🪐 Agent2World: Learning to Generate Symbolic World Models via Adaptive Multi-Agent Feedback - View it on GitHub