So, this is related to the OP's original post. I'm working on an underwater adventure set in the world of today. The players are scientists onboard of a deepsea habitat, when all hell breaks loose: sections are breached, flooding occurs, loss of O2, electrical fires, power outages, you name it. The habitat also automatically seals a number of airlocks, to prevent the entire thing from flooding. Now the players need to restore functionality, and save each other from whereever they happened to be at the time of the disaster.
I want skill checks to be more meaningful and involved. I'm thinking of a system where a failed check may result in an escalation, and where neglected tasks result in further damage to the habitat. I want this disaster to feel like a bunch of firefighters trying to contain an out of control fire. But perhaps the opposite should also be true; a success should help bring the situation under control. It shouldn't be hopeless, just stressful.
How this might work, is that I keep track of the state of every module of the habitat. If the players focus only on containing the flooding, the electrical fires may get out of control, causing an explosion that destroys a module completely, and spreads the fire to an adjacent module. But I don't know yet how to track this in a balanced way exactly.
I'm also thinking that certain tasks may cause an extra loss of O2, especially when failed. And being low on O2 and while under stress, may result in sanity effects, as in Call of Cthulhu. The GM (me) will be the one keeping track of everyone's oxygen, and the players will need to take an action to check how much O2 they have left, but depending on their sanity, they may read it wrong. Muhaha.
I'd consider taking a page from a cooperative board game.
How they usually work is two little tricks.
There is (a) a goal of some kind, (b) an escalating background problem, and (c) a board state of immediate problems.
Every round the level of (b) makes (c) worse. Early on, you have enough resources to solve each round's (c) completely, and have resources left over to work on (a).
But almost completely. A bit of (c) tends to build up. You'll get to it later.
As the game progresses, (b) gets worse, until it will overwealm you. Towards the end of the game, you have to give up almost completely on fixing (c) in order to finish off (a), then bail.
An example of such a game that would map to a "underwater base is failing" might be "forbidden island".
(a) Your goal is to get the 4 macguffins. You make progress towards them via position on the (tile based) board and collecting and trading clues.
(b) There is water's rise scale, the rate at which the island is sinking. Each turn there is a risk it goes faster.
(c) Each tile on the island becomes flooded (flip the tile) or sinks (if you pick it again) and goes away in a pseudo-random process. Players can "shore up" flooded sections (repairing them), but the tile mechanics make shoring up usually temporary.
You could imagine a "board" of the various components of the station, and things become endangered. Fixing then takes time. The overall station damage track is something you cannot fix, and continues to get worse, making the rate of system failure go up. Initially failure happens slow enough they can repair it, eventually too fast.
The "skill check" in Forbidden island is "I fix it", but the skill challenge is still complex and interesting.