"Concealed" difficulties aren't actually different from how Fate operates in general.
As they're mentioned in Masters of Umdaar, they're not concealed during the roll, just concealed before the roll. But that's not that strange. Here's an example of play from the SRD:
Landon stalks around the siege tower of the Red Emperor's fortress, trying to sabotage the ballistas. If he succeeds, the army who hired him has a much better chance in the field when they attack tomorrow morning.
Amanda says, "Okay, so you make it to the top of the tower, and you start working. But then, you hear footsteps echoing below you in the tower -- sounds like the next guard patrol got here just a bit early."
"Damn," Lenny says. "Figures I'd get the one guard squad with real discipline. I need to disable these and get out -- if they find me, General Ephon already told me he'd disavow my existence."
Amanda shrugs a bit and says, "Work fast? You’re looking at passive opposition here—crunched for time, and dealing with intricate machinery bits, so I’ll call that Great (+4)."
Landon has the Crafts skill at Average (+1). Lenny grumbles and says, "Should have convinced Zird to do this." He rolls, getting a +2, for a Good (+3) result. Not good enough.
Lenny knows how hard it is to finish the sabotage when he's making the Crafts roll, but when he knows how hard the Crafts roll will be, he doesn't have the option of going back and deciding that Landon will do something different. He also can't just propose a bunch of courses of action to Amanda and see which one has the lowest difficulty before deciding what to do.
What is different is the larger number of pre-committed actions.
Instead of demanding success as a challenge would, or involving rolled opposition as a contest or conflict would, the pressure in a cliffhanger is "time", represented by the limited number of rolls available to you. You have five total rolls to get three successes on Overcomes, including rolls to Create an Advantage. (This is one reason why successful Create an Advantage rolls during a cliffhanger also get a hint as to the success rate of an approach, even if they're not revealing a cliffhanger aspect.)
In a conflict, everybody gets to say what they're doing when their turn comes up, and as a result they know the difficulty of the thing the last person in the conflict tried. They can pile on if it was easy or try something else if it was hard. However, to manage the tension of a cliffhanger when multiple characters are involved, the GM is encouraged to batch up the initial actions, such that the PCs all declare what they're trying at once, when none of them have any information about the difficulties of the various approaches:
If several players want to make attempts, listen to what they want to attempt and put them in the most logical order. Or, if you want to increase the tension, order them from least likely to succeed to most likely to succeed.
[...]
The GM decides to "lock in" these first three exchanges, leaving the last two open; the players can decide what they’ll do after seeing the first three.
-- Masters of Umdaar, p.30, "Cliffhangers: Multiple Characters"
This encourages them to "spread out" and try different approaches, or spend time Creating an Advantage to get information. They'll still find out how hard the approaches are when they try to make Overcome actions using them, but nobody in a batch of actions can change what they decided to do in response to that information. (This is another point in favor of resolving actions in order of highest difficulty to lowest, so that nobody makes their roll knowing for sure how poorly they chose.) After the initial actions the PCs are better-informed, and can choose approaches to get the successes they need more confidently.
If you have a large player group, you probably want to limit these pre-committed actions to an initial batch of two or three, so that the remaining actions can set the group up to succeed, or at least to fail less completely.
