What problem does it solve? Passive reading and direct answers rarely build durable understanding. This Skill turns a topic, text passage, webpage, screenshot, or PDF into an interactive Socratic learning dialogue where the learner reasons through exactly one question per turn, with adaptive follow-ups, gentle misconception repair, and a cumulative blackboard that builds from concrete examples to formal symbols. ## Core Features & Use Cases - One-question-per-turn tutoring: Each active turn asks a single concise learner-facing question, adapts to correct, partial, mistaken, or stalled answers, and exits immediately when the learner asks to stop or wants a direct explanation. - 12-keyword orientation map: For broad topics, it first builds a map of exactly 12 plain-language keywords plus 3-5 expert and book entrances, then lets the learner pick a starting point; expansion happens only on explicit request. - Progressive blackboard and visual scaffolding: Abstract, spatial, geometric, and calculus topics advance through cumulative visual frames with text alternatives, using authorized built-in image generation only when it adds real structure. - Self-contained learning webpages: On explicit request, it packages a lesson into a single offline HTML file with inline CSS/JS, no CDN, no telemetry, and no default deployment. - Use Case: A learner pastes a PDF chapter on opportunity cost and asks to be taught one question at a time; the Skill anchors questions to page numbers, distinguishes source statements from inference, and tracks mastery only after the learner explains concepts in their own words and transfers them to new contexts. ## Quick Start Ask your agent to teach you a topic or document Socratically, one question at a time, for example: use qiaomu-learning to teach me opportunity cost from this PDF, asking only one question per turn.
