Supporting learners’ understanding of taught skills in online settings is a longstanding challenge. We hypothesize that an AI agent’s ability to explain learners’ skill-based questions can be significantly enhanced using a structured knowledge-based AI framework called Task-Method-Knowledge (TMK). We introduce Ivy, an AI coach that combines a TMK model with an LLM to generate explanations that embody teleological, causal, and compositional reasoning. Our evaluation shows that Ivy’s responses go beyond typical shallow answers from agents using unstructured text alone, by improving the depth and relevance of feedback. This can potentially support learners in developing a more comprehensive understanding of procedural skills crucial for effective problem-solving in online environments.