study-quizlisted

# Quizzing & Grading **Output language: ALL learner-facing content MUST be in Simplified Chinese** — questions, explanations, grading feedback, reports. Two jobs: **exam-style analysis** and **quizzing & grading**. Shared iron rule: never answer questions for the learner in conversation, never leak answers before they attempt. ## Bank first: check the question bank BEFORE writing questions The workspace keeps a reusable question bank (`question-bank.json`, managed by `bank.py`). Consulting it costs a few lines of context; regenerating questions from chapter text costs thousands of tokens. Flow: ```bash python3 ~/.claude/skills/study-quiz/scripts/bank.py <study-dir> list --point 3.1.2 3.1.3 # cheap index python3 ~/.claude/skills/study-quiz/scripts/bank.py <study-dir> get Q0003 Q0007 # full entries python3 ~/.claude/skills/study-quiz/scripts/bank.py <study-dir> use Q0003 Q0007 # after assembling ``` - Prefer unused / least-used entries; for repeat practice **adapt** bank entries (new numbers/scenario, same point) instead of reusing verbatim. - Write brand-new questions ONLY for points the bank doesn't cover. Source them from the section's lecture JSON in `lessons/` when it exists (already-distilled content, far smaller than chapter text); fall back to `textbook/` only when there is no lecture. - After building any new quiz, ingest it: `bank.py <study-dir> add <quiz.json>` — the bank grows as a side effect of normal use. ## Delivery format: interac