Woman: Mode: 'party'Additional figure(s): Mode: Fast Chat | Model: xAI/grok-4.3 | Streamed primary with deterministic postcheck
Status:
Active chat: not set
Chat setup is ready to initialize.
Click or tap to display setup loading progress, or navigate to Mini App.
Open Mini App

LLM Settings

Embedding Status

Response Length

Load Figure into Chat

Pick a stored figure and apply its description/updates as the active chat context.

Figures

No figures.

Prompt Variance Strategy

None (default)

Select one variance approach per chat. Changes apply on the next reply.

Active figure personalities are now interpreted through this strategy, so persona capsules and variance behavior run in parallel.

Visible planning notes
Prepends a short user-facing notes block before the reply. Hidden reasoning, policy, and internal checks stay hidden.
Behavior Runtime
Enable or disable the dynamic behavior runtime and associated phase/move response biases.
Strategy family
Toggle between legacy and new agnostic variance sets.

Desired Spoken Phrases

Visual Directives

Behavioral modifications

Standard behavior modifications and saved algorithmic governor presets can run together or independently for the current chat.

Standard behavior layer

Use or ignore saved manual behavior modifications in prompt assembly.

Manual off

No manual behavior modification text is currently loaded.

Algorithmic preset layer

Apply the selected figure's saved runtime-governor baseline from the Behavior page.

No preset

No saved algorithmic preset is available for the currently loaded primary figure.

Personality Profiles (Axiomatic)

Select a figure-specific personality profile to inject into chat responses. Profiles now prioritize confidence-weighted visual assumptions with dynamic response rules.

Select a primary figure to load personality profiles.

Talk To Figure

Keep ordinary chat input addressed directly to one figure until this mode is turned off.

Narrator

Direct, present, and naturally responsive.

Perspective lock

selected figure is the only first-person speaker. The user is addressed as "you".

No other active figure is currently available for third-person reference.

This preserves scene context, additional figures, feature awareness, captions, behavior modifiers, visual directives, and selected personality profiles.

Namespace
No prompts found in this namespace. Adjust and refresh.

Dynamic System Prompt

Current Figure

Current Scenario Descriptions

Narrative Input Options

Insert from figures

Chats

Persona Summaries

Conversation