不分析的模考是浪费。下面 15 个 Prompt 把考后散乱变成结构化诊断：哪些错在概念、哪些粗心、哪些被时间逼，再给未来两周的针对性训练。
适合哪些场景
高压考试备考者（SAT / GRE / GMAT / MCAT / LSAT / 司考 / 医考）、AP 和 IB 学生、大学期末复习者，以及做个性化复习计划的辅导老师。
什么时候不建议这样写 Prompt
至少一次完整模考都没做的人不用。考完超过 3 周的也别用——记忆和情绪都褪色了，诊断质量下降。
Prompt 结构公式
弱点诊断 Prompt 一定要带这六个要素：
- 角色：AI 扮演谁——研究导师、同行评议人、考试教练、辩论对手、图书馆员。
- 上下文：水平、学科、deadline、论文数量、引用风格、课程或项目。
- 目标：一个具体交付物——12 道题、1 页文献矩阵、5 条反论、4 周复习计划。
- 限制：字数、深度、允许的来源类型、跳过什么、绝不主张什么。
- 输出格式：编号清单、表格、JSON 或分级块（E / M / H），能粘到 Notion / Anki / Word。
- 示例 / 信号：1-2 段参考或反例（“不要像维基那样讲”）。
这套 Prompt 适合用在哪
- 完整模考复盘
- 考前定向复习
- 辅导师生诊断面谈
- 科目分项弱点
- 考前 2 周倒计时
15 个可直接复制的 Prompt 模板
1. 逐题错误分类
默认首轮，每道错题归类。
You are an exam coach. Below are the questions I missed on my {test name} practice exam. For each: classify the error as (a) conceptual gap, (b) misread / careless, (c) time pressure, (d) test-taking strategy, (e) genuine knowledge gap. Recommend 1 specific drill per category.
{paste questions + my wrong answers + correct answers}可替换变量： test name, questions, answers
优化建议： 分类模糊时追加：“Be ruthless. If a question could be classified as both careless and conceptual, default to conceptual until proven otherwise.”
2. 概念簇分析
Below is my error log. Group my misses by topic / concept cluster. For each cluster: number of misses, the underlying concept, 1-line root cause, and 3 practice problems I should redo before drilling new material.
{paste log}3. 时间管理分流
Below is my time-per-question log for the exam. Identify: (a) questions I spent too long on, (b) questions I rushed, (c) the score impact of better pacing, (d) one pacing rule for next time.
{paste timing}4. 分项热力图
For my recent {test name} score breakdown: {paste section scores}, identify the 2 sections with the highest leverage for improvement and the 1 section to maintain. Suggest hours-per-week per section for the next 4 weeks.5. 反复错检测
Below are my last 3 practice tests with error logs. Identify the top 5 mistakes I keep making, the underlying root cause for each, and whether they are getting better or worse over time.
{paste 3 logs}6. 信心 vs 正确率审计
For each missed question below, I will provide my pre-answer confidence (1-5). Identify questions where I was high-confidence but wrong (overconfidence zone — most dangerous) and questions where I was low-confidence and right (could practice trusting instinct).
{paste questions + confidence + outcome}7. 策略 bug 抓取
Below are the questions I missed. Identify strategy bugs: (a) starting with the hardest question, (b) reading every answer choice instead of predicting, (c) over-engineering a simple question, (d) not using elimination. Quantify how many missed questions each bug touched.
{paste}8. 主题 x 难度矩阵
For my recent exam, build a 4x3 matrix: rows = top topics, columns = easy / medium / hard. In each cell, write my accuracy %. Identify the 3 cells where extra prep would yield the highest point gains.
{paste data}9. 2 周复习计划
Based on my diagnosis ({paste}), build a 2-week revision plan. Each day: 1 focus topic, 1 timed mini-section, 1 review activity. Friday: full timed section. Sunday: weekly review. Constraints: max {hours/day}.10. 定向题目生成
I keep missing questions on {topic}. Generate 8 practice questions in {test format} on this topic, increasing in difficulty. Include the answer key and 1-line explanation per answer. Mark which 2 should be drilled multiple times.11. 模考复盘单
Below is my full mock exam result. Produce a 1-page debrief: overall score, section breakdowns, top 5 missed topics, top 3 strategy bugs, top 2 strengths to lean on, and 3 concrete commitments for the next 7 days.
{paste mock results}12. 诊断面谈脚本
I am about to walk through 5 missed questions with my tutor. For each, generate 3 probing questions she could ask me to surface my reasoning (e.g., "what did you think the question was asking?"). Goal: distinguish careless from conceptual.
{paste questions}13. 焦虑 / 配速干预
My recent exam pattern: strong in section 1, fading in section 3. Likely causes: fatigue, anxiety, time pressure. Suggest 4 interventions (between-section reset routine, hydration, micro-pacing checks, practice under fatigue). Each intervention as 1-line action.14. 分数预测模型
Based on my last 3 practice tests ({paste scores}), project my likely score range on test day. Identify the score I need for {goal school / cert} and the 3 highest-leverage actions to close the gap.15. 终周作战计划
The exam is in 7 days. Based on my diagnosis: {paste}. Build a final-week plan that includes: 2 full-length mocks (which days), light review, sleep / nutrition / logistics, what NOT to do in the last 48 hours. End with 3 morning-of-test rituals.容易踩的坑
- 只复盘错题——答对题里的”瞎蒙”也要核。
- 把错都归”粗心”——大多数”粗心”其实是概念。
- 不记时——配速是标化考试一半分。
- 反复错没补就刷新题。
- 只做新模考不分析旧的。
- 只盯最弱主题，不看它在考试里的频次。
- 考前一周猛冲新内容而不是练高性价比的弱项。
优化技巧
- 每次模考后先做模板 1（分类）再干别的。
- 答题前记信心（模板 6）——过度自信是最危险的 bug。
- 建一份”反复错”档案——分数就藏在那里。
- 概念训练配带计时的小节（模板 9），训练压力下的提取。
- 考前 24 小时只做轻复习和模板 15 的事务清单。
- 辅导课带上模板 12 的提问，时间花得更值。
- 罕见主题不要追求完美——高频主题更重要。
FAQ
- 多久做一次完整模考？: 备考期每 2 周一次，考前最后一月每周一次。再频繁分析跟不上。
- 反复犯同样错怎么办？: 很少是粗心——底层概念或策略坏了。用模板 5 重建地基。
- AI 能批作文题吗？: 总体结构和评分量表反馈可以；正式考试还需对照官方量表和人工分。
- 分数预测有多准？: 看趋势可用，准确预测不行——真实分数还看休息、压力、当日难度。
- AI 能替代辅导老师吗？: 诊断和训练常常可以；动力、问责、个性化策略，好老师仍占优。