First Steps

After installation, open http://localhost:9200.

1. Create Your Account

The first user to register becomes the admin. Choose a strong password — this protects your entire KULVEX instance.

2. Chat with Mnemo

Click the chat icon to start talking to Mnemo, your local AI. It runs entirely on your GPU — no data leaves your machine.

Try asking:

“What can you do?”
“Turn on the living room lights” (if home devices are configured)
“What’s the weather today?”
“Analyze this code: …” (paste a snippet)

3. Toggle Cloud Mode

In the chat interface, you’ll see a Cloud toggle. When enabled, your messages go to Claude API instead of Mnemo. This is useful for complex reasoning tasks.

You need an ANTHROPIC_API_KEY in Settings for this to work.

4. Configure Voice

KULVEX supports voice interaction with STT (speech-to-text) and TTS (text-to-speech).

STT priority:

mnemo:voice — GPU-accelerated Whisper (if a voice node is configured)
Deepgram — Cloud STT (needs API key)
Whisper CPU — Local fallback (slower)

TTS uses EdgeTTS by default (free, no API key needed).

5. Set Up Smart Home

Go to Settings > Home to configure your smart home devices:

Zigbee devices via zigbee2mqtt
WiFi devices (Tuya, Sonoff local API)
Matter/Thread devices

6. Install KCode

KCode is the AI coding assistant CLI. It’s pre-installed in the KULVEX terminal:

# From the web terminal (/code)
kcode
 
# Or use the aliases
kk    # Interactive REPL
k "explain this function"  # Single-shot query

See the KCode documentation for full details.

7. Mobile App

Download the KULVEX mobile app for iOS:

Scan the QR code in Settings, or
Download from the App Store (search “KULVEX”)

The app connects to your home server via Socket.IO with persistent sessions.

Configuration Hardware Guide