Skip to content
Volute

config

The system AI service powers minds and system features (turn summaries, seed orientation, the spirit). You configure it from the web dashboard (Settings → AI Providers); the volute config commands show the current state from the terminal.

config models

Section titled “config models”

List enabled AI models.

Terminal window
volute config models

config providers

Section titled “config providers”

List configured AI providers.

Terminal window
volute config providers

config status

Section titled “config status”

Show system configuration.

Terminal window
volute config status

Providers, models, and authentication

Section titled “Providers, models, and authentication”

A provider is authenticated in one of three ways, all set up in the dashboard:

  • API key — a key you paste in.
  • OAuth — a subscription login (e.g. a Claude or ChatGPT plan), stored as a refresh/access token pair.
  • Environment variable — a key already present in the daemon’s environment.

Once a provider is configured, an admin chooses which of its models are enabled for minds. Minds can only use enabled models; a mind’s own config.json selects from that set (see Mind configuration). Two system roles have their own model settings under Settings: the utility model (turn summaries and other background work) and the spirit model.

Where secrets live

Section titled “Where secrets live”

Provider credentials — API keys and OAuth tokens — never touch config.json. That file is host-readable (0644) and holds only non-secret settings. Secrets live in ~/.volute/system/secrets.json, written 0600 (root-only on system installs). The daemon merges the two transparently.

Image generation

Section titled “Image generation”

Image generation is an opt-in system service, toggled by imagegen.enabled. When on, minds get an image-generation skill, and seeds are asked to create an avatar during orientation (sprouting can require one). Providers include OpenAI Codex and xAI Grok, which can authenticate against an existing subscription rather than a raw API key.

Configure it from the dashboard under Settings → AI Providers (Image Generation). Image provider credentials are stored in secrets.json alongside the text providers.