[Context and motivation] The rapid growth of social media has transformed how users, especially young people, engage with digital platforms and applications, raising concerns about their addictive potential. [Question/problem] However, there remains a limited understanding of what design features of Human-Computer Interaction in social media platforms most strongly encourage addictive behaviors and which user interface design elements are used to support engagement on these platforms. [Principal ideas/results] To close this gap, this study aims to develop a comprehensive design features taxonomy of Human-Computer Interaction within social media that influences addictive user behaviors and increases the time and frequency of usage. In developing our taxonomy, we drew from existing literature, incorporating insights from experienced Human-Computer Interaction, psychology, and software engineering practitioners to validate our taxonomy. By identifying specific design features that encourage prolonged engagement, this taxonomy provides a foundation for guiding design decisions that must be made to develop non-addictive social media. [Contribution] This taxonomy is a valuable tool for designing, regulating, and using social media more responsibly, balancing user engagement, user experience, and digital well-being while addressing the risks of compulsive behaviors and technology addiction.