/exam-forecast

1. Load ~/.claude/plugins/config/claude-for-legal/law-student/CLAUDE.md → class, professor, exam format, syllabus.
2. Apply the workflow below.
3. Intake past exams (PDF, paste, or paths). Confirm sample size.
4. Analyze each past exam: format, subject coverage, question style, fact-pattern density, recurring traps.
5. Cross-exam pattern analysis — what's stable, what varies.
6. Combine with current syllabus to produce forecast: subject weights, format, hobby horses, study emphasis.
7. Write ~/.claude/plugins/config/claude-for-legal/law-student/exam-forecasts/[class]/forecast-[YYYY-MM-DD].md. Framed as weighting heuristic, not prediction.

Purpose

Every professor's exam has fingerprints. The same hypo structures recur. The same traps come back. The same subject ratios repeat. Students who have prior exams study smarter; students who don't, study harder. This skill analyzes the prior exams you have and surfaces the patterns.

Not magic. A forecast, not a prediction. The skill cannot tell you what's on the exam — it can tell you what's been on past exams and what's likely to recur based on syllabus coverage.

Confidence discipline

• Pattern analysis (what subjects appeared, how many questions per topic, how often policy vs. rule-application) — confident where the exams are clearly in front of me.
• Inference about likely emphasis on upcoming exam — [UNCERTAIN] is the default; these are forecasts, not certainties. Explicitly frame as "based on the [N] past exams you shared, [topic] appeared in [M]. Your upcoming exam may emphasize it, or the professor may rotate — use this as a weighting for review time, not a prediction."
• If only 1-2 past exams are available, say so explicitly — any pattern inferred from 1 exam is noise.
• If the professor is new (no past exams available), skill can't forecast. Say so; fall back to syllabus-based "these are the subjects covered" only.

Load context

• ~/.claude/plugins/config/claude-for-legal/law-student/CLAUDE.md → current classes, exam formats, syllabus if captured
• User-provided past exams (PDF, pasted text, paths)
• Optional: syllabus for the current class (for "what's been covered to date")

If the uploaded past exams have a professor's name, use it to match patterns (same-professor exams are the highest-signal input). If not, match on subject and structure. Don't ask the user to type in the professor's name — use what's in the materials. If the user volunteers it in conversation that's fine; don't prompt for it.

Workflow

Step 1: Intake

• Which class are we forecasting for?
• How many past exams from this professor are available?
• Are they from the same course, or different courses by the same professor?
• Are any of them the take-home / open-book / different-format variants, vs. the typical format for your upcoming exam?
• Syllabus for your current class?

If fewer than 3 past exams: flag as thin sample. Pattern inference is weaker. If exams are across different courses: some patterns transfer (question style, policy vs. doctrine ratio); subject-specific patterns don't.

Step 2: Read each past exam

For each past exam:

• Format (number of questions, length, time limit, open/closed book)
• Subject coverage (which topics tested, in what proportion)
• Question style (issue-spotter, single-issue deep, policy essay, short-answer MBE-style, mix)
• Fact pattern density (fact-heavy hypos, sparse facts with doctrinal focus, or policy prompts with no facts)
• Recurring traps (e.g., professor always hides the jurisdictional issue in an otherwise-clean fact pattern; professor always asks about the exception rather than the rule)
• Policy vs. doctrine ratio
• Unusual structures (essays + MBE hybrid, moot court scenario, etc.)

Step 3: Cross-exam pattern analysis

Roll up what's consistent across exams:

Stable patterns (appeared in most/all past exams):

…