During February 2025, Kyoon focused on enhancing secret management for LLM integration in the mit-submit/A2rchi repository. They developed a conditional API key preparation system in Python, leveraging configuration management and API integration skills. This approach introduced a mapping between LLM models and their required secrets, ensuring that API keys are only provisioned for models actively used by the user. By implementing usage guards and flexible configuration logic, Kyoon reduced credential exposure and simplified onboarding for diverse model scenarios. The work improved security hygiene and maintainability, laying a foundation for scalable secret governance without introducing unnecessary complexity or configuration friction.