You can't optimise in Dungeon World, not like you can in D&D.

To do what you want to do, you have to approach it differently. It's a different game, after all.

Unlearn what you know about D&D

There are a bunch of assumptions in the question that make sense for D&D and make no sense for Dungeon World. Trying to play Dungeon World according to optimisation skills and assumptions honed by a completely different game with completely different philosophy of mechanics will fail and be frustrating.

Save yourself that trouble, and assume that everything you know about playing D&D well will lead to a poisoned-spike pit trap full giant bees in Dungeon World. Metaphorically speaking.

There are no skill checks, only triggered moves

There are no skill checks. When you want to do something, do it. You never go straight to the dice and a Special Word or number in the rules or on your character sheet. (Actually, you can, but the GM's job when you do is to respond to that with “cool, you want to Parley. Tell me what that looks like. What are you saying?” before you're allowed to touch dice.)

To do something, say what you character is doing. To intimidate a city guard, say what your character is saying to the guard that is intimidating. (You don't need to say this in first person. You can also paraphrase the message you're sending. The rule that “to do something, describe doing it” isn't there to make everyone pretend to be Shakespearean actors in their living room, it's to require specific committed action from the character before finding out what the result is.) If you want to persuade someone, describe your character saying something they hope is persuasive. To bluff, ditto.

None of these things require a roll. Maybe you'll say something in the course of describing your character's words and actions that matches a trigger of a move — then you immediately resolve the move. The important thing to remember is that moves trigger based on actions you describe your character making, not on things you want to have happen. You don't get to make a move because you want to intimidate someone; you get to make a move because in the course of actually doing intimidating things, a move's trigger is matched. And, critically, it might not be the move you were thinking of at first!

You will not get to use your best abilities at-will

Because moves are triggered by doing things in-world, moves will happen whether or not they're what you want to happen, in mechanical terms. If you're climbing a rope and get attacked by large flying things, and suddenly you're in danger of falling, you won't be able to Diplomacy Mind Control in response no matter how much you want to. It'll probably involve strength, or maybe dexterity, or maybe even constitution (depending on exactly how you describe respond to the threat of falling, as above — swinging yourself out of the way is likely to be dex, while just hanging on for dear life would be strength, etc.).¹

Because moves are triggered, never chosen, you can't “press” them like a button whenever you want to activate them. You have to narrate your character's actions in a way that works in the situation at hand, which might not always permit your preferred abilities to come into play.

Characters in DW are competent, but not because they have great stats like in D&D

Characters in DW are much more competent at 1st level than they are in any edition of D&D at first level, but they are competent despite there being absolutely no guarantees of succeeding on any given roll.

Competence is not about rolls, in DW. Competence is about having control over a situation and having options when it goes out of your control. Every PC gets these things for free just for being a PC in a game of Dungeon World. That's part of why Dungeon World is awesome (IMNSHO).

You won't see the source of this control or options on your character sheet or in the player moves though. These things exist in the rules the GM must follow, and those rules are hidden from the players during the game session (unless they choose to read them beforehand, which is fine, but they still aren't talked about during play). The GM's rules are about 90% of the actual game engine, so it's not surprising that major details of PC competence are encoded in those rules.

So how do you be competent in Dungeon World? Try things! (Describe them, of course.) You'll succeed way more often than you would in D&D, just because of how the dice odds work and the lack of “that's hard so you get −X” modifiers, even if it's something that might normally be considered hard. (One of the GM's rules is that their job is not to make things hard, their job is to make things interesting. So you as a player can attempt difficult things, and you'll succeed more often than expected based on D&D experience, but win or lose the result will always be interesting and fun. And, losing will never suck, only change your options.)

There is no such thing as optimisation

You can't optimise a Dungeon World character. You can do things that look like optimising to a D&D veteran, but it doesn't actually do what you think it does. In particular, making sure you win certain rolls (e.g., ones with +CHA) as often as possible isn't actually an optimisation. Because of the way the GM's rules work behind the scenes, all that does is change what your options are after the roll. It doesn't make you succeed at adventures better, it just allows you access to different situations than otherwise. Succeeding on a certain kind of roll more often just changes the kinds of adventures you have, because one bedrock rule of Dungeon World is that there will always be something adventurous happening, and neither failure nor success eliminates that. They just determine what kind of adventure you'll have.

That probably seems nonsense, so a pair of examples is in order.

Convincing the Goblin, example 1:

You're exploring the Forbidden Dungeon and meet a goblin. You have terrible Charisma, but try to Parley anyway.

Player: I Parley! I want him to join us.
GM: Cool, okay. What do you actually say?
Player: Oh right. Okay, so we've just stepped into the room? Right then, before the goblin freaks out at us invading, I raise my hands and take a step forward. I tell him we're hiring and we'd like to pay him shiny gold to be our guide.
GM: Okay, that does sound like a Parley, using the pay as the leverage. Cool. Roll +CHA.
Player: A seven! Yes... wait, no, my CHA is −1, so that's a six. Parley doesn't say anything about less than a seven, so does nothing happen?
GM: No, I get to tell you what happens. (Here the GM consults their move sheet and picks one. In this case, they picked Reveal an Unwelcome Truth.) So, you're offering the goblin pay and you see that it's not even paying attention. Actually, it's kind of staring at something behind you, just before turning and running in fear. Ranger, you're in the back, right? So just as the goblin turns and flees you hear heavy breathing of something big behind you — What do you do?

The adventure here has taken a turn, and for the next little bit will be about dealing with the big monster that can somehow sneak up on the party. That's interesting! (Maybe it can be talked to too?)

Convincing the Goblin, example 2:

You're exploring the Forbidden Dungeon and meet a goblin. You have awesome Charisma, and try to Parley with the goblin.

Player: Before the goblin freaks out at us invading, I raise my hands and take a step forward. I tell him we're hiring and we'd like to pay him shiny gold to be our guide.
GM: Okay, that sound like a Parley, using the pay as the leverage. Cool. Roll +CHA.
Player: A seven. Wait, no, my CHA is +2, so that's a nine. Still a partial hit. Parley says I have to “give concrete assurance of my promise right now,” right? Okay, then I say—
GM: Hold up, we need to make the move's result happen in the game world, first, so you know what kind of reluctance you're dealing with. So yeah, the goblin gets this glint in his eye and leans forward, obviously hooked a bit. “Gold? Gold is good, but lightbringers might just kill Gribnak and take his gold back. Show Gribnak you trust his guiding first then Gribnak trust you. Go back to statue and open door with blood orcs symbol, bring blood orc head back. Then Gribnak guide you, and we kill the blood orcs together!”
Player: Really? He just wants us to kill some orcs we were going to kill anyway?
Ranger: Wait, what if it's a trap?
GM: Hold on, remember the move. Parley on a nine says you have to give concrete assurance of your promise, but if you do, “they do what you ask.” Let me ground that in the game world again, okay? Alright, *ahem*… This goblin apparently named Gribnak is agreeing to guide you for gold, but only if you prove you trust him by dealing with an enemy of his. You can tell from his body language he's telling the truth — he looks all earnest and like he really wants that gold, and is willing to welcome his new lightbringer overlords.
Player: Oh, okay then. So let's go get him that head.

(Some time later, after more adventures, the party obtains a blood orc head. Gribnak steps out of the shadows — he's been following them — and does a little dance. Then he swears a weird but serious-sounding oath to “lead them true for holy gold.”)

The adventure here has taken two turns: for a bit it became about the blood orcs and dealing with them, and then it became about having a goblin as a henchman and all the shenanigans that could lead to. That's interesting!

So you see, both failure and success lead to interestingness. Optimising away failure doesn't actually optimise for success, it just optimises for a completely different set of adventures — which will happen regardless, because it's a big game world where unpredictable things are going to happen anyway. The optimising therefore is not actually improving anything, just changing things. It's a totally legitimate choice to go for a maximal bonus to something (it's fun!), but if you think of it as “optimising” you'll be super-disappointed that it gets you no closer to having a “win” button. Dungeon World just does not have those.

Trying to “optimise” in DW as if it's D&D will lead to unexpected failures

Can you be good at something in Dungeon World? Absolutely! You can be a great diplomat, even. What you can't do is optimise for being a diplomat. You will fail in unexpected ways if you try.

1. There are no dump stats in DW, only choices.

   Choosing to be great at Charisma-based stuff is not optimal for anything except “I want to have a lot of interesting conversation,” because it comes with trade-offs in things that you won't be good at, and you won't be able to use Charisma-based activity to avoid them. There will be Danger to Defy, there will be Slashes to Hack, there will be Realities to Perceive, and having a high Charisma will let you circumvent a few of them perhaps, but not all of them.

   High Charisma won't even get you out of every or even most fights, like you might be used being able to build for in D&D. It's impossible to completely avoid failure even on your best stat, just because of how the dice work and the limited availability of bonuses. Your best stat will fail you sometimes. You will have to face your weakest stats! Attempting to optimise Charisma so that you can avoid non-Charisma stuff just isn't a strategy that works, and will often create kinds of situations you thought it would let you avoid.

2. “Optimising” for one stat and then trying to use it for everything is also a great way to create trouble.

   If you go into Dungeon World hoping to play an invulnerable Diplomacy Mind Control machine, this will be frustrating. (With a mindset harmonious with Dungeon World though, it can be fun.)

   Putting yourself into situations where you hope to use Charisma, but the situation changes so that you can't? Being overconfident in an ability that isn't ever going to be a “win” button can have that result. There's nothing wrong with creating trouble for yourself like this, but if you start with the attitude that it will work, it will just be frustrating. In normal Dungeon World play, getting into messy situations by making strategic mistakes like that can be a source of fun and surprising twists, so the danger here is only frustration due to mismatched expectations. Doing that knowing what you're getting into is fine. My group's Ranger has been getting in trouble on purpose and is loving it.

3. Avoiding all but your best stats means you will lose out on XP and levelling.

   One of the major sources of XP in Dungeon World is missing on a roll. If you max out Charisma so that you often succeed, and studiously avoid situations in the game world where non-Charisma moves are triggered (which, as above, is not possible to do entirely, but you can certainly fight the current of the game a bit and reduce non-Charisma moves being triggered), then you will gain XP just that much slower than everyone else. After a few sessions, your fellow party members will be levelling up and you won't be.

   So trying to minimise misses when rolling is not only impossible to achieve perfectly and a great way to frustrate yourself, but actually makes you worse off in the long run. Trying to “optimise” in Dungeon World is a trap! The system just doesn't have any holes that make mechanical optimising possible, and trying to find the non-existent optimal numerical strategy will lead you to do things that make your character perform worse than those of players willing to face failed rolls.

4. Avoiding all but your best stat is how to critically succeed at dodging fun.

   Even if you're successful at avoiding dangers that you can't handle with Charisma, despite the statistical improbability of that, the result will only be maximising a lack of involvement in the adventure. Fanatically avoiding failed rolls just results in keeping yourself out of the events, complications, and heroic reversals that Dungeon World is creating on purpose because that's the fun stuff. Avoiding the dangers of adventure is what sensible farmers who stay home and leave dungeons alone do. It's fine to be good at something and use that when you can, but trying to use it to avoid situations where you can't use it is just opting out of the game.

   The game is not made to be fun for someone trying to avoid getting caught up in the ups and downs that make the adventure adventurous. Be an adventurer instead! Be fearless, and accept that you will get into trouble, and that the point of being in trouble in DW is so that you can be amazing and awesome getting out of it.

Hopefully you just want “good,” not true “optimal.” Good is possible, and a valid strategy. It's just not better (nor worse) than the player who decides to go face-first into everything their character is bad at.

Being the Face in Dungeon World

So being the Face in Dungeon World is not actually about maxing your Charisma modifier. It's about choosing to solve problems with talking, and that really only needs a decent, not maxed-out Charisma. A maxed-out Charisma may also be fun, but it's not necessary; and the Parley and Defy Danger moves are much more effective at creating fun, interesting adventure than they look like to someone viewing them through a D&D veteran's eyes, because they fundamentally function differently in this different system.

To be the face, choose to talk. Make your character bluff, intimidate, and persuade by actually bluffing, intimidating, and offering persuasive arguments. Sometimes that won't even trigger a move, and it will just work — or the NPC will push back in an interesting way. (Again, this has to do with how the GM's rules work.) Other times it will trigger a player move and interestingness will result. Always it will be adventure, and new exciting situations to deal with. Never will you find that doing the same thing, over and over again, is effective — there is no Diplomacy Mind Control in Dungeon World.

And really, any character can do this. The Fighter can be the Face just as well as the Cleric or the Bard. The Bard has a few extra fun tricks for it (like moves that let you create NPCs so you can go manipulate them), but isn't going to be astronomically better at it. The difference between a good Face and a bad Face is the player: the choices they make, and what audacious things they try to get NPCs to do with and for them.

Be daring! Be bold. Be charming! That is how to be the Face in Dungeon World. The stats just help you figure out when it works, and what twist happens when it doesn't.

1. Actually, I thought of a way you could use Charisma to Defy Danger when the danger is “these big winged things are banging into you and you're about to lose your grip on the rope, what do you do about that?” You could respond by saying you're going to persuade one of these winged things to catch you! Daring, very daring, but possible. Assuming they understand you (which isn't a given, but is possible too), and would require something you could say in very few words.

   How you could do that is tricky, because you need to figure out leverage, and what kind of leverage can you offer big winged things buffetting you? Maybe “Noo! I'm your long-lost prince that will fulfill the prophecy....!” as you fall? That might do it. Maybe “Catch me or my ghost will haunt you forever!” as you plummet? Maybe, could be interesting. Better roll well on that Defy Danger +CHA though, because failing will be very interesting. It would be totally awesome if it worked though!

   See how the details matter though? The details being credible are what permit the attempt, when normally Defying Danger with Charisma would seem to be impossible. The different outcomes is also why saying how is vital and required for a move to trigger — if either of those get a rolled hit, the outcome would still be getting saved from falling, but the consequences afterwards would be very different. One will lead to adventure entangled with a prophecy (that you just made up?) among these possibly-worshipful creatures, another will be dealing with possibly-hostile creatures who think you have magical powers to curse them. Very different outcomes!

Ok. First, a Rakshasa is not a low level encounter for a level 6 party. In fact, the DMG uses this specific example when talking about CR:

In addition, some monsters have features that might be difficult or impossible for lower-level characters to overcome. For example, a rakshasa has a challenge rating of 13 and is immune to spells of 6th level and lower. Spellcasters of 12th level or lower have no spells higher than 6th level, mean ing that they won't be able to affect the rakshasa with their magic, putting the adventurers at a serious disadvantage. Such an encounter would be significantly tougher for the party than the monster's challenge rating might suggest.

So a Rakshasa would be a strong opponent for a 12th level party.

That said, if this character is supposed to be of a more tactical bent, he should fight like a modern commander. That is to say: get someone else to do it for him. This would minimize the effect of his statistics on the fight, whether they're too strong or too weak.

I'd recommend you start by reading about Tucker's Kobolds, since they exemplify the approach you should be taking. The short version is make use of favorable terrain, hit and run tactics, and other similar, asymmetric, tactics. Fortunately, a thieves guild should be extremely good at this sort of thing (I don't know the statistics for chuuls, beyond them being cr4 individually, to comment on their usefulness).

An example of this type of strategy would be having the thieves ambush the PCs from rooftops (using ranged weapons) and then retreating once the PCs get their act together and respond propperly. Ideally, they should also focus fire on either the caster or the healer types. This is unlikely to actually kill anyone in one encounter, but it can drain their resources and wear them down over the course of several encounters. The key thing is to never let the PCs turn it into a knock-down drag-out fight, because they will win those and they aren't strategically sound (which the Rakshasa would disdain) or profitable (which the thieves would want to avoid).

One nice thing about using the thieves guild as a cat's paw is that they provide plausible deniability. "Why are the thieves making the new hero's lives hard? Because they're thieves and heroes are bad for business." The heroes themselves might know better, but it'll be a pretty hard story to sell to the townsfolk.

Another fun thing he can do is mess with their PR. Maybe trick them into fighting and killing someone that turns out to be a well regarded member of the town (or frame them, if that's not possible, though actually getting them to do it is better). Now the PCs are criminals, the entire town will be against them and the mayor can deploy more official, and visible, forces to deal with them, possibly at the behest of the townsfolk.