Creator AI Settings
These are world-level settings that creators configure in the editor. For player-controlled settings (model selection, temperature, context, BYOK), see the Player Guide: AI Settings.
playerName
| Type | Default |
|---|---|
| string | "User" |
The player's display name in the prompt. All {{user}} macros are replaced with this value. Set it to match your world's fiction -- "Commander" for a military sim, "Traveler" for an adventure, etc.
Players can also set their own name via personas, which overrides this default.
Lorebook settings
These control how the lorebook keyword scanner works. Configurable in the editor under Lorebook > Entry Settings.
| Parameter | Type | Default | Description |
|---|---|---|---|
lorebookScanDepth | integer | 2 | How many recent messages to scan for keyword matches. Default 2 = only the last 2 messages. Increase for deeper context triggering, but higher values add matching overhead. |
lorebookRecursionDepth | integer | 0 | How many levels deep to re-scan triggered entries for more keywords. 0 = single pass (no recursion). Set to 2 and triggered entries' content gets scanned again, up to 2 layers deep. Good for interconnected lore, but watch token budget with deep recursion. |
structuredOutput
| Type | Default |
|---|---|
| boolean | false |
When true, forces the AI to reply in JSON format via response_format: { type: "json_object" }. For mechanic-heavy worlds that need to strictly parse AI output. With this on, AI replies are JSON objects, not natural language.
Advanced
structuredOutput is not exposed in the editor UI. To enable it, export the world JSON, add "structuredOutput": true to the settings object, and re-import. Most worlds don't need this.
See also
- AI Directives & Macros — how the AI produces output that the engine processes
- Writing Great Entries — the content the AI reads, which these settings shape
- Player Guide: AI Settings — model selection, temperature, context, BYOK
All WorldSettings fields → World Spec: Schema