Summary

Creature mood, animations, and app background on the phone should be controlled by the server, which gets this information from Claude. Currently the creature visuals are static - instead, Claude should influence how the creature looks, behaves, and what environment it's in based on conversation context.

Proposed Flow

1. Claude - During responses, Claude outputs mood/emotion metadata and environment info (e.g. happy, thinking, confused, excited + forest, space, ocean)
2. Server - Parses mood/environment from Claude's output and sends it to connected clients via WebSocket (new message types, e.g. creature_mood, creature_background)
3. Phone app - Receives updates and dynamically adjusts CreatureView animations, colors, particle effects, expressions, and the app background

Creature Mood Examples

• User asks a complex question -> Claude is "thinking" -> creature shows contemplative animation
• Claude successfully completes a task -> "happy" mood -> creature bounces, glows brighter
• Claude encounters an error -> "confused" mood -> creature shows puzzled expression
• Idle state -> "calm" mood -> gentle breathing animation

Background Examples

• Claude is working on code -> dark terminal-like background
• Casual conversation -> warm gradient background
• Claude is debugging -> red-tinted alert background
• Night time / idle -> starry sky background

Open Questions

• How should Claude communicate mood and background? (structured output field, tool call, inferred from response tone)
• What set of moods/states and backgrounds should be supported?
• Should mood and background transitions be smooth or instant?
• Should backgrounds be predefined themes or fully dynamic (colors, gradients, particle systems)?