World Model Playground

This is a personal exploration platform for world-model-inspired agent systems. It focuses on how agents can maintain, update, and reason with internal simulations of their environment.

Key Features

• Simulation-based Planning: Agents can run multiple hypothetical scenarios before taking an action.
• Dynamic Memory: A hierarchical memory system that stores and recalls relevant world-states.
• Visual Feedback: Real-time rendering of the agent's internal state.

Tech Stack

• Next.js 15
• TypeScript
• Three.js
• Supabase