Clozed sentences to optimize reviews

I’m doing Fast Track Polish/Russian. It’s somewhat close to Polish, so I’m doing voice input.
In Fast Track - one word has approximately 1 sentence for it. For a completely unfamiliar word, I feel like it provides too little context overall. I noticed that I can’t retrieve that word in real life. For familiar words it’s ok, though.
Also, I wouldn’t want to go through hundreds Common Words, marking sentences/words as known manually.

In Clozemaster what are we learning? Words, Sentences, Words in context?
What if Clozemaster would show us “clozed” word but with a random sentence each time from array of 5 sentences. Would it help to learn words more efficient? Or would it be too hard?

I’ve made tampermonkey script to make sentence fully clozed.
Doing voice typing sometimes I can translate clozed sentence partially/fully. Would be nice if those other clozed words got +1 day added to review date as a reward, since I’ve retrieved the word from my memory.
Maybe it would help with the problem of a lot of reviews piling up.

My proof of concept for English/Russian. When I pronounce words correctly, the review date pushed +1 day (not actually implemented). When I move mouse over clozed word the word is just revealed.

TLDR: In Fast Track there is too few sentences for unfamiliar words and too many reviews for familiar words.