Exam Timetable Architect

Overview

Turns "I have 6 exams in 5 weeks and 40 topics" into a concrete day-by-day plan. Topics are ranked by exam weight × self-rated weakness × recency of first study, then scheduled backwards from each exam date so every topic gets an initial study block plus spaced reviews at expanding intervals (1–3–7–14 days). Days are load-capped by available hours; the tool tells you the truth about infeasible plans ("you need 4.2h/day but declared 2h/day") instead of silently producing fantasy schedules.

When to Use

• "I have exams in N weeks, help me plan" / "make me a study timetable"
• Multiple exams with dates and a syllabus/topic list per exam
• One big exam with many topics (bar exam, boards, finals, driving theory)
• Re-planning after a missed day ("I lost 3 days to flu")
• Deciding what to drop when there isn't time for everything

Don't use for: single-topic tutoring plans, habit building (habit-trigger- detective), or timeblocking whole weeks beyond exam study.

The Engine

1. Input: exams (name, date), topics per exam (name, difficulty 1–5, self-rated weakness 1–5, estimated hours), daily available hours.
2. Priority = weight(normalized exam proximity) × weakness × difficulty — near exams and weak heavy topics float to the top.
3. Spacing: each topic's first study is followed by reviews at +1, +3, +7, +14 days (review cost = 25% of first-study hours). Reviews must land before the exam; clips to exam-eve if needed.
4. Load capping: no day exceeds declared available hours (study blocks are 50 min + 10 min break units; max 3 units before a long break is forced).
5. Feasibility check: total required hours vs total available hours before the last exam. If over capacity, the tool ranks what to cut (lowest priority first) and rebuilds — triage, not silence.
6. Output: plan (per-day schedule), today (what to do now), done (check off, triggers re-plan of the remainder), progress (coverage %, hours burned vs remaining).

Numbered Workflow

1. init with exam dates and daily hours.
2. topic add each syllabus item with difficulty/weakness/hours (or bulk from a JSON/CSV dump).
3. plan → review the day map; if infeasible, follow the triage suggestions (topic drop <id> or raise daily hours).
4. Daily: today → study → done <block-id>; missed blocks auto-reschedule.
5. Weekly: progress — coverage and pace vs plan.
6. Exam eve: the plan converges to light review-only days by design.

Commands

python3 scripts/exam_planner.py init --hours 3.5 --start 2026-10-01
python3 scripts/exam_planner.py exam add "Calculus II" --date 2026-10-20 --weight 40
python3 scripts/exam_planner.py topic add "CalcII: integration by parts" \
  --difficulty 4 --weakness 4 --hours 3
python3 scripts/exam_planner.py plan
python3 scripts/exam_planner.py today
python3 scripts/exam_planner.py done c1-r2
python3 scripts/exam_planner.py progress
python3 scripts/exam_planner.py replan   # after missed days / new topics

Store: exam_plan.json next to the script (--store to override).

Spacing rationale

The +1/+3/+7/+14 ladder approximates expanding-interval retrieval practice (Cepeda et al. 2006 meta-analysis: spaced beats massed by ~10–30% on retention at test). Reviews are cheap (25% of original hours) because material was seen before. Reviews never schedule after their exam; they compress into exam-eve if the plan is tight — late compression is still better than none.

Pitfalls

1. Rating everything weakness 5 — the ranking collapses. Force honesty: weakness is relative to you, 3 = average.
2. Planning 8h/day — sustainable for almost nobody. The tool caps units/day and warns above 6h; believe it.
3. Studying in date order instead of priority order — the classic fatal mistake (chapter 1 gets mastered, chapter 14 gets nothing). Follow plan's ordering, not the textbook's.
4. Skipping reviews to cover new material — coverage feels good, retention collapses. Reviews are locked unless the day is genuinely overloaded >150%.
5. Never re-planning — real life invalidates plans weekly. Run replan after any 2+ missed days; it re-spaces from today, not from the past.

Example

5 exams in 4 weeks, 40 topics, 3.5h/day: feasibility check reports needed 4.1h/day → tool suggests dropping the two lowest-priority topics (weakness 2, weight-light exam) → plan fits at 3.4h/day. Each hard topic appears ~4 times (study + 3 reviews) before its exam; exam week itself is review-heavy by construction.