Supplementary Material: A Multi-Agent LLM Framework for Design Space Exploration in Autonomous Driving Systems

This repository serves as the Supplementary Material (Online Appendix) for the paper titled "A Multi-Agent LLM Framework for Design Space Exploration in Autonomous Driving Systems", accepted to the 41st ACM/SIGAPP Symposium on Applied Computing (SAC '26).

Paper Links

• Official Conference Version (ACM Library): https://doi.org/10.1145/3748522.3779714
• Extended Version (arXiv): arXiv:2512.08476

---

Repository Structure

This repository provides the complete prompts and example outputs that were omitted from the main manuscript due to space constraints.

.
├── examples/                                       # Input/Output Examples in our case study
│   ├── dse_agent_output_example.txt                # Example of DSE agent's reasoning output
│   ├── performance_deciphering_output.txt          # Example of performance deciphering agent analysis output
│   └── user_input_interpretation_agent.txt         # Example of user instruction
│
├── prompts/                                        # Full System Prompts and Instructions for LLM Agents in our case study
│   ├── dynamic_instruction_post_simulation.txt     # The example dynamic instruction after a simulation of our case study
│   ├── dynamic_instruction_pre_simulation.txt      # The example of dynamic instruction before a simulation, offering a few randomly-chosen design points to refer.
│   ├── system_prompt_dse_agent.txt                 # System prompt for the DSE Agent
│   ├── system_prompt_text_llm.txt                  # System prompt for text LLM in performance deciphering agent
│   └── system_prompt_visual_llm.txt                # System prompt for visual LLM in performance deciphering agent
│
└── README.md