One of the cool things about 5e is the refocusing they've done with feats. Feats were a mess in 3e, varying all over the board in relative power, purpose, and complexity; 4e feats may have been improved in some ways, but there were still "feat penalties" and lots of boring +1s and minor effects.

In 5e, feats seem both more powerful and more exciting. With very few exceptions, they open up new options and even new playstyles. The "specialty" packaging system is cool in itself, but it would be pointless if the feats didn't hold up too.

So when considering homebrew 5e feats and fixes to existing feats, I thought it would be worth trying to figure out what the guidelines are/should be for a good 5e feat. Here's my interpretation:

1. Feats are combat-oriented. Since feats are part of "specialties" which determine how you play your class, they're almost without exception applicable to combat situations. (The only "exception" that springs to mind is Jack of All Trades, but obviously there are plenty of ways to apply most trained skills in combat.) This is, in my mind, a GOOD thing for roleplaying: if you want to be a shopkeeper or have a bunch of followers, that's either strictly roleplaying or it's part of your background. No need to sacrifice combat prowess for the sake of in-depth roleplaying and character development.

2. Feats fit into a specialty. IMO there should not be a list of stand-alone feats. If a feat can't be formed into a cohesive specialty that defines how a certain type of character does his job, it's not specific and exciting enough to be a 5e feat.

3. Feats are not class-specific. While some feats work best for a narrow range of classes/builds, no feat should ONLY apply to a single class. There shouldn't be a fighter feat that gives you more CS dice, or a cleric feat that gives you more uses of Channel Divinity; while these things might be useful ways to distinguish a character, they belong under the tent of in-class customization (styles, domains, schemes, etc). This is as much to keep class design on track as feats: if the cleric NEEDS help to get his Channel Divinity up to par, the class shouldn't be sucking up a player's main other mode of customization (specialties) to get that done. It's a cop-out to rely on specialties to make the individual classes sufficiently flexible.

4. Any feat that affects HP or money should scale with level. Thanks in part to bounded accuracy, most abilities and effects that are powerful at first level are pretty decent at higher levels too. The big exceptions are hit points (and damage and healing), which increase drastically with level, and wealth, which does even more so.

There are two big rulebreakers here in the current playtest: Toughness and Herbalist. Both of these feats are awesome at first level but barely noticeable by 10th level, much less 20th. This is unnecessary. Toughness could be adjusted to give you 1 or 2 hp per level (along with the fixed 1d8 HD), and the maximizing of Healer's Kits could be moved over from Healer's Touch to Herbalism (perhaps with some nerfs), so that Herbalism is awesome for non-magical healers of any level, even if they don't care about dropping a few extra GP on store-packaged healer's kits and potions.

5. It should take a dedicated action to gain advantage on an attack roll. No feats that give you advantage when flanking, or when fighting goblins, or whatever. Advantage is awesome and hard to get, because with it you almost never miss - not to mention rogues getting to sneak attack every round.

6. New powers must be useful options for the appropriate build. There are three basic types of feats in the 2nd playtest: feats that give you a new action or reaction to use, feats that dramatically alter or improve an existing action or reaction, and feats that give you a generic passive bonus. The first type (e.g. the Defender feats) is my favorite, and the rule of thumb should be that the new action types will be about as good as a normal (at-will) class ability of the same action type. So if I can use a reaction to give an enemy disadvantage on an attack against my ally, as a defender, that's at least as useful to me as using that reaction for an opportunity attack. The Necromancer feats? Maybe not quite as good, but still viable in the right circumstances.

7. Passive bonuses must be interesting and character-altering. The other extreme type of feat is the purely passive bonus: Jack of All Trades, Toughness, armor/weapon proficiencies, etc. These aren't as flashy as other kinds of feats, but that can actually be a good thing for a player who doesn't want to drown in options. If I want to just focus on smashing things with my fighter, I can pick Toughness and not worry about special reactions and whatnot. But the trick is that these feats shouldn't be flimsy little +1 bonuses or other barely noticeable advantages. The biggest offender here is Two-Weapon Defense, which may serve a good purpose for the system (making a dual-wielder about the same AC as a sword-and-board fighter) but is snoozeworthy to the max. You're not doing anything differently due to that feat. Jack of All Trades is borderline, but at least a new trained skill can alter the way your character behaves (especially for rogues). Toughness is currently great at first level, completely altering the way a wizard or other low-HP character plays at low level, but as I said above this is more of a scaling issue. I would argue that +1hp and +1d8 HD at first level would still be very sweet, and you'd appreciate is much more when it's +10 HP at 10th level.

8. Feats that improve existing powers need to be carefully examined across all applicable classes. Healer's Touch and the Necromancer feats come to mind here. But I was also thinking of this when trying to design a Swashbuckler specialty for fighters using one finesse weapon with nothing in the off-hand. For a fighter, this style comes with big tradeoffs, since you're giving up 1d12+mod (greatsword damage) for 1d6+mod (rapier damage) with no big corollary benefit - so a fighter specialty that gave you +1d6 damage might seem good, right? But then, for a rogue, fighting with a finesse weapon in one hand and no off-hand is the default fighting style, so this would basically be free damage. The only solution (I can think of) would be to make the benefit something that would appeal to fighter-swashbucklers more than rogue-swashbucklers, like maybe a marking effect.

What rules would you add (or subtract)? What feats in the current playtest break these rules, and how could they be amended? And finally, what new homebrew feats would you add, using these rules as guidelines?

For starters, here's a feat from my homebrew Swashbuckler specialty:

Artful Dodge. When you fight with a finesse weapon and no off-hand weapon or shield, as a reaction you may grant an opponent disadvantage on a single melee attack against you.

"Artful Dodge" gives you a new reaction, which is useful and appropriate to a character with the swashbuckler archetype. It's useful for both fighters and rogues (and presumably rangers, etc) without overpowering any one class or build: fighters get the bigger benefit since they're more likely to get hit in combat, but a rogue with the swashbuckler archetype probably gets mixed up enough in melee to appreciate it as well.