All of your games feature a mechanic that has a "random chance" element to it.
While this isn't inherently bad on its own, and each use of it in your games DOES have some interesting parts to it, something that is very frustrating about it in your games is that (1) the feature has a high time investment cost to be able to do the "chance roll", and (2) is VERY swingy, with potentially insanely good or extremely mediocre results.
In "Squares", the issue was line bonuses: the ideal is max bonus per line, or nearly. Swingy in the early game, when you're struggling to fill 1 or 2 lines, and just as swingy in the late game; the difference between `9-9-9-9-9-9-9-9-7-7-5-2` and `9-9-9-9-9-9-9-9-9-9-7-7` is that the latter is over 10 times faster, a full order of magnitude.
In "More Tiles," the "Bless (3x bonus to random tile)" lottery is absurdly swingy depending on what you roll, and there's a high time investment per-run to be able to roll, and once again, you might get good rolls all the way up until the very last where you might lose it, since the way the bonuses stack matters.
In the previous iteration of THIS game, the issue was, of course, "Requests," because (once again) the bonus multipliers compound. The request rewards are random; for example, one critical bonus is the "x2 card drop rate" reward. If you luck out and roll multiple requests in a row on the same run that reward that, that 2x becomes 4x, then 8x, then 16x, then 32x if you get five "x2 card drop rate" requests. Even more swingy is the "cost" of the request: pray it wants 100k T1's and not 100k speed crystals, as that's the difference between being able to do more requests or having your run stop there. Very swingy.
And in this game, the "swingy high roll high-time-cost-to-be-able-to-roll" issue is "merge gems."
Each of these design choices highly incentivized me to savescum. I do not personally mind it all that much, because that (manipulating save-loads to get the random result I desire) is something I'm personally willing to do, because otherwise a bad roll after a high time investment would just lead to me quitting immediately.
I see this as a recurring "design problem" in your games, which are otherwise competently put together (though, no offense, but very clearly AI assisted, and I really think you'd be better served sticking with and improving one project than rapidly abandoning and letting the AI totally rewrite/refactor shit constantly; using AI can also be a skill, but part of developing that particular skill is figuring it out how to get it to improve existing things instead of churn out endless new stuff that will also have things you want improved), so I felt like I should take the time to write out why I feel like it's been an issue.