Designing all the individual weapon cards is a huge task, I made way more cards than what ended up included in the game, trying lots of different things and then selecting down to a set that worked well together. And of course most cards went through many different iterations. Really hard to test. I gained quite a good understanding of the game by playing it a lot but I knew it would probably end up being played by people who were better at it than me and they might break it in ways I hadn't been able to. And my judgment is likely to be a bit off in some cases from having played different versions and held onto intuition from those when things change, or simply from assuming the things I did for a purpose actually succeed in fulfilling that purpose. Sometimes I really felt the lack of a team; usually games like this are made by a substantial group of people working together and I expect that really helps with thinking through the complicated problems that come up. A bunch of people helped test it and it was really useful to discuss with them but not many of them played enough to have the full perspective on high-level balance, and the ones who did still didn't have time to keep up with all my different experimental versions - so if my intuitions kept being out of date, theirs would be even more so. So it was all really hard!
Going to write about the work that went into specific weapons.
This was the first weapon I put in the game. I wanted positions on the board to matter, I wanted interactions between different weapons, here you go. It started with 11111 damage but dropped to 11110 very quickly to balance it with risk. It didn't used to have the restriction from affecting other Whetstones, I thought it was cute that you could counteract the penalty by stacking them in the same row, it was pretty ridiculous.
After Whetstone, I thought "what would combine well with that?" - something that gives extra hits to use the extra damage more often of course. First combo, still a good combo. Used to be 10/30/60/100% but I found that a hit effect at anything less than about 25% felt bad because it often doesn't occur within the space of a level.
Other triggers get to have lower percentages because they can occur more times each level. This used to have 10/20/30/40/50%; starting already at level 0; but it was pointed out that every single other percentage-based item started at level 1 so they all had floating 0%s in front of them which made things less clear.
I wanted to see if I could make an interesting and useful weapon without needing any extra rules on it, just damage. It's weird for a weapon to do 4 damage because it doesn't even matter what colour it is, it just kills everything in one hit, so that needs some kind of extra disadvantage to balance it out. 3 is really strong but it's okay because there's always an enemy type you still take hits from. I tried a damage profile of 11223 but it felt so easy to level up, 11123 presented a nicer challenge. Establishing 3 damage as a baseline with no abilities limits everything else to 2 but that's okay. I was worried nobody would play with this because it's less straightforward than cards with rules written on them to tell you what to do with them, so I'm pleased to see that it has shown up in some winning boards.
Contrasting Cleavers slow-growing damage, I wanted a clear early-game weapon that peaked early and stayed there. For a while this was another no-rules weapon, just 12222 damage, but I ended up wanting every weapon to gain something at IV, it just felt wrong to have those levels there if they didn't do anything. Technically runes solved this problem but there was still a sense that you ought to get a thing, it feels fine for some of the intermediate levels to only give a rune but it never feels like enough for the final level. So I put "stun all enemies" which I thought was minor enough; turns out with the possibility of stun combos it's not minor at all so I cut it to "stun random" and then "stun nearest".
4 damage with a penalty. It turned out really weak so I made it also start at 2 damage so it has some early-game utility too. Tried it with a few extra curse-based abilities but it didn't need them. Used to be called "Witchpact Broadsword"; changed the name because it was a source of confusion that the witch hero couldn't effectively use it, also because it was slightly too long and was overlapping the card borders. Jeska's disadvantage eventually changed because it was still confusing so now she can use it but I guess the name change stuck.
Again, 4 damage with a penalty. I tried a bunch of different effects on it, things like "when this levels up, kill all enemies", but eventually decided it was overcomplicating what could be another nice distinctive weapon with needing any text. Some people hate it; I constantly had testers tell me it seemed really bad because 0. Tried ending it at 1 but that's just way uglier because two levels have to be the same. Now it's showing up in winning boards so I think it's okay.
I tried things that gained range as they leveled up but it felt quite weird because it's not a strict advantage (in that occasionally you're forced to hit with it when you'd rather move). As things level you expect them to get better, it's confusing to find that something has gotten sometimes worse, whereas it feels fine if that disadvantage was built in from the start. But in the situations where range is advantageous it can be quite strong indeed, so it's unsafe to pair with other strong abilities. Giving them damage is especially strong: 2 damage at range is almost like 4 damage since you get to hit twice without getting hit back (though it does cost an extra turn). This considerably limits the design of ranged weapons.
Like range, spending mana is a disadvantage so it feels bad to gain it with a levelup. So most mana weapons follow the same basic pattern: a spend at level 0 and then something at 4 to justify leveling up. This is the first blue weapon encountered even though spending mana feels pretty complex because it's such a standard thing that blue weapons do that I wanted to make it apparent straight away. Plus all the alternatives are complicated too in their own ways.
Paying to curse turns out quite weak because you're putting yourself at risk to set up a fragile combo instead of just doing damage; doing it at range is better so you at least don't get hit as well as paying. Meanwhile I wanted to distinguish blue from red; this establishes that blue weapons only get to attack at range at a cost, you have to pay the cost even if you don't care about the curse because really what you're paying for is blue range. You can get out of paying it if you or the enemy only has 1 mana, nice. By the reasoning above, both the range ability and the mana ability start at level 0, and then level 4 has to have something too.
Wanted to characterise red vs blue weapons, red weapons do more damage blue ones do more stuff, red weapons curse on kills blue ones curse with weird triggers. Turns out I had more ideas for weird triggers so this ended up being the only red curser but it fills that role fine. As usual I wanted something at IV but I wanted damage and cursing early on so I just gave it a second copy of the effect so as not to complicate things by adding something different. It used to have the same text written twice but I changed it to "repeat the previous effect" to make it more succinct and also open up the possibility of a future weapon repeating things in a more general way.
This was a stun effect for a long time, started as "Crippling Poison: when a cursed enemy moves, % chance to stun them". Was cool on its own but made it way too easy to trigger stun combos (especially with multiple copies in play) so I moved it to stun on attack which was better because they'd at least hit you once before freezing forever. Still too effective, wearing down an enemy that's four times stunned is boring, so I removed the combo potential and went with a simple miss chance. It can still lead to slow grindy endgames where enemies only rarely hurt you but at least it's not multiplied by other things that do that.
The idea was to have just a straightforward two-part combo: curse and then kill. First version was simple, "when this hits a cursed enemy, kill them". That seemed a little strong so I added a mana cost (before I made the rule that only blue weapons do that), then it was a percentage chance, or only at certain times of day. Then at some point I decided for reasons it would be best not to have any direct kill effects and went through and replaced them all with extra damage (see also Slingshot), but I already had an effect of "extra damage to cursed enemies" so I tried this version instead. Then as usual it needed something more at IV, and I was in need of more ways to attack multiple enemies and keep up with increasing spawn rates, so this became a thing.
Fairly late addition when I had too many problems with another effect and needed to replace it (it might come back one day). I wanted a bit more variety of curse interactions, different boards you could build around them. Witchpact Blade had the first effect at one point, the second effect gives a possibility to chain it even if your initial source of curses is unreliable. I've generally avoided stacking multiple kinds of costs together (e.g. mana spend, percentage chance, curse requirement) because each one cuts down the utility, but I think this is a big enough effect to merit it. Still it has been fairly unpopular on the scoreboards, I'm not sure if people are undervaluing it or if it's still a little weak; in principle once you have some cursers going it should be very turn-efficient. This initially had the "spend mana to damage cursed enemies" ability at level 1, which I said to avoid under Rimeclaws because it's introducing a disadvantage. I thought it was okay here because the curse requirement makes it a specific situation that you have to set up, but now I've patched it back to level 0.