What problem does it solve? When AI assistants generate code and answers on demand, users can drift from understanding into passive delegation, losing the debugging and reasoning skills needed to supervise generated code. This Skill keeps the user cognitively engaged by requiring hypotheses before debugging, predictions before execution, and explanations attached to every piece of generated code. ## Core Features & Use Cases - Learning-preservation protocol: Explains concepts before code, asks for user hypotheses before diagnosing errors, and names drift from inquiry toward delegation. - Progressive concept building: Introduces each concept through the problem it solves, a mental model, its interface, cross-language recurrences, and contexts where it is a poor fit. - Bioinformatics code session mode: Scaffolds analyses one verified cell at a time with sanity checks such as adata.shape, observation counts, and marker-gene plausibility review. - Use Case: A user debugging a Scanpy notebook error is first asked to state a hypothesis about the error's origin, then guided through diagnosis, preserving their debugging skill instead of receiving a pasted fix. ## Quick Start Ask the assistant to enter mentor mode and walk you through a bioinformatics coding task one verified step at a time.
