(First, I'm not a lawyer, this isn't legal advice, retain a lawyer if you have any doubts.)
Referencing things has nothing to do with copyright law
Let's get one thing out of the way: Dungeon World's references to bits of pop culture and literature and stuff is in no way a violation of copyright, trademark, or patents. Trademark and patent law don't even remotely apply, and copyright prevents copying and transforming a copy, not plain referencing. Those are the three prongs of IP law, the entirely of what creators have to worry about, and none are relevant to referencing pop culture. This will probably make the rest of this stuff make more sense.
You can make stuff for Dungeon World. Please do!
Dungeon World is licensed under the Creative Commons Attribution 3.0 Unported License. That means you can make anything you want using even its text, and you can sell it so long as you give attribution to Sage and Adam as the creators of what you're using. You could even just copy the, say, Druid class wholesale, call it "The Shaman" without changing anything else, stick an attribution line at the bottom, and put it up on RPGNow for $5 and that's legal. See "How do I properly attribute material offered under a Creative Commons license?" for details on how to correctly attribute Dungeon World as the basis of your work, if you end up using any of its text (like, say, borrowing a Wizard move whole).
But for just a class, you don't even need that. Publishing a class that's compatible requires no use of copyrighted material from the original, so you don't even need the CC license in that case. You just publish.'
Even saying "This is compatible with Dungeon World" is legal: that's called "nominative use", and the use of a trademark to indicate compatibility is a protected right that you have. TSR went to court over this ages ago, and lost hard, so it's even a fact that's been tested in the RPG world. Kenzer Co openly publishes material for D&D without using the OGL, because they aren't copying any text and it's legal to say "This works with Dungeons & Dragons" without getting WotC's permission.
Besides which, Sage and Adam encourage people to make stuff for Dungeon World, an have provided a "For Use With Dungeon World" kit that you can (optionally) use if you want a recognisable shared graphic.
Lots of people have already created new classes for use with Dungeon World and are selling them just fine. Not only are you in the legal clear several ways from Sunday, Sage and Adam aren't interested in suppressing that kind of thing.
You should be paying more attention to the actual threat
All that said, you do have to be careful basing a class on the Witcher. File the serial numbers off, make sure that people can recognise it but you're not using any words, names, images, visual character designs, or anything else that is owned by Atari, and you should be okay. If you're planning on making any money off it, you may want to research your legal rights and obligations here. You can produce something based on someone else's work, but it has to be pretty substantially separated and only recognisable by people who really get the references.
A DW class PDF for sale on RPGNow based on the Witcher is highly, highly unlikely to draw Atari's attention, but you never know. They've sued people into bankruptcy for less. Your safest best is to not make it obvious that it is a Witcher, because if you use Atari's trademark to draw attention to your class, if you advertise "You can play the Witcher in Dungeon World!" as a feature of what you're selling, you've violated their trademark and you are—if they notice you—potentially in serious trouble. At best they send a Cease & Desist to RPGNow and they pull your PDF. Worse, RPGNow closes your account. Worst, Atari goes for you directly.
So don't get noticed, and don't do anything that will make anyone notice that your class is based on the Witcher until they've already bought it, and you should be in the clear.
Of course, if you're not selling it, then you can probably get away with saying you've made a Witcher class and here's the PDF, since there's a lot of precedent for ignoring fan-made stuff that's not for profit, but you still have to watch out, because you're on the wrong side of the law then, and you're only OK so long as the company a) doesn't notice you and b) turns a benevolently blind eye.
New answer for a renewed question. The old one is here
I'd say that a hidden move is quite against the grain of *World games'idea of moves. It robs the players of their agency in determining their characters' story.
And in your case, the intra-party conflict that may occur if the infiltrator is exposed seems to be more important than whether the infiltrator will be able to commit acts of sabotage and betrayal behind their backs. I'd base the move on that and add another move for the other party members to respond.
Infiltrator
You are an agent of your deity with a secret agenda. When you commit acts of sabotage and betrayal behind your comrades' backs, mark experience, and choose one comrade that doesn't already have it to receive the "Suspicious" move immediately.
Suspicious
You are suspicious of the infiltrator's weird behavior. When the Infiltrator commits acts of sabotage and betrayal behind your back, roll+INT.
- On a 10+, you have seen enough, but just need some more evidence. Replace this move with "Deeply Suspicious"
- On a 7-9, you noticed something unusual. Take +1 forward to your next "Suspicious" roll, but getting there wasn't easy. Choose one:
- You had to get into trouble to get your information and come back with 1d8 damage.
- Information is pricey. You had to spend 2d6x20 coin to find some clues.
- Lies are everywhere. You wrongly implicate another comrade (The infiltrator will tell you who). You cannot aid or interfere with them until the next time you roll "suspicious"
Deeply suspicious
Replaces Suspicious
Same as "Suspicious" but add these to the list of possible outcomes.
- On a 12+ , you have enough evidence to expose and implicate the Infiltrator. Now everyone in the party knows better. You decide the fate of the traitor.
- On a 10-11, you collect a solid piece of evidence, slowly building a strong case. Give another comrade the "Suspicious" move if they don't already have it, or replace their "Suspicious" with "Deeply suspicious"
This is based on the assumption that the infiltrator is a well known and trusted member of the party in the beginning. The first act of the infiltrator is essentially free. His or her comrades suspect nothing in the beginning and they trust the infiltrator. But then it gets harder and riskier every time.
And yes, all this can and will eventually leave the fate of the infiltrator in the hands of another player. The question is, when?
Best Answer
So, you want to make more of your game's class moves into "attack" type things.
You can totally do that. There's a big caveat though: you won't be playing a "Powered by the Apocalypse" game anymore, in that it won't run and flow and evince all the emergent gameplay stuff that Dungeon World (and its parent Apocalypse World) are celebrated for. It won't play like Dungeon World at the table, and won't hand the GM the material and hooks they need to riff off the characters' actions. Your GMing section will end up quite different because of that.
Why? Because making more of your class moves about attacking means you're moving most of the fiction-generating mechanics into combat, leaving the out-of-combat events more undefined mechanically. With fewer moves triggering outside of combat, the flow of the game overall will be located more often in the GM's sole hands instead of located in the back-and-forth between players and GM, requiring (or allowing! two sides of the same coin) the GM to make up more of the game's overall structure and direction by themself. This will require a rejiggering of the GM's moves, since what move to make and how to make it is usually heavily informed by what the players have been doing – but with fewer out-of-combat class moves, that becomes less dynamic and more predictable, giving the GM less inspirational material to work with on the fly when making their GM moves. Along with that, a rewrite of the Agenda and Principles would probably be necessary – specifically, you won't be able to play to find out what happens anymore; and that in turn will require changing the Agenda to allow for presenting more of a GM-controlled plot.
Moving more class moves into the sphere of combat would shift the game into a design space that's halfway between Dungeon World and D&D 4e. If 4e is your bag, that might be a cool project to work on. On the other hand, that makes it completely unexplored design space, so you're pretty much on your own. That might be a downside, or it might be an exciting opportunity. You might be able to get some advice from the designers on the Apocalypse World / Dungeon World hacks design forum, but you might not – you might be doing totally alien things and teaching them new stuff.
So if that sounds good, go for it. Sadly, I can't give you any advice on making good moves then – and probably no-one who is good with DW can either.
If that doesn't sound good though, then your next step is probably to step back from your project and learn more about what makes Dungeon World really tick. (This might all be irrelevant, so I'm sectioning it off.)
If that's what you choose, for starters you could look at how moves should work and the relationship between moves and fiction as they have to be used during play. As written, those two moves aren't good Dungeon World moves, mostly because they're backwards: they lead with mechanics, when Dungeon World always begins and ends with the fiction. Another place to start research if this is you, is to dig into or ask about your perception that classes have few moves they can use in combat – that's actually not true, and points to a common misunderstanding of how moves work in general, and during combat specifically.
A great way to just absorb wisdom about Dungeon World is to browse through the Dungeon World subreddit. There are a lot of people who've been playing for a while and just "get" the game fundamentally, and the creators hang out there too. There's also the fan-written, creator-endorsed Dungeon World Guide which is specifically about explaining how Dungeon World ticks to established RPGers coming from very different systems.
For an example of working up a full DW move, let's look at your first move:
This has no trigger, so it will never happen during combat. However, it does look at lot like a 4e power. The difference is that moves are never chosen: the player describes what the character is doing, and only if what they're doing matches a move trigger (any move trigger), does a move happen. Conversely, 4e powers are chosen and then just happen. Additionally, it doesn't tell you what to roll, so technically the roll will never happen (you only ever roll when a move tells you to) and the results won't resolve.
So let's give this a trigger:
Notice that the trigger doesn't say when you do kick an opponent – that condition would (almost) never trigger during combat, since how can you kick someone successfully before you've checked to see if it's successful? There are ways it could happen, but it would be a weird two-move dance using Defy Danger or an unarmed Hack & Slash that was aimed at the sensitives, and that would actually stack the dice against ever bothering to try. Or, of course, out of combat while kicking a defenseless opponent (which requires no move). So in order to make it trigger, we have to give it something that could actually plausibly happen in the fiction during the flow of combat – an attempt to kick in the crotch fits.
Now let's give it a roll. Dexterity sounds good, because this is a precision move.
Now, because the fiction leading up to the move is different, we need to adjust the outcomes and wording:
A bit better now. But what if there are no allies near? The 7–9 doesn't make sense all the time, and a move should always just work as long as the trigger was met. Also, it's important to remember that a 7–9 result is a success – a qualified success, but still success. So we need to give the player success, with conditions.
Better. The 6- needs some work though: passing out is a pretty weak for a miss result. We should add that the passing out is in addition to the normal miss result, which is the GM making a move.
There. That's a move in the mold of Dungeon World.
But there's still something to consider before giving it a stamp of approval and putting it into a game: does this do anything that an existing move doesn't? On the surface, without any more work: no. This is pretty much an in-combat Parley, using the leverage of "I'm going to kick you in the sensitive bits otherwise." It's a bit different in that it takes +DEX and stuns the kickee, but that's not really worth a move with this much write-up. (You could do a simpler stat-substitution move instead: "When you inflict pain to intimidate, use +DEX to Parley instead of +CHA.")
Now, that doesn't mean you can't make a custom move for fighting dirty. It just means that, to bother with a custom move about fighting dirty, you have to consider what you want to give a player consistency guarantees about, while fighting dirty, that they don't already have from the existing moves. (That's what moves do: they give consistency to fictional efforts. They don't actually make the character more powerful.)
This also demonstrates something about combat: it's way more flexible than it at first appears. Any move you can trigger during a fight, you can use. That includes the "social" moves – and I put that in scare quotes because calling them that is misleading in exactly this manner, making them sounds like they don't apply during combat.
If I were to write Dirty Fighter myself, I'd first start by thinking what advantage fighting dirty gives a combatant. People are mostly going to be fighting dirty already (who fights honourably when life is on the line?), so we don't actually need a move to make the dirtiness happen. A move could, though, codify the ability to really put an opponent into a spot when they're vulnerable. On the other hand, most of the time when an opponent is vulnerable, you can just hurt them or do whatever without having to go through a move like Hack & Slash. (Remember that, if you're not in a melee – your opponent isn't endangering you and neither are other combatants – you don't do Hack & Slash, you just murder them without a move.) That makes thinking of a trigger that doesn't actually make it worse than normal kinda difficult. In that case, Dirty Fighter would probably end up being a passive move: something that just helps you when fighting unarmed, maybe.