Empowering K-12 teachers through GenAI-supported vibe coding for educational software development

Teachers increasingly design digital resources to meet diverse student needs, yet educational software development has historically remained a high-barrier domain. Generative-AI-supported vibe coding holds potential to enhance teachers’ curriculum design work by lowering such barriers, but little is known about teachers’ experiences with vibe coding and how human-GenAI interaction unfolds during this process. This mixed-method study analyzed vibe coding prompt logs and development notes from 12 K-12 Korean teachers over a three-week period to explore what they created, how they navigated the development process, and what opportunities and challenges they experienced. Results showed that vibe coding enabled teachers to create deployable tools primarily for instructional and administrative purposes. Prompt-intent sequence analysis revealed that development was primarily driven by content refinement and error handling, with limited engagement in conceptual understanding and evaluation—indicating shifting configurations of human control and AI automation across the design process. The findings suggest that vibe coding can support teacher-as-designer work by enabling teacher-driven educational software creation, while also introducing new constraints, such as a technical knowledge ceiling, and potentially stratified benefits associated with prior experience and digital competencies.