Those things go without saying as far as I can tell!
So, here's my take on what's worth doing as a playbook - first off, it has to be something that hasn't been done before. Which is harder than it seems! The first-generation AW playbooks do a pretty good job of covering the bases and it's hard to squeeze in a basic concept that's not already covered by one of them in some way.
As far as I can tell, there are two ways to go about finding space for new playbooks - look for game mechanics gap you can fill or try and create a second generation playbook, one which assumes familiarity with first generation playbooks and adds a serious twist unique to that one playbook that changes how we approach the Apocalypse.
So, gaps in game mechanics! Let me try to explain this by an example - the Skinner. Of the original playbooks, it was the only Hot-based playbook. What this meant was that, via the choose-a-move-from-another-playbook, the game mechanics space a Skinner could explore would be pretty limited - there was a move you could take from the Battlebabe, and maybe some moves to your non-core-stat that other playbooks could be doing better, and that's pretty much it.
So the Maestro'D takes that game mechanics gap and fills it - now, across it and the Skinner, there is a whole range of Hot moves that you can take, broadening the possible game mechanic-y combinations characters could take. But note - the Maestro'D doesn't really change the rules of the game - what's on that playbook is mostly a combination of things that are on other playbooks and not much more. Whether the Maestro'D is in or not is not a game-changer.
For example, my own playbook, the Abacus, attempts to be one other such gap-filling playbook. I figured that, between the Battlebabe and the Operator, there aren't really enough Cool-based moves to play around with a character. So I took a look at the possible descriptors of 'cool' and tried to find one that isn't covered by the existing playbooks. I found one, 'cool=rational', and built a playbook around that, best I could.
When you're doing a gap-filling playbook, your main job is to respect the space you're working in, I think. Which means a bunch of little things that I only kinda-sorta feel, but here are a few I can actually verbalize. First, your moves have to be ones other playbooks could take - it's easy to make rules that are only usable by your character (see early iteration of the Metal Beast/Juggernaut, for example) or make the coolest move on your playbook also be the coolest move on an already existing playbook (any fighter-type playbook with a reskin of NTBFW, I'm looking at you).
Second, your moves should respect the stat that they're based on. This one's more of a taste thing, but here's two opinions of varying agreeability. First one is... if you make a straight manipulation move that's based on Hard, it's a mistake - not because you can't make fictional sense of it (rugged, strong-willed people=sexy, why not?), but because it messes with the mechanical stat economy somewhat, allowing a Hard person be good anything from violence to facing danger to gathering information to being sexy. (There's a workaround, of course! Weird playbooks don't get to roll Weird to seize by force, but they can get a bunch of conditional violence moves, so you could do a conditional Hard manipulation move with different requirements or outcomes to a straight manipulation roll).
The second thing, for me at least, are Sharp-based playbooks that allow an option of +3 Sharp and a whole lot of awesome moves, anything from violence (roll +sharp to seize by force!) to social (+sharp to seduce!) to special-effect, on top of the thing that the basic sharp moves are about - giving every other moves bonuses.. Giving Sharp cool moves doesn't respect it's place amongst the stats - it's a meta-stat that gives bonuses to other moves, and it works great as that. That's the reason the first generation Sharp playbooks don't get cool sharp moves and get cool gear instead to balance it out. Sharp as a stat isn't a doing-things stat, it's a setup-a-move stat.
So filling gaps with playbooks is hard! You have to see a gap, which there aren't that many of to begin with, plus you have to fill it without breaking the overall balance of the stats.
The other option is to forget about most of that stuff and just go as wild as you can. Basically, instead of trying to work within the limitations set out by first generation playbooks, you take them as a given and bolt on something entirely different, making a good and proper second generation playbook. Look at the Quarantine's psi-harm rule, or the thing that allows him to answer questions about the Apocalypse, or the Touchstone's thing that allows him to walk among the people, respected and loved by default (I don't have the Grotesque, but my guess is it does a similar thing). These playbooks are game-changers, you include them and what what the Apocalype is about becomes different.
If you do one of these, you need to go bog crazy with your inspirations (living incarnation of the maelstorm! Alien overlords here to see you through the mess you made!), and then enforce them in your writeup either through special rules (psi-harm), gear (your symbol is...), or writeup (when you walk among the people). That can be easy or hard, depending.
I don't know if there're lessons to be learned about this type of playbook - each one of them is supposed to be unique, conceptually and mechanically, after all.
Now the problem with these is that any of these playbooks, just by virtue of what they are, is that they are show-stealers and the game will eventually revolve around them, one way or another. So what you do is build in dependencies on other characters - that's why the Quarantine has the advice move and starts out alone, instead of knowing all she needs to know and being surrounded by a powerful infrastructure ready to support her. Make them needy and dependant, and they'll naturally try to rope in whatever other players they can into their future, to fun results.
So yes! Working within first generation playbooks and respecting the limitations or working outside of them and respecting them as-is and bolting on something new and cool on instead.
And... There's also the Faceless, which isn't really a snugly fit for this way of splitting things up. Don't have anything to say about how it works or why, though.