Configuration

Config Files

File	Location (Windows)	Purpose
config.json	%APPDATA%/aether-desktop/config.json	Application settings
auth.json	%APPDATA%/aether-desktop/auth.json	Authentication state

On macOS/Linux, replace %APPDATA% with ~/.config.

Settings Panel

Open Settings from the dashboard gear icon. Six sections are available:

General

• Auto-start containers — Start IDE and Terminal when the app launches
• Auto-start on boot — Launch Aether Desktop when your system starts
• Workspace path — Default project directory

Runtime

• Container backend — Select between nerdctl, Docker Desktop, k0s, or WSL
• WSL distro — Configure the WSL distribution (Windows only)

The app auto-detects available runtimes on first run. You can override the selection here.

AI & Models

• Hardware accelerator — Shows detected GPU/NPU capabilities
• Embedding models — Download and manage models for GrepAI
• LLM models — Download and manage models for local inference
• OVMS control — Restart the model server to load new models

Model storage location: %LOCALAPPDATA%/aether-desktop/models/

Network

• Traefik proxy — Enable/disable *.aether.localhost routing
• Tailscale overlay — Connect to a Tailscale/Headscale network for remote access

Security

• Device certificate — Register a TPM-backed certificate
• Auth tokens — View and manage authentication tokens

Storage

• Model storage path — Change where AI models are stored
• Workspace path — Change the default workspace directory

CLI Options

# Launch with debug logging
aether-desktop --debug

# Launch a specific project
aether-desktop launch --project ~/my-project

# Mount additional volumes
aether-desktop launch \
  --project ~/my-project \
  --volume ~/.ssh:/home/coder/.ssh:ro \
  --volume ~/.gitconfig:/home/coder/.gitconfig:ro

Environment Variables

Services inside the containers can access these variables:

Variable	Description
OVMS_ENDPOINT	OVMS REST endpoint (set by Aperture)
OVMS_GRPC_ENDPOINT	OVMS gRPC endpoint
WORKSPACE_DIR	Mounted workspace path

First-Run Wizard

On first launch, a 5-step wizard guides you through:

1. Welcome — Introduction
2. Runtime detection — Auto-detect Docker/nerdctl/k0s
3. Workspace selection — Choose your project directory
4. Image pull — Download container images (~1-2 GB)
5. Complete — Ready to code

The wizard can be re-triggered by deleting config.json.