Also, in search for SRS that works, and so I have deviated from the ways of Anki. I think it can be made to include Japanese, but it’s unlikely for me to go that far.
Reading SRS only. (Optional meaning quiz with LLM API key, non-SRS.)
Multiple answer input (; separated), with options to require multiple important readings
No real synonyms. Warning shake at best. All answers must exist in the dictionary.
Typing and auto mark as right/wrong. No changing from wrong-to-right or vice versa.
Mark as repeat with Esc (midway between right and wrong).
Undo (via Ctrl + Z) is sent to the end of review, rather than immediate retyping.
Skip current vocab with F1 if you don’t want to do similar vocabs immediately. It will come back in the next batch anyway.
Wrap up every 20 items, but can be done earlier. (F4, and I sometimes wrap-up at 10 items. No need for otherwise decision paralysis.)
Restudy (drill) wrong items after wrap up.
New vocab (10) after some reviews (40), but before backlog is cleared. New vocab can be turned off in Settings.
Search for inclusive vocabularies, and put studied items first. (Personal preference for hiding reading only if due in 1 hour. Tomorrow or later reviews don’t worth hiding reading.)
Hanzi break-down and build-up. Also put studied items first.
Vocab and Hanzi searching can include reading alongside to narrow the search
No separating vocab and Hanzi. (Not really an issue for Chinese, but could apply to Japanese.)
External links and Notes with rich text and images. I encourage taking notes, even if only readings are quizzed.
Levels, of course. Sorted by language proficiency test levels (HSK), then by most common Hanzi.
You can start from a mid level. It’s just a vocabulary list. Go down level if you feel like it.
Currently, it’s a desktop app, and it’s kinda unlikely that I would make it into a website or include Japanese (as it doesn’t have much benefit to myself). Anyway, contributions welcome.
Download the zip file, unzip, and open the exe file.
I updated the project, for the text analyzer to exclude Hanzi in learning (i.e. fresh new Hanzi), and also to make names more distinguishable (ateji in Chinese is fierce).
Also, Add button – so that (almost) all Hanzi will be learned soon enough.
AI translation can be used if you have an API key, to cover for monolingual definitions, colloquial usages, and grammar explanation. DeepSeek is relatively cheap and cost effective.
Personal plan: After adding all preset vocabs (to Level 60) and most Hanzi from reading, I might be adding vocabs with 4 or more Hanzi next. (in order to be careful not to overload the SRS.)
Be familiar with sounds of 四字熟語 or 成語. Those are really everywhere in Chinese, probably even more common than in Japanese.
This app is weak on meaning quiz, but I tried to answer every words in SRS at least once (including lone Hanzi and more multi-facet ones).
A different option of not LLM… I could have used a diff engine to EN definitions – to mark as either correct or ungraded (but that would be a new feature)
Major difference from my previous Chinese SRS project, is the right/wrong criteria is clear, and another big difference is that I managed to finish it. (Preset vocab lists, in any case.)
About the details, what it actually teaches, is indeed Hanzi reading. Another thing is vocab mining process to be put to SRS. Meaning aren’t taught that well, but I consider LLM for meaning comparison helpful. Maybe a better different way is indeed SRS, but Anki mode + reset card for misunderstood meanings.
No reset, no resurrect, vocabs in this project, but Hanzi reading can be reinforced by adding more new items. The point is don’t fear to have tried to learn as many new lessons.
In a way, I respectfully disagree with going completely slow. I don’t disagree with steadiness. Even if lifelong learning, milestones still have to be reached, only that such attempt to reach may far deviate from original plan. Long running means there will be unexpected events, good and bad.
or ungraded 
(but that would be a new feature)