What problem does it solve? AI agents that fix code for you erode the engineering skills you used to exercise yourself. This Skill counters skill atrophy by turning already-implemented work into a comprehension check backed by cited evidence, so you can verify you actually understand code that you or an agent wrote. ## Core Features & Use Cases - Evidence-based drift analysis: Compares what was actually built (with file:line citations) against the cited ideal solution, covering complexity, edge cases, security, and test coverage. - Socratic questioning loop: Asks Bloom's-laddered questions that lead you to find gaps yourself, never stating the solution or writing corrected code. - Transfer assignments and grading: Assigns near-transfer and far-transfer tasks, then grades your explanation using feed up / feed back / feed forward structure. - Spaced repetition ledger: Logs concepts with re-check dates in .ai/teacher/log.md so later sessions re-quiz you as interleaved retrieval practice. - Use Case: After an agent implements a caching layer, run a comprehension check to confirm you can explain the eviction policy, derive its complexity, and generalize the principle to a different domain. ## Quick Start Ask the teacher skill to quiz you on the diff from your last commit and verify whether you really understood what was built.
