Anki Skill

Workflow

1. Verify connectivity first

python3 <SKILL_DIR>/scripts/manage_decks.py --list

If this fails, tell the user to open Anki and check AnkiConnect is installed.

2. Check note types

Before creating cards, discover what note types and fields the user has:

python3 <SKILL_DIR>/scripts/get_note_types.py

The default model names are Basic, Basic (and reversed card), and Cloze, but the user may have custom ones. Adapt field names accordingly.

3. Creating Cards

Read references/card-quality.md before creating any cards — it has rules for writing atomic, high-quality medical flashcards.

Always present proposed cards to the user for approval before creating them.

Basic card:

python3 <SKILL_DIR>/scripts/create_cards.py \
  --deck "Medical::Cardiology" \
  --type basic \
  --front "What is the first-line treatment for stable angina?" \
  --back "Beta-blockers (e.g., metoprolol) — Class I, LOE A (ESC 2023)" \
  --tags "cardiology,angina,treatment,ESC-2023"

Basic + Reversed:

python3 <SKILL_DIR>/scripts/create_cards.py \
  --deck "Medical::Cardiology" \
  --type basic-reversed \
  --front "Troponin I" \
  --back "Cardiac biomarker; rises 3-6h post-MI, peaks 12-24h, normalizes 7-10d" \
  --tags "cardiology,biomarkers,MI"

Cloze:

python3 <SKILL_DIR>/scripts/create_cards.py \
  --deck "Medical::Cardiology" \
  --type cloze \
  --text "CHADS₂-VASc: {{c1::CHF}}, {{c2::HTN}}, {{c3::Age ≥75 (2pts)}}, {{c4::DM}}, {{c5::Stroke/TIA (2pts)}}, {{c6::Vascular dz}}, {{c7::Age 65-74}}, {{c8::Sex (female)}}" \
  --tags "cardiology,afib,scoring"

Batch creation from JSON (preferred for multiple cards):

python3 <SKILL_DIR>/scripts/create_cards.py --from-json cards.json

JSON format:

{
  "cards": [
    {
      "deck": "Medical::Cardiology",
      "type": "basic",
      "front": "Question",
      "back": "Answer",
      "tags": ["tag1", "tag2"]
    },
    {
      "deck": "Medical::Cardiology",
      "type": "cloze",
      "text": "The {{c1::left ventricle}} pumps to the {{c2::systemic circulation}}",
      "tags": ["anatomy"]
    }
  ]
}

4. Searching Cards

# By deck
python3 <SKILL_DIR>/scripts/search_cards.py -q "deck:Medical::Cardiology"

# By tag
python3 <SKILL_DIR>/scripts/search_cards.py -q "tag:angina"

# Full-text
python3 <SKILL_DIR>/scripts/search_cards.py -q "troponin"

# Combined
python3 <SKILL_DIR>/scripts/search_cards.py -q "deck:Medical tag:treatment metoprolol"

# Raw JSON output
python3 <SKILL_DIR>/scripts/search_cards.py -q "tag:ESC-2023" --json

5. Editing Cards

# Update a field
python3 <SKILL_DIR>/scripts/edit_card.py --note-id 1234567890 --field Front --value "Updated question"

# Add tags
python3 <SKILL_DIR>/scripts/edit_card.py --note-id 1234567890 --add-tags "reviewed,2024"

# Remove tags
python3 <SKILL_DIR>/scripts/edit_card.py --note-id 1234567890 --remove-tags "draft"

6. Deleting Cards

python3 <SKILL_DIR>/scripts/delete_cards.py --note-ids 1234567890 1234567891

Always confirm with the user before deleting. The script also prompts interactively.

7. Deck Management

# List all decks
python3 <SKILL_DIR>/scripts/manage_decks.py --list

# Create (use :: for hierarchy)
python3 <SKILL_DIR>/scripts/manage_decks.py --create "Medical::Endocrinology::Diabetes"

# Delete empty deck
python3 <SKILL_DIR>/scripts/manage_decks.py --delete "OldDeck"

8. Moving Cards Between Decks

# By card IDs
python3 <SKILL_DIR>/scripts/move_cards.py --card-ids 123 456 --to-deck "Medical::Cardiology"

# By search query (moves ALL matching cards)
python3 <SKILL_DIR>/scripts/move_cards.py --query "tag:renal" --to-deck "Medical::Nephrology"

Creating Cards from Medical Guidelines

When the user provides a guideline (PDF, pasted text, or URL content):

1. Read references/card-quality.md for flashcard quality rules
2. Extract key clinical facts, recommendations, and decision points
3. Organize by topic using :: deck hierarchy (e.g., Medical::Cardiology::HeartFailure)
4. Choose the right card type per concept:
   • Basic — factual recall ("What is X?", "First-line for Y?")
   • Basic+Reversed — bidirectional pairs (drug ↔ mechanism, disease ↔ finding)
   • Cloze — lists, criteria, scores, stepwise procedures
5. Tag every card with: guideline source, year, clinical topic
6. Present ALL proposed cards for user approval
7. After approval, write a JSON file and batch-create with --from-json

Deck Naming Convention

Suggest this hierarchy (adapt to existing decks):

Medical::
  ├── Cardiology::
  │   ├── HeartFailure
  │   ├── ACS
  │   └── Arrhythmias
  ├── Pulmonology::
  │   ├── Asthma
  │   └── COPD
  └── Endocrinology::
      ├── Diabetes
      └── Thyroid

Always run --list first to match the user's existing structure.

Environment Variables