When called in a lesson, it would
When called in a course, it would
i would go even further and have an option for 1 unknown word too but this would be nice
That sounds efficient and very practical, keeping it focused like that would be really useful.
In its simplest implementation, this feature is simply:
This feature could also be extended a bit (more targets and/or more filters):
** Summarise:
** When summarising, show only the sentences where there is at least:
The filter UI could be a bar/sequence of (sticky) chechboxes:
Only show lines with [✓] blue [1] [2] [3 [4] word(s)
Out of curiousity, what hinders you to just skip the sentences or parts where all or almost all words are known? Do we really need a feature for this? Couldn’t it happen that a sentence is hard to undertand if you haven’t read the one or two sentences before it? Some languages are very context sensitive and some material is rather complex topic-wise, so reading those sentences despite them containing no unknown words might be necessary.
Do you ever watch a film or series with subtitles and write down in a notebook dialog parts with new words, or idioms, or interesting bits, so that you can review them later? And again and again and again?
That’s exactly the same: all and only the interesting bits in one place.
.
No, I don’t.
How does this relate to the original suggestion? You were talking about restricting displayed sentences to the ones containing not fully known words, not about restricting it to interesting ones (however an algorithm is supposed to identify them anyways).
All I am saying is that I don’t see what hinders you on skipping the sentences that don’t contain yellow words.
Everytime the LingQ staff adds a new “feature” (as you described yourself in the thread you linked in you OP 🡒 “Completel your Lesson” […]) it will be buggy, people will want tons of adjustments made to suit their individual need, and time will have been invested that could have been spent on dealing with existing issues.
Therefore I am somewhat sceptic for adding features for the sake of adding them, especially if the users are perfectly capable of performing the required task themselves. Unless I am overseeing something, hence the original question, which, btw, you haven’t answered. 
A 15-minute podcast takes 30 pages on my phone. After reading the entire lesson, I just want to click on 1 button and see the compact list of sentences with the 12 new words in them, so that I can review them rapidly one more time m
I thought the benefits were obvious. Apparently I’m wrong on this one.
Ah, now this makes more sense. You should have described it like that in the original post. So basically a more useful version of what we currently have, where just long lists of words are displayed at the end of the lesson?!
It would definitely be an improvement over the current end screen. That beeing said, how applicable your suggestion is highly depends on the language one learns. On my end, even with 80,000+ known words in Korean, even intermediate language learning podcasts contain new words in approx. 10~20% of the sentences. So on my end one long list of words would be replaced with one long list of sentences. 
(Japanese would be even worse, with 50% unknown word counts in the lessons I am using).
But if the intention is to replace the current end screen with something more useful, I am not opposed to that. “Add summary mode” somehow didn’t imply that for me.