Windows: Double-click start.bat
Mac/Linux: Run ./start.sh
Or manually:
cd modules/jane-ui
npm install
npm run dev
The onboarding wizard will ask for your GPU type:
- NVIDIA → downloads CUDA build of llama-server
- AMD → downloads Vulkan build
- Intel Arc/Iris → downloads Vulkan build
- CPU only → downloads AVX2 build
llama-server is downloaded automatically. No manual setup needed.
Models live in ~/.visrodeck/models/
To add a custom GGUF model:
- Open model selector → click the + card
- This opens the models folder
- Paste any
.gguffile there - It appears automatically in the selector
Press Alt+J to toggle the FLOAT overlay (bottom pill).
It expands into a mini chat with quick actions.
| Key | Action |
|---|---|
| Alt+J | Toggle FLOAT |
| Alt+Space | Toggle overlay chat |
| Ctrl+Shift+I | Dev tools |
jane-complete/
├── modules/
│ ├── jane-ui/ ← Electron app
│ │ ├── main.js ← Main process
│ │ ├── preload.js ← IPC bridge
│ │ └── renderer/ ← UI pages
│ └── jane/src/ ← AI agent
│ └── vre_client.js
├── vre/src/ ← VRE kernel (tools engine)
├── config/ ← License & VRE config
├── start.bat / start.sh
└── package.json
Place your logo file at: modules/jane-ui/renderer/splash-logo.png (or .ico converted to .png — any 256×256 or smaller image works) The splash screen will automatically use it. If missing, the default Visrodeck "H" mark shows.