fabricatio-anki

MIT

AI-powered Anki flashcard deck generation and compilation for the Fabricatio framework. Uses LLM-driven templates to produce card layouts, model definitions, and topic analyses, then compiles them into .apkg files compatible with Anki 2.1+.

Installation

pip install fabricatio[anki]
# or
uv pip install fabricatio[anki]

For the full Fabricatio suite:

pip install fabricatio[full]

Overview

The package splits into two layers:

• Rust (fabricatio_anki.rust) — fast filesystem operations for project scaffolding, template serialization, and deck compilation via deck_loader.
• Python (fabricatio_anki) — LLM-driven generation of deck metadata, card models, HTML/JS/CSS templates, and topic analyses, integrated with Fabricatio's Propose capability system.

Project Structure

A deck project created by create_deck_project follows this layout:

deck_project/
├── deck.yaml           # Deck metadata (name, description, author)
├── models/             # One subdirectory per note type
│   └── basic_card/
│       ├── fields.yaml # Field definitions (e.g. Front, Back)
│       └── templates/  # One subdirectory per card template
│           └── card/
│               ├── front.html
│               ├── back.html
│               └── style.css
├── data/               # CSV files with card content
│   └── basic_card.csv
└── media/              # Global images, audio, etc.

Rust Functions (fabricatio_anki.rust)

| Function | Description | :|---|---| | compile_deck(path, output) | Compile a deck project into an .apkg file. | | create_deck_project(path, deck_name?, description?, author?, model_name?, fields?) | Scaffold a new deck project with sample templates and data. | | save_metadata(dir_path, name, data) | Write a Python dict as YAML into a project directory. | | add_csv_data(project_path, model_name, data_path) | Copy a CSV file into the project's data/ directory. | | save_template(dir_path, front, back, css?) | Write front.html, back.html, and optional style.css for a card template. | | extract_html_component(html) | Parse an HTML string into (layout, js, css) by separating <script> and <style> content. | | extract_content_by_tag(html, tag) | Extract inner text from all occurrences of a given HTML tag. |

Python Models

Python Capabilities

GenerateDeck

LLM-driven deck generation. Extends Propose. Key methods:

Accepts a single requirement string or a list for batch generation. All methods pass through Fabricatio's validation system.

GenerateAnalysis

Generates structured TopicAnalysis objects from topic strings.

AppendTopicAnalysis

An Action that reads a CSV file, runs generate_analysis on each row, and appends a "Topic Analysis" column.

from fabricatio_anki.actions.topic_analysis import AppendTopicAnalysis

action = AppendTopicAnalysis(
    csv_file="questions.csv",
    output_file="questions_with_analysis.csv",
)
result = await action.execute()

Configuration

All options below are read through the fabricatio configuration chain (see the Configuration Guide). Set them under the [ext.anki] table in fabricatio.toml, equivalently under [tool.fabricatio.ext.anki] in pyproject.toml, or via FABRICATIO_EXT__ANKI__<FIELD_UPPER> environment variables.

[ext.anki]
generate_anki_card_front_side_template = "built-in/generate_anki_card_front_side"

Access at runtime: from fabricatio_anki.config import anki_config.

Usage Example

from fabricatio_anki.rust import create_deck_project, compile_deck, add_csv_data
from pathlib import Path

# 1. Scaffold a project
create_deck_project(
    Path("./my_deck"),
    deck_name="French Vocabulary",
    description="Essential French words for beginners",
    author="Language Team",
    model_name="french_vocab",
    fields=["French", "English", "Pronunciation", "Example"],
)

# 2. Add your CSV data
add_csv_data(Path("./my_deck"), "french_vocab", Path("./words.csv"))

# 3. Compile to .apkg
compile_deck(Path("./my_deck"), Path("./french_vocab.apkg"))

For LLM-driven generation via GenerateDeck:

from fabricatio_anki.capabilities.generate_deck import GenerateDeck

gen = GenerateDeck()
deck = await gen.generate_deck(
    "Create a deck for learning Japanese JLPT N5 vocabulary",
    fields=["Kanji", "Reading", "Meaning", "Example Sentence"],
)
# deck is a Deck with auto-generated models and templates

Dependencies

• fabricatio-core — core interfaces and utilities
• fabricatio-capabilities — capability framework (Propose)
• deck_loader (Rust) — Anki deck serialization and project management

License

MIT — see LICENSE