Agent Skills
› mohitagw15856/pm-claude-skills
› exam-prep-planner
exam-prep-planner
GitHub生成基于提取练习与间隔重复的真实备考计划。通过加权薄弱点、交错学科及预留缓冲时间，提供每日具体执行方案与紧急调整策略，拒绝无效复习。
Trigger Scenarios
制定考试复习计划
询问多科目备考方法
需要具体的学习时间表
Install
npx skills add mohitagw15856/pm-claude-skills --skill exam-prep-planner -g -y
SKILL.md
Frontmatter
{
"name": "exam-prep-planner",
"description": "Build a realistic exam-prep schedule with spaced repetition and retrieval practice — the plan that survives contact with an actual week. Use when asked to plan my exam prep, make a study schedule, I have N weeks until finals, or how do I study for multiple exams. Produces a day-by-day plan across all exams: spaced blocks, retrieval-first sessions, weak-topic weighting, and built-in slack for the days that go wrong."
}
Exam Prep Planner Skill
Most study plans are written for the student the planner wishes existed: six perfect hours daily, no bad days, rereading counted as studying. This skill plans for the student who exists — retrieval practice over rereading, spacing over cramming, weak topics weighted, and slack built in, because a plan that dies at its first missed day was never a plan.
What This Skill Produces
- The day-by-day schedule — all exams interleaved, sessions with named topics and methods
- Retrieval-first sessions — every block says what to do (blurt, past paper, teach-back), never "review chapter 4"
- Weighting — weak and high-value topics get the prime slots and the most spaced repeats
- Slack and triage — catch-up buffers, plus the pre-agreed what-to-drop list if time collapses
Required Inputs
Ask for these if not provided:
- Exams, dates, and formats — all of them; single-exam plans that ignore the others aren't plans
- Topic lists with self-rated confidence (red/yellow/green per topic — honest ratings)
- Real available hours — after work, sport, life; the actual number, not the aspirational one
- What "studying" has meant so far — rereaders need the method change named explicitly
Framework
- Retrieval beats rereading — every session is active: closed-book blurts, past papers, self-testing, teach-it-aloud. Rereading appears only as a 10-minute pre-retrieval warmup.
- Spacing beats massing: each topic returns 3+ times with growing gaps (day 1 → day 3 → day 7 → pre-exam). The schedule interleaves subjects — blocked days feel productive and test worse.
- Weight by (weakness × exam value): red topics on high-weight exams own the morning slots; green topics get maintenance touches only.
- The 80% rule: schedule at most 80% of stated available hours — the other 20% is the buffer that saves the plan when Tuesday explodes.
- Pre-agreed triage: decide NOW what gets dropped if two days vanish (green topics first, then low-weight material) — triage decided in a panic always cuts the wrong thing.
Output Format
Exam Prep Plan: [N] weeks · [exams]
Real hours/week: [stated] → scheduled: 80% = [n]h · Method note: [the change from current habits]
The schedule
| Day | Slot | Exam/topic | Method (what you'll actually do) | Repeat # |
|---|
Weekly retrieval checkpoints
[End of each week: one mixed self-test across everything touched; scores drive next week's re-weighting]
If it goes wrong
Missed 1 day → [absorb via buffer] · Missed 3+ → [the triage list, in drop order]
Quality Checks
- Every session names a retrieval method — zero "review/reread" blocks
- Every red topic appears ≥3 times with growing spacing
- Total scheduled ≤ 80% of stated hours
- All exams share the calendar, weighted by date and value
- The triage list exists before it's needed
Anti-Patterns
- Do not schedule rereading as studying — recognition feels like knowledge and tests like its absence
- Do not plan 100% of available time — a plan without slack is a countdown to abandonment
- Do not block whole days per subject — interleaving tests better than it feels
- Do not spend prime hours on green topics because they're pleasant — comfort studying is procrastination with flashcards
- Do not build guilt into the plan — a missed day triggers the buffer protocol, not a spiral; the plan's job is to survive
Version History
- 54fad50 Current 2026-07-19 13:12
