Feature Inventory
Every capability packed into the phantom. From hardware-level screen protection to real-time voice transcription, Imposter is engineered for absolute discretion and maximum productivity.
Stealth Window System
The core innovation. A window that's physically present on your screen but invisible to every recording tool on the planet. Not a software hack — this uses OS-level DRM hardware flags.
No title bars, no borders, no system chrome. The window feels like part of your wallpaper.
90% opacity glassmorphism layer that blends with any background using hardware-accelerated compositing.
Pinned at screen-saver z-level — above fullscreen apps, system dialogs, and everything else.
Invisible in the Windows taskbar and Alt+Tab menu. Zero trace in system navigation.
setContentProtection(true) — App content is invisible in screenshots, screen recordings, and screen-sharing sessions (Zoom, Teams, Discord, OBS). Uses Windows DXGI hardware-level DRM — not a software trick.
Window cannot be accidentally resized in stealth mode. No shadow for a truly invisible floating panel.
Prevents multiple copies of the app from running simultaneously. Only one phantom at a time.
Dual Application Modes
Default • Maximum Discretion
- › Hidden from taskbar
- › Always on top (screen-saver level)
- › Not resizable
- › Content protection ON
Configuration • Standard Window
- › Visible in taskbar
- › Standard controls (Minimize/Close)
- › Resizable
- › Content protection OFF
Switch instantly via Settings → App Settings → Visibility & Tracking.
AI Chat Engine
Not just a chatbot. A context-aware AI engine that knows your background, targets your job descriptions, and adapts its tone with 12 pre-built interview personas.
Full conversation history maintained in-memory and sent with every query for context-aware responses.
A custom 'Systematic Override' prompt is prepended to every LLM query automatically.
Software Engineer, Frontend, Backend, Full Stack, Mobile, DevOps, ML/AI, QA, System Architect, Engineering Manager, Analyst, PM.
Paste your resume — the AI cites your real background, projects, and experience in answers.
Paste a JD — the AI prioritizes that role's specific technologies and skills.
GitHub-Flavored Markdown with syntax-highlighted code blocks, language labels, and one-click copy buttons.
Copy the entire raw AI response to clipboard instantly with one shortcut.
For models that support it, a collapsible 'Logic Trace' shows the AI's internal chain-of-thought.
Personalized welcome: 'Good morning/afternoon/evening, {Name}' based on your local clock.
Multi-Provider AI Backend
Run AI completely locally with zero internet, or tap into the most powerful cloud models available. Your choice. Your keys. Your data.
HTTP REST • /api/chat • 127.0.0.1:11434
Zero-latency, 100% private inference. No API key needed. Models auto-detected on boot. Run Llama, Qwen, Mistral, or any model Ollama supports.
HTTPS REST • /api/v1/chat/completions
Access any model in the OpenRouter ecosystem. API key stored locally, never sent to Imposter servers. Supports free-tier models for zero-cost usage.
Model Management System
On startup, scans local Ollama instance via /api/tags and populates the model dropdown automatically.
Add any model from Ollama (custom URL) or OpenRouter with display label, model ID, API key, and base URL.
One-click delete for any custom model. Changes persist in localStorage under imposter_models.
Visual dot indicator — green = Models loaded / Ready, red = No models found or Ollama not running.
All custom models saved in localStorage. Survives app restarts. No cloud sync needed.
Same UI, same workflow — whether you're running a 7B local model or GPT-4 through OpenRouter.
Voice Transcription & Dynamic Island
Capture every word spoken on your system in real-time. The Dynamic Island — inspired by Apple's design — floats as a draggable pill overlay showing live transcription and AI responses.
Captures desktop audio (what your speakers output) via getDisplayMedia, processes through a custom AudioWorklet (100ms PCM buffer at 16kHz), and streams to AssemblyAI V3 via WebSocket.
Pulsing green mic icon + live transcription text. Partial results appear faded, final results in bold green.
Pulsing blue brain icon + AI response text in bold blue. Shows when AI processes a voice query.
One keystroke: grab last finalized transcript, auto-fill prompt, trigger AI query, display response in Island. End-to-end voice-to-answer.
Settings panel includes a "Test Connection" button to validate your AssemblyAI API key before use.
Screen Snip + OCR
See text on screen you can't copy? Snip it. Imposter extracts text from any window using local OCR — no data leaves your machine.
Press Ctrl+Shift+S — a transparent fullscreen overlay appears over your entire desktop.
Draw a rectangle over any area to capture it. Precise pixel-level selection.
Captured region processed by Tesseract.js locally. English language model bundled (~5MB).
Extracted text automatically appended to the AI prompt input. Ready to query with one Enter.
Press Escape to cancel the snip and dismiss the overlay instantly.
The snip overlay itself is also content-protected. Even the snipping tool is invisible to recordings.
Settings & Configuration
A split-pane settings modal with sidebar navigation and 7 configurable tabs. Every setting persists locally.
Display name for personalized greetings.
System prompt, persona selection (12 types), resume context, and job description targeting.
Add/remove custom models, configure providers, set API keys and base URLs.
Toggle between Stealth Mode and Normal Mode.
AssemblyAI API key configuration with live connection test.
Full read-only reference of all 14 global + 4 in-app keyboard shortcuts.
Keyboard Shortcuts
Every action at your fingertips. Global shortcuts work from anywhere on the system — even when Imposter isn't focused.
Global Shortcuts
| Shortcut | Action | Category |
|---|---|---|
| Ctrl+Shift+I | Focus Input — Jump to AI prompt | Navigation |
| Ctrl+Shift+Enter | Stealth Search — Trigger AI query from anywhere | AI |
| Ctrl+Shift+C | Deep Copy — Copy raw AI response to clipboard | Clipboard |
| Ctrl+Shift+Q | Emergency Exit — Quit instantly | System |
| Ctrl+Shift+D | Debug Override — Toggle DevTools | Debug |
| Ctrl+Shift+S | Screen Snip — Launch OCR tool | OCR |
| Ctrl+Shift+B | Toggle Voice Recording + Dynamic Island | Voice |
| F10 | Voice-to-AI — Grab transcript and ask AI | Voice + AI |
| Ctrl+Arrow Keys | Move Window — Nudge by 10px | Window |
| Ctrl+= / Ctrl+- | Resize — Grow/shrink by 50px | Window |
In-App Shortcuts
Security & Privacy
Privacy isn't a feature — it's the foundation. Every design choice ensures your data stays on your machine.
All config (name, system prompt, resume, JD, API keys) stored in browser localStorage. Nothing leaves the machine.
contextIsolation: true, nodeIntegration: false — renderer has zero access to Node.js APIs.
Strict CSP headers whitelist only self, specific CDNs, local Ollama, openrouter.ai, and assemblyai.com.
All Main ↔ Renderer communication goes through curated contextBridge API with hard-coded channels.
--disable-gpu-cache flag prevents GPU data persistence between sessions.
Prevents duplicate app processes. Only one phantom runs at a time.