I'd prefer to do a Strategy or Puzzle game. The former requires semi-smart AI and the latter requires either procedural generation or authored levels. I'm not sure I have the time to do any kind of planning AI, let alone do it in a framework which I don't have that much experience in yet (Unity), so I pretty much have to rule Strategy out by default.
So, Puzzle or Other?
Games I think I could do with this concept:
* A puzzle/platformer like Karoshi, but with bodies lingering after death. You have to kill yourself in creative ways or specific sequences in order for one of your respawns to escape.
* An action/puzzle game, where you have to kill creatures in specific ways, or at specific times, to accomplish goals. Inverse Lemmings. I'm thinking there's a rolling deadline (ball, creature, something that moves inexorably forward) that you have to protect or shepherd, and instead of using special ammo to create or modify the terrain in front of it, you have to manipulate creatures into its path to have their death create those conditions. Give the player time bombs, repelling sonic devices, freeze rays, etc.
* A kinda misanthropic project-planner game. Imagine a NASA planner game where instead of your resources being Funding, Time, and Popular Opinion, they are instead Funding, Time, and Test Subjects. When people die, they do so in incidents that you learn lessons from faster than by careful project planning and the expensive isolation of worst-case-scenarios. Therefore: have the players identify a set of safety criteria, scoring them on Test Runs and Test Subjects, letting them optimize how to learn the most safety lessons in the lowest number of sacrificial lambs. A puzzle is failed if a test subject survives the experiment, but sees another test subject die.
... I like that last one. Has some elements that I love about Zachtronics Industries' games, and can be done in a simplified, cheesy art style. I'll post some concept art at next availability.