I've entered Ludwig jam, and there are 5 days left to create a Foddian game- something "unforgiving, with no real checkpoints.. a constant threat of loosing huge chunks of progress".  I have a somewhat grand idea for this and don't know if I'll finish in time with all the features I'd want. I'm dealing with intense anhedonia/lack of motivation from my depression, and need to remind myself that even a tiny bit of progress is good; and it's the process of learning that's most important anyways. The game definitely does not need to be polished; as long as it's kind of fun and presentable I will be ok with the prototype. 

I'm following a "Hill Climb Racing" Godot tutorial by LucyLavend, hoping I can transfer ideas from that into the mechanic of my game. Worst case scenario, it'll  involve skating or floating downhill super fast (the opposite of typical climbing Foddian games) and having to maneuver around obstacles. Best case, players will skate or float downhill super fast while having to time pole vault takeoffs or grapple hook maneuvers over obstacles below. I think I'll allow applying brakes and moving the character yourself instead of an endless-runner type scenerio, but the slope will be steep and low-friction to keep things hard. When applying brakes, you still move forward a bit too. 

Later I might make some extreme skate park inspired obstacles- like a 360 degree loop-de-loop. 

Often I put in a lot of time (on art, planning, or attempting to program a mechanic) just to realize I should change gears to something totally different. I think this is the sunk cost fallacy; good to keep in mind. 

Another idea involved moving through a cave with two or more extendable grappling hook arms. This sounds a little less exciting than my downhill pole vault/grapple hook idea, but maybe easier to implement.