1. Load

   ~/.claude/plugins/config/claude-for-legal/law-student/CLAUDE.md

   → current classes, weak subjects, outline locations.
2. Apply the framework below.
3. Route by flag:

   • --generate

     : build cards from source (outline path, notes, casebook) per card-writing rules. Write to

     ~/.claude/plugins/config/claude-for-legal/law-student/flashcards/[subject]/cards.md

     .

   • --drill

     (default): prioritize due cards + new; show Q, wait for answer, show A, take self-assessment, update buckets + next review.

   • --review

     : browse deck by bucket.

   • --stats

     : progress snapshot; flag stuck cards for verbal drill.

   • --session <n>

     : focused N-card session, prioritized by prior misses + due cards; appends results to

     study-plan.yaml

     →

     session_history

     .
4. Apply confidence discipline: flag every card generated from knowledge-without-source with

   [VERIFY]

   .

"This sounds like a real situation, not a hypothetical. I can't give you legal advice, and you can't give it either — you're not a lawyer yet. If this is real, [the person] needs an actual lawyer: legal aid, your school's clinic, a lawyer referral service (your jurisdiction's bar association, law society, or legal aid body), or (if there's money) a private attorney. I'm happy to help you understand the general legal concepts involved, but that's study, not advice."

• If generating cards from a source you provide (outline, notes, casebook excerpt), the card's Q and A come from that source. Confident.
• If generating cards from my knowledge without a source, I flag every card that states a rule I'm not fully confident on with

  [VERIFY: rule — confirm against source]

  . You should check before committing to the card as a learning target.
• If I don't know an area well, I generate fewer cards rather than inventing. Better to have 8 good cards than 20 where 5 are wrong.

• ~/.claude/plugins/config/claude-for-legal/law-student/CLAUDE.md

  → current classes, weak subjects, existing outlines

• ~/.claude/plugins/config/claude-for-legal/law-student/flashcards/[subject]/cards.md

  if it exists (incremental build)
• User-provided source (outline path, notes, casebook excerpt) if given

/law-student:session Contracts 5 --flashcards

• Load

  ~/.claude/plugins/config/claude-for-legal/law-student/study-plan.yaml

  if it exists and read

  session_history

  for this subject.
• Prioritize: cards previously marked wrong > due cards > new cards.
• Run N cards one at a time per the

  --drill

  flow.
• At session end, append results to

  study-plan.yaml

  →

  session_history

  :

session_history:
  - date: 2026-05-08
    subject: Contracts
    type: flashcards
    n_cards: 5
    right: 3
    partial: 1
    wrong: 1
    stuck_topics: [parol-evidence-rule]

• If no

  study-plan.yaml

  , write to

  ~/.claude/plugins/config/claude-for-legal/law-student/session-history.yaml

  instead.

• Subject (class name or topic)
• Source (outline path, notes, or "use my existing outline from ~/.claude/plugins/config/claude-for-legal/law-student/CLAUDE.md")
• Optional: card count target (default 10-20 per session)

undefined

1. Cards where

   next_review <= today

   AND bucket != mastered
2. New cards not yet attempted
3. If no cards due and no new cards: ask if user wants review of mastered cards (for decay prevention)

1. Show Q. Wait for answer.
2. User answers (or types "skip" / "don't know")
3. Show A.
4. User self-assesses:

   right

   /

   partial

   /

   wrong

   /

   don't know

5. Update bucket + next review per the table below:

• outline-builder: after building or extending an outline, offer to generate flashcards from the new material
• socratic-drill: if a card has been wrong 2+ times, route it to

  /law-student:socratic-drill

  for verbal working-through — flashcards aren't enough for concepts you don't actually understand
• bar-prep-questions: bar prep subjects with poor flashcard stats weight higher in MBE drilling

• Replace Anki. If you already have a flashcard habit, keep it. This is for when you're in chat and want to drill without switching apps.
• Invent cards to hit a count target. If I can only generate 8 confident cards from your source, you get 8. Padding with

  [VERIFY]

  -heavy guesses is worse than a smaller deck.
• Enforce study discipline. Missed review days compound; the skill just shows what's due. You decide whether to drill.
• Teach you the rule. Cards are for drilling what you've already studied. If a card is consistently wrong, the problem is upstream — use

  /law-student:socratic-drill

  or re-read the source.