Since someone mentioned QTE and shimmying, I'll take it a few steps further:
Games that reduce complex actions down to a single button push that's context-sensitive while temporarily disrupting control of the character with an uncancelable animation sequence. That includes stealth takedowns, snap-to cover mechanics, traversal animations (climbing/grappling/mantling/crawling/shimmying), QTE, Doom glory kills, L4D-style boss attacks, etc. The fact that these situations often have to resort to obnoxious onscreen prompts to explain to the player why the character is no longer responding to the same control scheme they were a second ago is an obvious red flag, imo. I'd mostly attribute this trend to console controllers being inadequate to handle the variety and complexity of verbs/actions the game design is wanting to have. What all of this tends to lead to is having the most intense, interesting, and involving moments actually have the least amount of player agency and input (ie. gameplay). You're likely to have less control over your character when climbing a precarious cliff, hiding behind a rock in a gun fight, or grappling with an enemy than you do when sifting through your inventory deciding which rat hide to sell to a vendor first.