One of my least favourite things in the entertainment industry is when a new innovative or interesting thing is in an otherwise mediocre/terrible product, its used as a justification to never do it again. Of course, you can go far in the other direction and assume that because something is different means it is "good", whatever that means. I appreciate artistic risk taking tho
There's probably a decent number of random mechanics in RTSes that got used once and then never again. I remember Act of War had the prisoner mechanic, which encouraged aggressive risk taking play. It was present in the sequel but I think they could have leaned into it further. Not that any random mechanic is going to be generally applicable. I always wanted RTSes to be more asymmetric but that's harder for balance and the marketing of "competitive" ranked play.
A mechanic I really enjoyed in a recent game was from Enlisted. It's a competitive (as in, you're playing against other humans) first person shooter, but your respawns are your bot squad. Most of the time when you're shooting at a rando crossing a field, it's a bot. This experience meant that even the worst players were getting kills and having an effect (by killing the better players respawns). I'm not terrible but not that good at FPSes either, but I found the experience way less frustrating overall. The rest of the game was mid. Freemium. But I'd be interested to see that in other games.
I remember an old game called Metal Fatigue, which had a battle layers mechanic. Obviously many city/base builders have had this for ages, but I could totally see it in a modern imagining of the current Palestinian thing. Having a multilayered tunnelling mechanic to fight over.
I think overall when I'm playing non-competitive indie games, my main gripe is usually "I wish this game was the same except for this thing from another game". Modding is generally less available nowadays, and I'm not the most amazing coder so remaking the whole game with the extra mechanic from scratch isn't really an option (probably).

