Standard RPG item effect, this was in there right from the earliest concepts. Now everything had to be a weapon and level up, so just give it a basic damage profile and withhold the hp bonus until it's leveled. There was going to be a corresponding blue item but it seemed too straightforwardly symmetric so I cut it, that effect might show up somewhere else?
It's a really subtle effect because it's not useful at all if damage and healing are evenly distributed; you have to have clusters of healing you can save up until you get a cluster of damage; how true that is depends a lot on your particular strategy. But I figured a beginner probably won't worry about that yet so it's probably not too complex to include in the initial weapon set (it helps that the first character is Harry with a power that does exactly what you need to make this useful).
Again going through mental lists of standard RPG effects to steal ideas; life drain is obvious. Restoring on hit is too much when enemies only do 1 damage so it happens on kill. At first this had "when this levels up, restore ♢" but then I went through a campaign of adding minor curse effects to help give reasons for a variety of strategies to consider including some cursing instead of it being a clear-cut matter of "include the curse things if and only if you're building the curse board", and so it became "when this kills a cursed enemy, restore ♢". (Most of those minor curse effects added too much complexity so they didn't stay, instead curses got the default ghost effect so they weren't entirely dependent on other weapons to have an effect.) Being able to heal both colours led to stalemates where you could stay alive on a spear constantly healing by killing auto-cursed enemies, but never move anywhere; having it be red-only sadly makes the curse bonus less special but gives a critical weakness to ensure that if you can't advance from a position then you'll die in it.
My initial idea of leveling up was that some weapons would grow when they kill enemies of a specific type; this idea was a way to circumvent those restrictions. That became less of a thing when I made most weapons get experience from any kill (but is still present with Blight Broadsword) but the effect still seemed useful as a way to level your board faster. Adjacency originally wrapped around so if you placed this on an edge tile it would affect tiles on the opposite edge; this seemed arbitrary and unclear so I settled for the simpler version where those are just poor positions for it. I tried them feeding everything else in the same row and column so it wouldn't matter, but that was ridiculously fast.
Most players seem to go through a phase of thinking these are essential and overpowered followed by one of thinking they're worthless trash. Hopefully they're somewhere in between?
In my early notes the idea was a pair of weapons that let you teleport between them, maybe to a random one if there's more than two copies, but a simple random teleport was easier to implement so I tried it first and stuck. Probably better this way, it gives more board flexibility, I'd rather not have complicated rules about what has to be included together.
As with range and mana costs, a random teleport on hit isn't a strict advantage so I wanted it there from level 0. I felt bad teleporting next to an enemy and getting hit so I wanted to let you take another turn after teleporting, to have a chance to react to the randomness before it hurt you; implemented this with a "stun all enemies" hit effect - thought it was cute that this would also trigger any stun combos but that ended up pretty broken. (Stuns were overpowered in general for a while, they tended towards stalemates where you could stop enemies hitting you but they could still stop you getting anywhere, I've had to make them much less prevalent.) It was okay but later I needed an IV effect so I added in target selection and it sort of became superfluous - if you've had a say in the random selection there's no need to give an extra chance to respond to it. The IV effect let you select any tile to travel to but that was too much of a departure from the rest of the game where you're always choosing between a small number of options, so I cut it smaller but made it start at an earlier level. I like that this means it's still always a gamble to use it, its character doesn't suddenly change once's it's leveled up (I've seen someone use it on 1 life with just one enemy on the board and be offered a choice between each of the four squares next to that enemy); that risk ended up making it feel a little weak though so I gave it extra damage on IV to prop it up (part of a general campaign of giving red weapons more damage to characterise the colours).
The target selection is an exception to the control scheme of the rest of the game and goes against my philosophy of not adding extra controls and separate input modes. It would have been more consistent with the rest of the game to have stuck with just teleport + stun or something. But it just felt like such a natural way to level up the effect, and it is an interesting decision without too much extra complexity. It sort of interrupts the flow of the game, but a big random effect makes a pause in that anyway so it's the least disruptive place to have a mode-shift. Plus I've had in mind that if I expand the game with more effects then this might be an interesting direction to explore further; and if I've already introduced the concept that will feel like more of a natural growth than adding messy new things for the sake of it.
I like the symmetry with Blink Dagger. For a long time this also had "when this levels up, teleport all enemies to random tiles", I liked the idea of stirring the board all up with a whisk, but it was messy and didn't add much.
In retrospect I'm not sure about the power balance between this and Blink Dagger; Dagger is also usable as a scoring and healing mechanism (by possibly skipping closer to the gem) while Whisk's teleport is generally safer because it doesn't leave your hero exposed - different strengths, but overall I think Dagger gets the better deal with having limited control as well. I felt okay with this when Whisk also had more damage, but I didn't end up re-evaluating it after increasing Dagger's damage (which happened very late in development). But it's still definitely distinct and useful, I don't have any changes in mind right now.
Amulet of Immortals
Before I had the concept of leveling up the idea was that this would trigger if you died in that spot, and it would only take effect once (being replaced with ashes). But then once there were levels it was simpler to just reset those than to transform the item, I was open to doing that if being able to reuse it proved too strong but it turned out fine.
At some point someone objected to so many weapons having their main effect at IV rather than being more accessible so I tried this at III, with the IV effect being to also curse everything on respawn. For a while it had extra damage too. Those versions all turned out too strong so I went back to the simplest one.
Through most of development the effect was "IV: Remove all ♢ costs". It was pretty tough to design around because I had to make sure everything with a cost didn't break the game when it was removed. That's still an issue with this version but you have to set up multiple copies so you're at least paying more for the big effect, and you don't get 100% reliability (even if on average you expect a profit), plus now you do have to have the mana available to spend even if it does get paid right back. Having it as a triggered effect feels better in play too because it actively connects the Ring with its bonus every time rather than it just passively rewriting things once. I'd been testing a bunch of other rewriting effects too but they all had the same passivity problem and once I saw how much of an improvement the new Ring was I started trying to find similar ways of transforming them instead. Whetstone is okay though because its effect is visible on the board, not just in the card text.
The old Ring made stalemates when combined with this - you could kill anything one-on-one without it hitting back, but it takes a long time and you don't get to go anywhere. I tried making it only affect red enemies to break the stalemate - unringed it's the same cost whether a blue enemy hits you or you pay to stun it. When I changed the Ring I put this back to the simpler version. You can still lock things down for a while but something always has a chance of getting through.
I like that this mostly translates into a straightforward resource exchange, pay a blue hit to prevent a red hit, it could almost read "spend ♢ to gain ♡".
After Witchpact Blade started counting cursed enemies I got interested in what else I could use similar counts for, so this started with "0: when an enemy moves onto this stun them for 1 turn for each cursed enemy". But long stuns were bogging the game down so was cutting down on those, and also needed to cut back on curse/stun connections in general - I'd been putting too many effects like "when this hits a cursed enemy stun them" because they were easy to make but they weren't really that interesting. So it went into the simpler form, simple and effective. It had a weird delaying effect at a higher level too but it was complicated, broken, and unnecessary.
Hasn't changed much since I first added the concept of ghosts, the only question was whether to have the damage or the ghost effect come first. With both this and Caltrops it seemed more interesting to access the weapon's unique effect sooner, rather than just damage damage more damage and then only something special at level 4.
I think it's easy to underestimate the value of targeting enemies of the same kind because it doesn't happen much early on, but late-game it's very likely you'll be able to hit several. And even early on, a lot of the toughest situations are when the same enemy spawns a few times in a row - so while the times when this effect pays off are rare, they're also the times you most need it. On the other hand, when it's a flood of Serpents you're up against it's not ideal because you might not have the mana to spend. Mirror was very strong under the old Ring where it basically became an ultra-Harp.
First version of this was a percentage chance to kill, the idea being to have a way of taking down an enemy that was otherwise too big to deal with right now. I never tied down the right numbers, 50% seemed like too much but still compared unfavorably to Crossbow which gets a definite kill in two hits rather than maybe in one, maybe in four. I decided percentages were more a blue thing anyway.
Tried "when this hits a minotaur, kill them" but what level does that go at? At IV it's too weak so it needs something else, I kept trying weird additional effects. One was cursing when it kills minotaurs; that eventually just became a property of minotaurs themselves. When I dropped "kill" effects I realised the damage could scale, so the kill effect is at IV but it's already helping you fight them sooner. Wasn't sure it was strong enough relative to Crossbow so for a while I had it affecting all blue enemies; broken. Ranged weapons are hard. I'm not too bothered if it's a little underpowered; it's a specific counter to a thing that might cause you problems and you can ignore it if it's not.
I tried to structure the unlocking items in a way that makes sense, maybe you see a problem first and then you find the solution later, maybe you see one piece of a combo and then find the right thing to complete it. At first I had this unlocking with Dominic because obviously you want to use it to constrain serpent spawns but I thought it would be nice to put it on the next character so that by the time you get it you really know why you might want to use it.
Dice of Omens
I just went through a list of every possible way I could think of to apply curses, something that does it automatically over time seems straightforward. The percentages are ridiculous, there's no reason other than sheer numerology to expect a 3% chance at level 3 to be the right balance. But with low integers there's not much I can do, it has to go up and there you go. If a 50% chance isn't quite enough I can push it up to 55% or whatever, but this would look awful if I went into fractions (1.5%/2.75%/3.875%/4.96875% yes).
Gave them extra damage because people were being put off by the low percentages, plus I generally wanted to strengthen cursers a bit to encourage experimenting with combos rather than sticking with safe obvious things. Probably unnecessary in retrospect.