Should Lichess improve puzzles and how so?

Hi everyone!
Today I am briefly going into the topic of whether Lichess should improve puzzles, why I think they should be improved and what Lichess can do to improve them.

Chess puzzles are great for improving tactical vision, calculation and speed. Lichess already has an amazing system for puzzles including puzzles that can be customised in terms of difficulty and theme as well as Puzzle Storm and Puzzle Streak.

Despite this, there is one major flaw of tactical positions.

No matter how complex or instructive a puzzle may be, it is only a snapshot of a game. A snapshot which we know that a solution exists to. A concrete solution that is the only way to maintain the current evaluation.

One suggestion is as follows:

Lichess plays through/allows the user to play through a few moves. Let us say that you are white. Somewhere, black has made a blunder and white has not capitalized it. The aim of the puzzle is for the user to pinpoint where exactly black has made the blunder, and play a move on the board to punish this.
However, this is probably time consuming, hard to code and does not completely address the fact that there is always a concrete solution in a puzzle. This only changes the location
.
Therefore, let us try and make this imaginary feature as similar to a real game as possible.

This feature will consist of playing a game against an engine of moderate difficulty. Let us say you play well until move 22. Let us say your queen is on c2, hanging, and you play exd5, allowing a bishop on g6 to take on c2. Engine labels your move as a blunder, so that you know that your move is a blunder. The game automatically reverts to the position one move earlier, allowing you to make a move. Say that you don't get why your move is a blunder.

Then, you click the explanation button, which displays a line which continues until checkmate or a clearly winning position. Ah, now you see why exd5 was bad! A few moves later on move 45, you play what seems like a reasonable move. It isn't losing, but it isn't immediately winning either. At once, the engine recognises a missed tactic, reverting to the previous position. And this does not mean that there is a concrete tactical solution either. There could be two tactical wins that you have missed! At the end of the training session, the feature will show your total amount of mistakes, blunders and misses including the ones that you have retried and succeeded in finding the solution to. Obviously, inaccuracies don't count - the evaluation loss there is far too low.

Yet the user still knows when exactly to make their move. The move that the puzzle reverts back on. To which, I have a solution to as well.

Now, imagine that if instead of returning to a position 1 move ago, it returns to a position anywhere from 3 to 10 moves ago a few moves after you have missed the continuation. It asks you to identify and find the correct continuation where you have missed it. If unfound, then it shows it for you after some time.

Undoubtedly, this 'imaginary feature' probably has many flaws which you will be able to identify, but the message should be clear.
Puzzles have a clear, concise solution and the player knows exactly when it is, so they are on guard. A true player should always be on guard, a quality that often lacks in real games. After all, how can one win a war if they never fully pay attention?

Maybe there should be a solution.

P.S. I am not suggesting that puzzles are bad in any way, sort or form. Puzzles are essential and important factors which anyone would highly recommend. Even if they may be flawed, they can be crucial factors in improving one's gameplay

Thank You for your time.
Please feel free to make suggestions in the comments!