Is there any (official or not) guidebook to the creation of classes, both prestige and base, that does not simply state what you must pick for a class (BaB, saves, concept, profficiences), but actually helps in balancing the class, even though I find balance a really shady aspect of 3.5.
The 3.5 DMG lists Modifying Character Classes and Creating New classes on page 174 and 175 respectively (PDF Doc).
There is also the Class Construction Engine from SHADOWCRAFT STUDIOS.
You can find more about Prestige Classes from Designing Prestige Classes - RPG.Net, and Designing Prestige Classes Part II - RPG.Net or Creating Balanced Prestige Classes - Guy Fullerton.
There’s a lot of material out there about class design, both specific to Dungeons & Dragons 3.5 and to games in general. Even stuff about non-role-playing game design can provide good insights. So reading is my #2 suggestion.
My #1 suggestion, though, is to put your work out there. There are a lot of forums dedicated to roleplaying games, and a lot of them have a lot of veteran 3.5 players, and a lot of them have dedicated homebrew forums. Homebrewing takes experience and it takes feedback. Stuff made in a vacuum is rarely good; I know this to be true even of my own work.
However, those are both really general suggestions, and probably apply to just about any endeavor out there. They remain the best suggestions anyone can give you, but some specifics about classes in 3.5 seem appropriate.
What follows is a somewhat-rambling discourse on some of the things I’ve gleaned from my time homebrewing. I have made a lot of classes, and my work was enough to get me noticed by Rule of Cool gaming, and my name now appears as a design credit in their system, Legend. None of this is in any way official, or the “One, True Way” or anything like that, but it is my best attempt to distill my skills for new ’brewers.
A Word on What This Is and What This Is Not
Because this is aimed at new homebrewers, I have tried my best to couch my personal opinions in phrases that indicate this, while things I believe to be more objectively true, I’ve tried to emphasize. Unfortunately, I am not always good at doing this. To this end, I’ve tried to explain why a certain thing is good or bad, so that you can make a judgment on whether or not the goal I’m trying to achieve with a particular piece of advice is actually worthwhile.
The other thing is, as should already be apparent, this is not a short document. This is not intended to be read from beginning to end; it’s supposed to contain a lot of specific tips on creating several different types of classes. I recommend that you approach this answer with an idea in mind, and read only the most relevant sections. I’ll be using section titles to try to indicate these.
Don’t ignore Joshua Drake’s answer. More ideas are always good.
Some of my own favorite discussions of the topic:
Some links to great homebrew, which can give good inspiration:
Most of the time, new homebrewers have a particular concept in mind when starting a class, rather than just suddenly deciding “hey, I want to make a class today!” (though I’m sure some novices do; experienced homebrewers certainly have that feeling often).
But even if you have an idea you want to make happen, most of the time you have only half the idea. That is, you typically start with either an archetype, character, or flavor in mind, or with a neat mechanical idea you want to support or extend.
Classes require both. Good design does not ignore either half, but tries to synergize them. If you only have half the concept, you need to start thinking about the other half.
If you have an archetype, think about the kinds of behavior he or she gets up to, what kinds of tools he or she needs to do them, the personality and approach that the class might want. Characters that are aggressive and up-front melee combatants need to be able to get in close, stay there, prevent enemies from retreating, and take hits in return. Sneaky characters need to move without being seen, and then they need to neutralize threats before being noticed. Spellcasters, well, they need more spells, but you need to give them something else or more that will get them to pull their heads out of the spellbook for a while – and there’s nothing they like better than becoming better spellcasters.
If, instead, you have a neat mechanic, you need to think about who would use it. Want to broaden the usage of Attacks of Opportunity, turn Combat Reflexes into a type of resource governing your actions for the round? OK, so that sounds like a very mobile warrior, a duelist, parrying blows and responding immediately to openings. Or maybe a kind of ninja, hiding in the shadows, waiting for the perfect moment to strike? There are usually a few ways you could go, so you should weigh them and consider them.
This is also a great time to start looking for feedback. Describe the archetype, and ask what mechanics others might find suitable for it, or describe the mechanic, and ask what sorts of characters might use it. Ultimately, however, the most fulfilling classes will be the ones where you come up with two perfectly-matching halves yourself.
Knowing Your Space
Quality homebrew generally requires a fair bit of system mastery. After all, you want your class to fit in with the others, you don’t want to break the game or be useless.
3.5 simultaneously makes this easy and difficult. One, system mastery in 3.5 in general is hard to achieve, because there’s just so much material, and it’s riddled with traps and unintuitive but overpowered combinations. For another thing, the balance is all over the place, and it’s furthermore heavily dependent on the system mastery of the people playing. But that does mean there’s a pretty wide range that you can fall into while still being “within” the bounds of 3.5’s balance.
System Mastery and Potential
For example, the strengths of spellcasters are not immediately obvious to everyone who plays. Many groups play for years and never notice anything problematic with them. Other groups start playing and within two levels everyone’s annoyed at the Druid doing everything. The potential of spellcasters is enormous; at mid-to-high levels of optimization, it becomes extremely difficult for non-magical martial classes (excepting the Warblade) to keep up. At high-to-very-high levels of optimization, it simply cannot be done (even the Warblade).
But the people using your class may not know that. You should, but it doesn’t mean that you should automatically make your class in line with the Wizard. In reality, most of the time, you shouldn’t – the Wizard is generally overpowered (at least in the upper half of his potential range). What you should do is understand why the Wizard is what it is, and then try to determine where – in the continuum that extends roughly from the Commoner to the Wizard (and his other “Big Five” buddies) – you want your class to fall.
A lot of this means knowing the Tiers and Why each class is in its tier. If you have a lot of play experience in your own group, that experience may or may not match the list – that’s OK, and it doesn’t mean there’s anything wrong with either your group or the list.
This is the reason for the second link. You can argue about this class or that class, but the really important thing is to understand why the classes are tiered the way they are. Those are very real phenomena, and they are important to successful homebrewing in 3.5.
The Tier system gives an indication of a general truth in 3.5: magic trumps everything. Spellcasters get far more spells than non-spellcasters get Bonus Feats, or Special Abilities, or even just class features. Individual spells can literally be licenses to “cheat,” to do things the system otherwise would not allow. Classes without magic are restricted by the usual rules, and spellcasters are not.
Magic is also frequently adjustable over time, another major advantage that shows up in the Tier list. Even if a Cleric makes awful choices with his spells, he can always wake up one morning and decide to pray for much better spells. This is inherently forgiving: mistakes can be fixed every 24 hours. Spontaneous spellcasters cannot do that, so they’re in a lower tier – but they do get an opportunity to swap out a spell every now and then, and they do get a lot of Spells Known compared to a Fighter’s Bonus Feats, so they have some room for error.
Meanwhile, the Fighter gets eleven Bonus Feats, which can only be chosen once, and never changed again thereafter (barring a forgiving DM or external assistance à la psychic reformation). He’s got to be very careful, or he’s in for a hard time.
Choosing Your Tier
Mostly, good design has goals, and in the context of 3.5, that means choosing a tier. Choosing a tier means knowing what makes a class a member of that tier, how able they are to respond to various challenges, and what mechanical features allow those responses. It’s important, in class design, to have a sense of what class features are out there – it can give you ideas, allow you to avoid reinventing the wheel, and people’s discussions of a class feature will give you great knowledge of the good and bad features of the ability.
Knowing Your Own Weight
One of the most common beginner mistakes in 3.5 class design under- or over-estimating how “big” a particular concept is. By “big,” I mean both literally, in terms of how many levels should be devoted to it, and figuratively, in terms of how narrow or broad the concept is.
Base classes are broad, and must be broad because it’s important that players can customize the class so that not every member of the class is the same. The class should have options and it should lend itself to at least a few different types of character.
Prestige classes are where you can get more specific, but the more specific you are, the fewer levels you should use. A character should not spend half his career mastering one little niche thing: that should be three or five levels. A ten-level prestige class is definitely focused on one thing, but that thing has to be flexible and allow the character to also bring other features to the table.
Base Classes – Breadth and Depth
Base classes are broad. If a concept can only really describe one sort of character, it generally should not be a base class, at least within the broader themes and traditions of Dungeons & Dragons 3.5. Base classes are far fewer in number than prestige classes, and cover numerous possible characters.
But mere breadth is not sufficient for a base class. A base class has to be reasonably deep, too, because it could potentially last 20 levels. Dungeons & Dragons 3.5 changes massively from level 1 to level 20, and a base class should ideally be competent but not broken at every level, while staying true to its own identity.
A good example is the Crusader (Tome of Battle). Crusaders are generally aligned with a god or religion, inspired by their faith and driven by their zeal. That’s a broad concept. Mechanically, they’re tough, able to take a hit, and their refreshing mechanic means they can continuously mete out divine retribution without stopping, which is exceedingly appropriate. The class does not restrict the Crusader to Lawful Good like the Paladin does, and its class features extend out across 20 levels much better than the Paladin does.
A particularly awful example is the Samurai (Complete Warrior). The Samurai is almost literally a Fighter, except that his feats are pre-chosen instead of chosen by the player. A single-classed Samurai has very few options for differentiating himself compared to other Samurai, and moreover his class features simply do not scale with level the way a twenty-level class must.
Some Things Are Too Broad, Even for a Base Class
This is a mistake Wizards themselves fell in, especially early on in 3.5. Some archetypes are just too broad, even for a base class. The quintessential examples of this are the Fighter and the Wizard, which attempt to cover nearly all “warriors” and “mages,” respectively. This is in many ways a fundamental flaw in their design (though it is, of course, not entirely without positives), that can lead to serious problems in gameplay. This is a bit tangential to the topic, but it is illustrative.
The best solution to these situations is to create multiple classes. For instance, the Wizard class is problematic because it’s entirely possible for the Wizard to take his “generic-ness” and turn it into the ability to do nearly anything. A better approach was the later Beguiler, Dread Necromancer, and Warmage classes, each of which is a spellcaster but has a particular theme that prevents them from being all things to all people.
Prestige Classes – Specialization at a Price
Prestige classes are much more focused than base classes. Many prestige classes are literally the graduates of a super-selective training program, whether it be the special forces operations of a kingdom, a secretive cabal of wizards, or an honorary order of the most courageous knights in the realm. Plenty of others specify particular required sources of knowledge – training by another of the order, or reading the ancient tome describing the class’s secrets.
Prestige classes are not required to be this specialized, however. For every Knight Phantom (members of Breland’s special forces, Five Nations), there’s a dozen Archmages, Loremasters, Horizon Walkers, etc. that certainly have a clear focus, but could be entered by quite a large number of unrelated candidates.
The ideal prestige class is one that becomes better at its forté, but a bit worse at the other things that a single-classed member of the entering base class could have done. A great example is the Malconvoker (Complete Scoundrel) – the Malconvoker loses a spellcasting level right at 1st level, no getting around it, so he’s not as good a general spellcaster as a Cleric or Wizard would otherwise be. His class features, however, make him a much better summoner than a Cleric or Wizard would be. He got a lot better at summoning, for becoming a little worse at everything else. This is an appropriate trade, and the Malconvoker balances out nicely.
The most important step in prestige class design is choosing how many levels it’s going to be. Sometimes there are thematic answers to this question: Initiate of the Sevenfold Veil (Complete Arcane) is all about the colors of the rainbow, so it has seven levels; Malconvoker (Complete Scoundrel) is all about controlling fiends, so it’s got a level for each of the Nine Hells. But usually, you don’t have this easy answer.
It’s at this point that you have to be most honest with yourself: how many levels is this concept really worth? Traditionally, prestige classes come in three, five, or ten levels. This is definitely not a requirement for your prestige class, but these are pretty good break points for different types of concepts.
Prerequisites – What Level Are We Talking?
Prerequisites primarily serve the function of gatekeeper, barring entry to a prestige class until an appropriate level. Most of the time, this is the only significant facet of prerequisites.
Usually, this is achieved by requiring a certain number of ranks in a skill. You can then qualify for the prestige class when your level is three less than those ranks. Since an enormous number of prestige classes start at 6th level, 8 ranks is a very common requirement – someone with the skill in-class cannot get 8 ranks until level 5.
Other types of requirements that set the minimum level do exist. Requiring BAB +5 prevents all but the truly dedicated warriors from getting in “on time”; Caster Level 5th or 3rd-level Spells require similar dedication from a spellcaster.
Sometimes Wizards has used other requirements that seem intended to require a certain level, but are a bit “soft,” and people can game them to get in early. For example, Sneak Attack +3d6 would require a Rogue 5 – but a Rogue 1/Scout 1/Ninja 1 (Scout and Ninja are both from Complete Adventurer) would qualify since Sneak Attack, Skirmish, and Sudden Strike all stack with one another and qualify as Sneak Attack for prerequisites. I strongly recommend that you include at least one “hard” requirement, probably a number of skill ranks, just to make sure.
The other primary purpose of prerequisites is to balance out the prestige class. Prestige classes are generally more powerful than base classes, but you “pay” for it by wasting (usually) feats to get in. This is, in my opinion, pretty poor design, since it makes a character underpowered at one level, and then overpowered at another, but it’s way too common in 3.5 to ignore.
Personally, I’d say you should try to keep this to a minimum. If you’re finding yourself trying to saddle a prestige class with a ton of bad feat prerequisites to counteract the abilities you gave it, you should tone down the class. Tying up all of a player’s feat choices railroads the player into making their character a very specific way, preventing differentiation between members of your class.
A note on system mastery: you can only do this well if you know which feats are bad. Common “feat taxes” like this include Combat Casting, Dodge, Endurance, Skill Focus, Weapon Focus, or any of the +2 to two skills feats. Feats that give small numerical bonuses to only one or two things, or even a single thing but only in certain cases, are generally bad. Players generally want feats that give them new abilties (like Natural Spell or metamagic), or power up their existing abilities in dramatic ways (like Power Attack or Rapid Shot).
Three-level Prestige Classes
Three-level prestige classes are usually very specific, but not a huge facet of a character’s identity. How could they be, with only three levels? But three levels does allow you to be extremely specific about the class. Even if it can only do one thing well, a character finishes it in three levels; doing one thing well can be worth three levels.
A great example of a three-level prestige class is the Exotic Weapon Master (Complete Warrior). The class gets to choose a weapon trick for his or her exotic weapon on each of three levels. This is about right: mastering a weapon, even an exotic one, is ordinarily a matter of a feat or two, not a bunch of levels. If you are devoting levels to this, and nothing else, you could be really narrowing your character’s scope – but with only three levels, it leaves you plenty of room to do something else. It’s a part of a character, but nothing like the whole character.
Three-level prestige classes are somewhat uncommon, so I don’t have a good example of a bad one at this point. There are plenty of examples of larger prestige classes that really should be three levels, though. It’s a shame there aren’t more.
The Hellfire Warlock (Fiendish Codex II) is a kind of problematic class that happens to be three levels long, but its problems don’t really stem from being three levels long (actually, they more stem from ways to artificially extend it to more than three).
Five-level Prestige Classes
Five levels is a quarter of a character’s potential levels; that is a big deal. But for the typical requires-5th-level prestige class, you can enter at 6, finish at 10, and have 10 levels left for a ten-level prestige class, which means it’s entirely possible that a character who has completed this class actually has another, larger prestige class that is making up a greater portion of his character identity.
This is where a lot of prestige classes ought to fall, to be honest. There are a lot of concepts that simply do not really have a place in the game after 10th level. If the Wizard’s getting the ability to force the universe to watch his back with contingency, your class should not be just halfway done mastering a particular form of wrestling; that’s something he or she should have already finished doing.
A great five-level class like this is the Abjurant Champion (Complete Mage), a full-BAB, full-spellcasting class with a few boosts to abjurations. In reality, there are a lot of players who would probably like this class to be longer, but five levels is very appropriate. One, full-BAB and full-spellcasting is unique; continuing to gain both spellcasting and full-BAB after Abjurant Champion takes some doing, and would probably be too easy if the class was longer. Furthermore, the concept of the class was as a master of the shield spell – that only goes so far, really (actually, it’s quite likely that the author thought mage armor was an Abjuration).
A rather different reason to use five levels, but equally valid, is for something like the Archmage: these are the last five levels of the game, you’re getting something good here. Five levels is very appropriate because a lot of characters will be just finishing a ten-level prestige class and looking for something to finish with. Sadly, the Archmage is probably the only good example of this sort of design.
While poor five-level prestige classes exist, like the Hierophant, the problem doesn’t usually come from their length: as a middle-of-the-road option, it’s hard for five levels to be dramatically “too few” or “too many.”
Ten-level prestige classes
These are literally half a character, and since they’re a prestige class rather than a base class, they’re usually more specific, less generic — more character-defining. These are the most common prestige classes in Wizards’ work, but I think that is a mistake. There are many things that should get ten levels, but don’t assume that your concept should just because it’s so common.
That said, there are a lot of examples of good ten-level prestige classes. For my example, I'm going to choose the Psion Uncarnate: a class for a psychic devoted to becoming pure psychic energy, untethered to the physical world. This hits a lot of checkmarks. It’s a powerful feature, to be sure; the Incorporeal subtype has a lot of potent abilities. It’s very fitting, too, for a character obsessed with the power of the mind. At 10 levels long, it’s definitely only for the committed, but that’s what it should be: this is clearly not the typical direction that manifesters go in. (Note: the class does have some serious flaws, but for the purposes of a discussion of prestige class length, it’s quite good)
Terrible ten-level classes abound, too, unfortunately. Dwarven Defender is my choice here: these are ten levels devoted to the concept of “being a dwarf.” That’s your race, you already are one. OK, it has some stuff about being good at defending caves, where dwarves live. Mechanically, it doesn’t actually accomplish that as well as it might, but that’s not the issue here. The real issue is that this is not a ten-level concept. Any character who finishes the class is, at a minimum, level 17. That’s the level that the Cleric, Wizard, and Druid get 9th-level spells, stuff like time stop and shapechange. The Dwarven Defender can take a fair bit of damage as long as he doesn’t move. The Dwarven Defender could have made a great and flavorful three-level prestige class, or a decent five-level prestige class (see the Deepstone Sentinel, Tome of Battle, for an excellent five-level Dwarven Defender replacement). As ten levels, your character becomes dominated by a really niche schtick that really is more relevant to NPCs guarding a location than adventurers trying to get in to the ruins/dungeon/temple of evil/whatever. The features of the Dwarven Defender stopped being relevant at around level 8 – which is the level this fiasco started.
You Can Be Too Specialized
This is primarily a problem with ten-level prestige classes. Ten levels is a long time. Many campaigns will not see ten separate level ups for a single character, in the whole campaign.
A concept for a ten-level prestige class must therefore be flexible enough to allow different members of the class to differentiate themselves. A common mistake with these sorts of classes is to start making it less a class, and more just a single character. Be sure that if you’re half a character’s levels, you’s giving him opportunity to make his mark on the class so he’s not the same as everyone else in it.
More than Ten Levels
In official material, prestige classes that are more than ten levels are very rare. The Necrocarnate (Magic of Incarnum) is thirteen levels, and the True Necromancer (Libris Mortis) is fourteen levels; those are the only two I can think of right now.
At any rate, this is usually a bad idea, for all the reasons why ten-level prestige classes have to be made with care – the class is now more than half a character. If a concept is really that detailed that it needs that much of a character, it really doesn’t seem to be a class at all: it seems like it’s just a specific character. It certainly wouldn’t be impossible to be an exception here, but it would be hard.
There is a category of prestige classes where the above paragraph does not apply, however. The so-called “theurge” prestige classes, named after the Mystic Theurge, that progress two different subsystems, could justify more levels as an answer to a vexing problem with the archetype: the “what now?” question that comes up at level 16 when the class is finished, and the dual-progression is no longer possible. Making the class longer is a half-decent answer to this for two reasons:
Theurge classes are pretty generic, for a prestige class. They’re simply marrying two different subsystems, usually without a lot of suggestions as to why. That makes them usually pretty safe from constraining a character too much.
Mechanically, theurge builds are typically already more-or-less “all in” on the concept anyway. These builds tend to have sacrificed quite a lot from each progression in order to have both. It only makes sense that they would be able to continue having both now that they’ve sacrificed for it.
Spellcasting Prestige Classes
If you give a Sorcerer full spellcasting, what is a Sorcerer giving up? He doesn’t get any class features other than his spells, his HD, BAB, and skills are very poor, and there’s almost nothing for him to lose (he does lose progression of his Familiar, but Familiars are frequently treated as a walking time bomb anyway, just waiting to cost you some XP).
The obvious, but unfortunately not very good, answer to this is to not give spellcasting at every level. Lose a level, probably at 1st, and now it’s a real choice, right? Unfortuantely, no, or at least not usually. From an optimization perspective, a missing spellcasting level means the prestige class is probably not worth even considering. Spells are the most powerful thing in the game. Spellcasters certainly can afford to lose some power, of course, but not a lot of players like actively nerfing themselves, and Wizards has provided them with a lot of alternatives.
There are some exceptions, of course, but justifying a lost spellcasting level is hard. The features have to be amazing. A better approach is to come up with other costs. Use the Archmage’s trick, taking spell slots to power class features. It’s sadly unique; you’d do well to make more prestige classes like that. Or ban a school, à la a specialist wizard.
So good examples of spellcasting prestige classes include the Archmage. Bad examples are everywhere; the Green Star Adept (Complete Arcane) is probably the most egregious of the partial-progression classes, demanding that you pay a lot of gold for the “privilege” of losing five spellcasting levels.
Note that Practiced Manifester (feat from Complete Psionics) and Augmentation means lost manifesting levels are not as serious as lost spellcasting levels, and moreover note that every psionic prestige class loses at least one. That’s part of the reason manifesting is much better-designed than spellcasting.
Creating a great class means having unique concepts that synergize well together and play as intended. Achieving that means knowing what’s out there, how they interact, and how powerful different options are, as well as having the humility to know the limitations of your concept. It also means knowing what you’re aiming for: what range of levels, what tier, etc. – and knowing what sorts of things fit in that target area.
All that said, it's OK to not have all that knowledge already. I only attained most of it by tinkering with my own classes and designs for a few years. And it’s also OK if you disagree with some of what I said. Not every group optimizes, and a lot of people despise the tier system. That’s OK, and it’s OK if you do too. I still suggest you take a look, at least, at the Why thread. It can give you insights even if you don’t really agree.
But most of all, and I do mean most of all, homebrew requires practice and willingness to accept feedback. Post your class, and let others comment on it: you will never know as much as everyone else combined, so try to leverage that knowledge.