Refine Reasoning for Hallucination Reduction

Analyze the `reasoning_trace` generated by Llama 3.3 during initial simulation runs. Identify instances where the agent's reasoning seems to deviate from the explicit supply chain model or produces illogical actions (hallucinations). Refine your LangGraph state transitions, tool definitions, and Llama 3.3 prompts (e.g., by adding more constraints, few-shot examples, or explicit validation steps) to mitigate these issues and improve reasoning coherence and accuracy.