the window - Navigate to the coding problem you want to solve
- Press
Ctrl+Hto capture a screenshot (repeat for more context) - Press
Ctrl+Enterto extract the problem and generate a solution - Review the solution — ask follow-up questions if needed
- Press
Ctrl+Eto reset session when done
Interview Assistant — Voice Mode
- Press
Alt+Qto start microphone recording - The interviewer's question is transcribed in real-time
- AI automatically generates an answer based on the transcription
- Press
Alt+Qagain to stop recording
System Audio Capture
- Press
Alt+Wto start capturing your computer's audio output - Audio from any application (Zoom, Chrome, etc.) is transcribed live
- Press
Alt+Wagain to stop
Keyboard Shortcuts Reference
| Shortcut | Action |
|---|---|
Ctrl+H |
Take screenshot |
Ctrl+Enter |
Process screenshots / Submit AI input |
Ctrl+L |
Delete last screenshot |
Ctrl+R |
Reset view |
Ctrl+E |
Force cancel & reset entire session |
Ctrl+B |
Toggle window visibility |
Ctrl+W |
Focus UI for scrolling/interaction |
Ctrl+Q |
Quit application |
Ctrl+Arrow |
Move window (left/right/up/down) |
Alt+Arrow |
Scroll content |
Alt+Q |
Toggle interview assistant |
Alt+W |
Toggle system audio capture |
Ctrl+[ / Ctrl+] |
Decrease / increase opacity |
Ctrl+- / Ctrl+= |
Zoom out / zoom in |
Ctrl+0 |
Reset zoom |
AI Configuration
Noxans uses three separate AI models that you can configure independently:
- Problem Extraction Model: Analyzes screenshots to extract the coding problem, constraints, and examples
- Solution Generation Model: Generates complete, optimized code solutions with complexity analysis
- Debugging Model: Analyzes errors and improves solutions iteratively
Tip: Use a vision-capable model (gemini 3 pro) for extraction, and a code-focused model for solutions and debugging.
Interview Assistant
The Interview Assistant provides two modes of audio capture:
- Microphone Mode (
Alt+Q): Records your microphone input. Best for capturing questions asked directly to you. - System Audio Mode (
Alt+W): Captures all audio playing on your computer via WASAPI loopback. Ideal for transcribing questions from video calls.
Both modes use OpenAI's Whisper for transcription. Ensure Python 3.9+ is installed with the required packages.
Best Practices
- Multiple Screenshots: Capture all relevant parts of a problem — statement, examples, constraints
- Use Context: When pressing Ctrl+H, you can add optional context before processing
- Choose the Right Model: Different models excel at different tasks. Experiment to find your optimal setup
- Keep Sessions Clean: Use Ctrl+E to reset between different problems
- Adjust Opacity: Use Ctrl+[ and Ctrl+] to find the right transparency level
- Move Window: Use Ctrl+Arrow keys to position the window where it doesn't overlap your work
Troubleshooting
- No screenshots appearing? Ensure the window is visible (Ctrl+B). Some screen capture software may conflict.
- AI not responding? Check Your Internet Connection
- Window not invisible? Stealth features are production-only. Run the installed version, not dev mode.
- High CPU usage? This typically occurs during AI processing. The app returns to idle priority after completion.