language-learning

Anki import files for practical language learning.

Structure

• spanish/ - Spanish decks and import files
• future languages get their own directories at the repo root

Importing into Anki

Most files are simple CSV imports.

Typical mapping:

• Front/Back vary by deck, and many decks include both directions in the same CSV.

Start small, keep decks practical, and optimize for real-world communication.

Current Spanish starter decks

• spanish/nouns.csv
• spanish/verbs.csv
• spanish/adjectives.csv
• spanish/useful_phrases.csv
• spanish/nouns_images.csv - experimental image -> Spanish noun deck

Root seed lists

The root nouns.txt, verbs.txt, adjectives.txt, and phrases.txt files are the reusable base packs for future languages.

When adding new vocabulary or phrases to a language-specific deck, also add the English source item to the matching root .txt seed list so the shared library keeps growing.

Anki audio workflow

This repo now includes a reusable pronunciation workflow for Anki across languages:

• audio files live in <language>/audio/
• deck CSVs store inline Anki sound tags like [sound:querer.mp3]
• the generation script is scripts/generate_anki_audio.py
• preset defaults live in anki-audio-presets.json
• the repo-local skill is skills/anki-audio/SKILL.md

Spanish verbs example:

uv run --with edge-tts python scripts/generate_anki_audio.py --preset spanish-verbs

Other current presets:

• spanish-nouns
• spanish-adjectives
• spanish-useful-phrases

To regenerate from scratch:

uv run --with edge-tts python scripts/generate_anki_audio.py --preset spanish-verbs --force

Local Anki sync workflow

You can sync the repo CSVs and media into Anki Desktop without manually dragging files into collection.media or re-importing each CSV.

Requirements:

1. Open Anki Desktop on the computer that owns the collection.
2. Install the AnkiConnect add-on (2055492159).
3. Run a dry-run first:

uv run python scripts/sync_to_anki.py --language spanish --dry-run

Then run the real sync:

uv run python scripts/sync_to_anki.py --language spanish

Current Spanish deck mapping:

• spanish/adjectives.csv → Spanish Adjectives
• spanish/nouns_images.csv → Spanish Image Nouns
• spanish/nouns.csv → Spanish Nouns
• spanish/useful_phrases.csv → Spanish Phrases
• spanish/verbs.csv → Spanish Verbs

By default, media is uploaded through AnkiConnect. If you prefer direct file copy, pass your local media directory:

uv run python scripts/sync_to_anki.py \
  --language spanish \
  --copy-media-dir "/Users/rasul/Library/Application Support/Anki2/User 1/collection.media"

The script creates missing decks, uploads/copies referenced [sound:...] and <img src="..."> files, adds new notes, updates existing notes matched by their Front field inside each deck, and triggers Anki sync unless --no-sync is passed.

It also cleans up two old manual-import artifacts when run against AnkiConnect:

• deletes accidental literal header notes where the note content is Front / Back
• normalizes stale notes whose Front had an old trailing parenthetical, like to want (querer), when the current CSV now uses to want

Anki image workflow

This repo also includes a reusable image-based deck workflow for any language:

• image files live in <language>/images/
• image-based deck CSVs reference bare filenames with HTML like <img src="el-cafe.jpg">
• the helper script is scripts/generate_anki_image_deck.py
• preset defaults live in anki-image-presets.json
• the repo-local skill is skills/anki-images/SKILL.md

Important: the script only scaffolds the deck CSV plus a prompt manifest. The actual image creation step happens separately via an image-generation tool/model or manual sourcing.

Preferred generation path: use OpenAI image generation through OpenClaw's OpenAI Codex OAuth subscription auth, for example image_generate with openai/gpt-image-2. Do not require or document a direct OPENAI_API_KEY for this workflow; the intended auth path is the same Codex OAuth subscription used by the agents/chats.

Spanish nouns example:

uv run python scripts/generate_anki_image_deck.py --preset spanish-nouns-images --overwrite-output

The script writes:

• an image-based deck CSV such as spanish/nouns_images.csv
• an editable prompt manifest in the image directory, such as spanish/images/nouns_image_prompts.csv

Use the prompt manifest as a starting point, then review and tighten prompts for ambiguous concepts before generating images at scale. Prompt manifests live next to images because they are generation metadata, not import decks.