There's a language technique in the polyglot community known as "stacking", which is essentially the process of learning an L3 through an L2. According to research on this topic, what are the benefits of using this technique to learn an L3 as compared to learning an L3 through your L1?
I would assume that the rationale for learning an L3 via L2 is to maximise time spent with the L2. In other words, the benefits sought by "stackers" may have more to do with maintenance of L2 than with effectiveness of learning L3. I'm unaware of any specific research on the effects stacking might have on L2, but given the general principle of time-on-task (the more time you spend with a language, the better you get at it), the intuition that stacking can be useful for L2 is coherent.
As far as the effectiveness of learning L3, there are but tentative leads, nothing that would allow you to take a strong position either way. That I'm aware of, the only model of bilingualism/multilingualism that currently predicts significant differences in L3 learning depending on whether L1 or L2 is used as medium is the "L1 regulation hypothesis", which predicts that the cognitive benefits of bilingualism in terms of learning would only come to bear in L3 learning if you learn via L1, not via L2. In Bogulski, C., Bice, K., Kroll, J. 2018. Bilingualism as a desirable difficulty: Advantages in word learning depend on regulation of the dominant language (DOI:10.1017/S1366728918000858) for example, they find some indications of advantages in L3 word learning via the dominant language as compared to via the non-dominant language. That would lean in the direction of saying that stacking is not necessarily an effective strategy for L3 learning, unless you're in a situation where your L2 has become your dominant language (which could be the case for example if you are living in-country and using your L2 more often in your everyday life than your L1). But again, these are just early, tentative leads, nothing more.
If these findings are further solidified, I suppose "stackers" would then need to consider some kind of cost/benefit analysis, since on the one hand learning via L1 might be more effective for L3 learning, but on the other hand learning via L2 might be beneficial for L2 maintenance by increasing time spent with the language.
About this exact topic, Cindy Berger (or Cindy Blanco, no idea which one) from Duolingo research team answered this when I asked the question around a years ago :
Definitely doesn't decrease quality of learning! There might even be some benefits for the language you're learning it through. No matter the combination of languages or the directions you learn them from, your brain creates connections between related concepts and words. (There's really cool research from psycholinguistics and neurolinguistics showing how all of a bilingual's languages light up in the brain, even when they're using just one!) The kind of learning you're doing is building even more direct connections!