New version of the Anki Addon Dataset (on 2026-04-01) was released:
New version of the Anki Addon Dataset (on 2026-05-01) was released:
New version of the Anki Addon Dataset (on 2026-06-01) was released:
- Addons number: 2943
- Quickly download report in Excel format
- Visualizations (charts/diagrams)
- Hugging Face dataset
- GitHub poject
Screenshots
Report in Excel format (download):
Visualizations (link):
New version of the Anki Addon Dataset (on 2026-07-01) was released:
- Addons number: 3071
- Quickly download report in Excel format
- Visualizations (charts/diagrams)
- Hugging Face dataset
- GitHub poject
Screenshots
Report in Excel format (download):
Visualizations (link):
New version of the Anki Addon Dataset (on 2026-08-01) was released:
- Addons number: 3207
- Quickly download report in Excel format
- Visualizations (charts/diagrams)
- Hugging Face dataset
- GitHub project
Screenshots
Report in Excel format (download):
Visualizations (link):
Click the button “Restart this Space” if the space is sleeping.
According to info from around 2021 there were about 750 add-ons in total, currently there are 3,000+ add-ons so the number has increased about 4 times over the past five years. I think it’s become more difficult to find useful or essential features for learning that aren’t already available in the native Anki or add-ons.
IMO there are concerns that the increase in add-ons may lead to more security risks, but recently no such dangerous add-ons have been discovered or reported on AnkiForums, I think this is probably because the new upload restrictions implemented by the official Anki are working effectively.
I’d argue this surge is also a consequence of the AI boom. Three of the five top-rated add-ons are vibe-coded, and if you sort by “Recently Updated,” you’ll have a hard time finding an add-on that hasn’t been touched by AI.
It might be worthwhile to provide new developers with a template project that has a sensible structure and perhaps even includes a CLAUDE.md (or similar) file. That would raise the baseline quality of new add-on releases. Right now, the official documentation is essentially just, “Create an __init__.py file and see what happens.”
I guess I’m suggesting a more centralized promotion of e.g. GitHub - abdnh/anki-addon-template: Anki add-on template for my projects · GitHub in the docs with an extra section to guide the LLMs
For now one of the add-on development communities that has been promoting AI is led by the developers of Ankimon. Ankimon is relatively active in recruiting contributors and it seems to assist with coding and learning using AI and provides support via Discord. The author of the popular add-on Onigiri is an Ankimon contributor, so I think they probably started developing from there.
Ankimon Github: I don’t know how to CODE! Where do I start?
It’s an interesting theory 
I added a new column to the Anki Addons Dataset which lists AI tools used in an addon. It uses explicit signs like presence of CLAUDE.md, etc. files.
So, only 54 addons use AI tools (explicitly).
The chart (link, click button Restart this Space if sleeping):
New column in XLSX (download):
I implemented this quickly (with AI 
). Later I’ll revise the approach, maybe numbers will change.
When developers use AI and want to let users know, the common approach is to explain it on their add-on page. e.g. one add-on developer who frequently uses AI is Eros Cardoso, they have released 120+ add-ons so far and explain on their add-on page that they used AI.
I think there are very few developers these days who don’t use AI tools at all. According to a 2025 Stack Overflow survey about 50% to 80% of developers use AI, so those who don’t use AI are already in the minority.
One more reason why more developers are using AI tools to develop add-ons is that students can often use paid AI tools for free. Many Anki users are students so they may be able to use advanced AI tools more easily.
New version of the Anki Addon Dataset (on 2026-09-01) was released:
- Addons number: 3379
- Quickly download report in Excel format
- Visualizations (charts/diagrams)
Click the button “Restart this Space” if the space is sleeping. - Hugging Face dataset
- GitHub project
Screenshots
Report in Excel format (download):
Visualizations (link):
Click the button “Restart this Space” if the space is sleeping.
In XLSX:
Versions: 26.08.1~
Site ankiweb.net: Supported Anki versions: 25.09.2-26.08.1+ (Updated 2026-08-17)
Is this how it should be?
What is the algorithm for obtaining the “Language” field?
Is this determined by the description?
It would be nice to have another field for a short description. Perhaps we could somehow parse the very beginning of the description, or perhaps from the “About” section, like on GitHub.
Yes, versions (“branches”) are extracted from Supported Anki versions: 25.09.2-26.08.1+ (Updated 2026-08-17).
In general, addons can have several branches with really confusing naming.
In your example, it should be just 25.09.2+, shouldn’t it?
Actually, I don’t know how to represent versions in a meaningful manner.
All parsed branches are present in the Parquet file (column anki_web.branches). XLSX file contains just the latest of them.
BTW I guess your link should point to a specific addon? Can you check it?
The “Language” is detected by the lingua-language-detector Python library from the description.
The full description is available in the Parquet file (column anki_web.addon_page_content).
I didn’t add it to the XLSX because it’s too long.
The structures of addon descriptions and its GitHub repos are arbitrary. So, it’s not easy to prepare “a short description”. In the future versions, I’m thinking to apply AI to do it.