Settings
Configure local service URLs and default model preferences.
The app can still open, save settings, manage tasks, and use local SQLite data. Start the missing host services below when you need model chat, custom voice, or memU sync.
http://host.docker.internal:11434
fetch failed
Local model chat and model downloads are unavailable. The rest of the workspace remains usable, and OpenAI can be used if configured.
- Start Ollama on the host machine.
- Pull the configured model, for example: ollama pull llama3.2.
- In Docker, keep the Ollama Base URL set to http://host.docker.internal:11434.
http://memu:18000
fetch failed
Remote memU sync is unavailable. Local SQLite memories still work in the Memory page and assistant context fallback.
- Start the managed memU-compatible service with Docker Compose.
- In Docker, keep the memU Service URL set to http://memu:18000.
http://host.docker.internal:17493
No Voicebox API responded at http://host.docker.internal:17493. health endpoint: fetch failed; profiles endpoint: fetch failed
Custom voice cloning and Voicebox-backed profiles are unavailable. Whisper.cpp transcription and default system/local TTS can still work.
- Start Voicebox on the host, or run docker compose --profile voicebox up -d voicebox.
Autonomous learning is enabled.
http://127.0.0.1:18100
fetch failed
Learned code skills cannot run, but chat and non-code skills remain available.
- Start the skill-sandbox Compose service.
OPENAI_API_KEY is not set
OpenAI-backed agents are unavailable, but local workspace features still work.
- Set OPENAI_API_KEY before starting the container if you want cloud-backed models.