I Copied and Pasted Anki Cards by Hand for 8 Years, So I Let Claude Code Do It

This post contains a referral link.

In this part of my Anki series, I automated adding the cards themselves. I used to add cards to Anki by hand, copying and pasting, but with AI as advanced as it is now, I figured it was about time to semi-automate it.

Related: cleaning up duplicate Anki cards with Claude Code

For 8 years, there’s one thing I’ve kept doing. When I come across a word I don’t know in an article or an English conversation, I first jot it down on a sticky note, in a notebook, or, being a devoted Apple fan and iPhone user, in a notes app. Then I look up the meaning in English-Japanese and English-English dictionaries such as Cambridge Dictionary, copy and paste it, and add a card to Anki.

I always thought how great it would be if this part of the process could be automated, something like “linking it to a dictionary.”

In this post, I’m writing up the process of building that skill, just as it happened.

Sponsored links

My workflow before automation (8 years of copying and pasting)

Here’s what I’ve been doing whenever I came across a new word in an article or a conversation:

  • Save the word in a notes app for now
  • Look up the meaning, part of speech and example sentences in dictionaries (Cambridge, English-Japanese)
  • Copy and paste the definition, example sentences and pronunciation by hand
  • Open Anki’s Add screen and paste everything in
  • Prepare a TTS audio file separately and attach it

5–10 minutes per word. Little by little, it adds up, and I kept it up for 8 years.

What I built with Claude Code

The goal this time was a skill that adds cards to Anki just by pasting a list of words. With the AnkiConnect API, you can control Anki from outside, so I had Claude Code build it as a skill.

① Checking my decks and note types

First, I checked my current deck structure and note types in Anki. I had Claude list them with deckNames and modelNames. Eight years of decks came up, and I felt nostalgic and a little conflicted. lol

② Designing the format (the Back field)

Next, I decided on the format for the Back field of my cards (where the meaning, example sentences and related words go). While showing Claude the cards I actually use now, we settled on this structure:

  • ① Definition (English)
  • ② Japanese translation
  • ③ Pronunciation (IPA) + TTS audio file
  • ④ Synonyms (A / B)
  • ⑤ Antonyms (⇔ C / D)
  • ⑥ Derived words (→ word 1, word 2)
  • ⑦ Example sentences (carefully chosen from Cambridge and other sources)

The key to the format is not copying copyrighted dictionary data as is. Following Cambridge Dictionary’s license terms, definitions are paraphrased or summarized. For example sentences, the rule is to quote from public corpora (such as COCA and BNC) or fill in with sentences generated by AI.

③ Pronunciation and TTS audio

The IPA comes from the Cambridge API. For the audio, Google TTS (gTTS) generates an mp3, storeMediaFile saves it to Anki’s media folder, and it’s embedded in the Back field as [sound:word.mp3].

So I can press “R” in Anki to check the pronunciation as I learn, just like before.

④ Duplicate check

It searches the existing cards in the target deck with findNotes and skips any card whose Front (the word) matches. So it’s safe even if the same word is already somewhere in my 8 years of decks.

⑤ How batch adding works

When I paste a list of words, I can check a preview of each word and choose whether to add it or skip it. At the end, everything is added in one go with addNotes.

⑥ The preview step

Instead of adding everything automatically, I added a step where it asks me “Is this OK?” one word at a time before adding it. The idea is not to blindly trust data generated by AI, and to check it with my own eyes at the end.

In fact, this step saved me once.

⑦–⑩ (fine-tuning)

From here on are the details we adjusted as we went back and forth.

  • ⑦ Prepositions: for words that almost always take a set preposition, like agree with and rely on, the preposition goes at the top of the Back field
  • ⑧ Synonyms: written as (A / B), or (A / B / C) if there are three or more; antonyms are always written as (⇔ D)
  • ⑨ Multi-word phrases: phrases like get the hang of were added without any problems
  • ⑩ A real case where the check step stopped a wrong card (the “dot” incident)

A little more about ⑩.

The word was actually dot used as a verb (to be scattered here and there), as in “cafés dot Tokyo.”

On top of that, the word to be added had become “. Tokyo,” so I had the Front fixed to dot and the definition replaced with the verb meaning.

I was really glad I had built in the check. And if a few days had passed, I might not have noticed even while checking, said “OK~~~ add it~ ♪( ´θ`)ノ”

and ended up memorizing it wrong. lol

⑪ Managing cards with an AI tag

Finally, I casually asked, “I’d like to add some kind of (AI) mark so I can search and bulk-edit the cards Claude made later. Would a tag work?”

Rather than writing something into the Back field, an Anki tag is more practical because you can search and bulk-edit with tag:AI, so we went with an AI tag. Claude also went back and tagged the 30 cards I had already added (3 in the first run + 27 in a batch), and the rule was written into the skill and my manual.

The AnkiConnect API actions I used

actionWhat it’s for
deckNames / modelNamesList decks and note types
findNotes / notesInfoCheck existing cards in the target deck and find duplicates
storeMediaFileSave audio files to Anki’s media folder
addNotesAdd cards in bulk while skipping duplicates automatically
addTagsTag notes I had already added

The result: where it stands now

A new skill called tango-seiri (roughly, “sorting vocabulary”) is done, and I’ve actually added 30 words (3 in the first run + 27 in a batch) to my “English notes” deck, which I’ve been using for 8 years.

The format (when to use ① and ②, synonyms (A / B), antonyms (⇔ C / D), the derived-words line, where the audio goes, and the AI tag) was all worked out little by little as we talked. The skill is written on the assumption that the format isn’t fixed: it’s something we keep looking at and growing together.

Thoughts and what’s next

Work I did by copying and pasting one word at a time for 8 years now takes nothing more than pasting the words and saying OK to the preview. I think that’s a quiet but huge change.

Since the pronunciation and TTS audio come as a set, I can still press “R” in Anki and check the pronunciation as I learn. I’m also happy with where we landed: not copying copyrighted dictionaries as is, but keeping a generous amount of meanings, synonyms and derived words within a practical range.

I’ll probably keep fine-tuning the format every time I paste in new words. I’d like to write about it again, including how it grows.

If you’ve also been building up your word cards bit by bit by copying and pasting from dictionaries by hand, I think you’ll find something to relate to here.

📊 I also used the same Claude Code and AnkiConnect setup to analyze 9 years of my Anki stats.

I Had AI Analyze My Anki Stats: My Most-Forgotten Words After 9 Years and 950,000 Reviews

🃏 I’ve also put all my Anki posts together on one page, sorted by what you need: getting started, settings, keeping it up, automation, audio and troubleshooting.

NO IMAGE

Referral link: if you sign up through it, I may receive free lesson minutes.

🃏 Words you add are worth using in conversation

Once adding words gets easier, it’s time to use what you’ve learned. I put words I come across in articles and conversations into Anki and try using them in my Cambly lessons. Through my referral link, you get 10 extra minutes when you buy a trial. I’ve been taking lessons on Cambly since July 2021.

Try Cambly with my referral link →

The referral program isn’t available in every region.

English expressions I learned (B2–C1)

I’m aiming for Eiken Grade 1, so I had AI pick out some useful phrases from this topic.

① streamline (to make a process simpler and more efficient)

We streamlined the process of adding new vocabulary cards.

This one describes exactly what I did with this skill.

② redundant (duplicated; more than is needed)

The system automatically skips redundant entries.

You can use it to describe the duplicate check (findNotes) in this post.

③ tedious (boring and tiring because it goes on too long)

Copying definitions by hand became tedious after a few years.

If I had to sum up how I felt after 8 years of copying and pasting in one word, this would be it.

Reference: Cambridge Dictionary

Related posts

3 Anki Settings to Change Before You Start: Next Day Starts At, New Cards/Day and Maximum Reviews/Day

Related: my daily Anki routine / how to make Anki cards and decks / starting Eiken Grade 1 vocabulary in Anki / flagged Anki cards with n8n and Claude Code

I’d also like to write about rewriting my flagged Anki cards someday.

☕ Enjoyed this post?

I'm a Japanese blogger who spent 8 years relearning English from scratch.
If this post helped you, a coffee would mean a lot — thank you for reading!

Leave a Reply

Your email address will not be published. Required fields are marked *

ABOUT US
Suitman
An office worker in Japan who has worked in three fields: education, manufacturing and sales. I started learning English at 35 and, after 8 years of keeping at it, passed Eiken Grade Pre-1. VERSANT score: 59. I write about work tips, Excel, AI, English and cameras, all based on my own experience.