This paper explores the design space for teachable social media feed experiences, aiming to empower users with agency and personalized feed curation. Drawing on interactive machine teaching (IMT), a framework for non-expert algorithmic adaptation, the authors conducted a think-aloud study with 24 participants from four social media platforms—Instagram, Mastodon, TikTok, and Twitter—to understand how users evaluate and curate their feeds. The study identified key signals users use to determine the value of posts, which were synthesized into taxonomies. These taxonomies, combined with user interviews, informed five design principles for extending IMT into the social media setting. The authors then embodied these principles into three feed designs, serving as sensitizing concepts for future research in this area. The contributions of the paper include cross-platform taxonomies of prominent signals, five design principles, and three proposed feed designs, all aimed at enhancing user agency and personalized feed curation.This paper explores the design space for teachable social media feed experiences, aiming to empower users with agency and personalized feed curation. Drawing on interactive machine teaching (IMT), a framework for non-expert algorithmic adaptation, the authors conducted a think-aloud study with 24 participants from four social media platforms—Instagram, Mastodon, TikTok, and Twitter—to understand how users evaluate and curate their feeds. The study identified key signals users use to determine the value of posts, which were synthesized into taxonomies. These taxonomies, combined with user interviews, informed five design principles for extending IMT into the social media setting. The authors then embodied these principles into three feed designs, serving as sensitizing concepts for future research in this area. The contributions of the paper include cross-platform taxonomies of prominent signals, five design principles, and three proposed feed designs, all aimed at enhancing user agency and personalized feed curation.