For what seems like endless ages I have fought with the 5e skill system. It has never been satisfactory for my tastes; too vague, too loose, too undefined. And so I have sought solutions, and none of them have been satisfactory for me.
Allow me to borrow from Burning Wheel and Mouseguard's parlance: Obs and Factors.
All tasks have an Obstacle associated with them; Factors increases the Obstacle accordingly. Factors include the base requirements for the task at hand, and they also function like Aspects. They're less formalized like that; consider it to be akin to the Burning Wheel process.
Example: You want to light a fire? Sure thing. You'll need kindling and a tinderbox (or some other way of generating fire). Ob 2 (kindling + fire generation). But if it's rained recently, the wood is sodden, which is a Factor, which increases it to Ob 3. On the other hand, if there's a drought, the kindling might be quick to catch, which will decrease it to Ob1. Nonsense cannot decrease the Obstacle further, so in this test, you'd still need kindling and fire, but you might be able to make things work properly with inappropriate kindling.
The goal for the players is to decrease the Obstacle to 0. Ways for them to decrease the Ob include things like taking extra time, having specialized training, getting help, exposing themselves to additional risk, and so forth. If something doesn't make any sense, then I ignore the system entirely.
Players can voluntarily increase the Obstacle, too, by proposing greater effects (bigger intent, generally).
At the end of everything, I calculate the remaining Obstacle.
Ob0 is success. No roll.
Ob1 is a roll with advantage.
Ob2 is a standard roll.
Ob3 is a roll with disadvantage.
Ob4+ is failure. No roll.
Typically, most checks are DC 15, depending on all the circumstances involved. Delving deeper into dungeons or into more dangerous wilderness increases DCs because risk/reward and no other reason.