DesignAdvisor Appendix D

This folder contains experimental configurations for Tracks A–C as described in Appendix D of the paper.

Structure

• prompts/: Role-specific prompt templates (JSON)
  └── e.g., compliance_officer.json

• logs/: Multi-agent interaction logs in .jsonl format
  └── e.g., trackA_session1.jsonl

• scripts/: Evaluation scripts for consensus rounds and JS divergence
  └── e.g., evaluation.py

DesignAdvisor Evaluation Scripts

This folder contains prompt templates, deliberation logs, and analysis scripts corresponding to Appendix D in our paper. Run scripts/evaluation.py to reproduce the JS divergence metric reported in Section 5.

OpenAI API Configuration

• Version: 2024-12
• Temperature: 0.3
• Max Tokens: 2048