chinese-sentence-miner VS hskhsk.com

I made a tool that somewhat solves your problem a while back, specifically for Chinese: https://github.com/Destaq/chinese-sentence-miner 18k sentences taken from Tatoeba, add a known word list, specify which words you’d like in the sentences, and it will present you with the ones most comprehensible to you from the corpus.

It's likely tough and slow-going because you don't know the words. Consider separately studying the HSK1-3 words (600 words), and filling in the gaps. I take the HSK vocabulary (lists are here) and put them on my wall, so I can see my progress and weak words.

HSK3 is indeed only A1-ish. You can judge for yourself if HSK4 is worth it by looking at the HSK4 vocabulary.

For preparing for the HSK5 and HSK6, I whipped up vocabulary lists HSK5 and HSK6. I used Unix commands like awk go convert the raw data into a LaTeX file for compiling. Or you could also just write them on a whiteboard by hand.

Especially early on, it's important to master vocabulary, as this accelerates the rate of improving all other aspects of Chinese. (Later on, we focus more on improving fluency.) Perhaps start by studying HSK3/4 vocabulary; I suggest a "breadth first approach", focusing on studying large numbers of words inaccurately (as opposed to a small number of words accurately), and let fluency improve naturally with repeated exposure.

If you haven't already mastered the HSK vocabulary (which seems likely, given you're asking about what vocabulary to study), this sounds wildly unrealistic.

Credit goes to Alan at HSK东西 for the idea of plotting word frequencies in a stacked histogram.