I think the skill tree could use a change in the sequel. Mainly so that players can’t max out everything. Players should have to decide what they want to put their skills into. One thought I had was to adjust the skill tree to have everything under 5 categories.
The main categories would be like most RPGs where you put points from leveling up into a category. So if you have a point, you may put it into strength. From there, you can use the system in place now where you can level up sword based on experience, but the max that you can level it up is to the current amount of points you have in strength. So if you have 10 points in strength, then you can max out at 10. Points would be limited based on your level, let’s say up to 60.
Another thing is that leveling up sword should give you unlocks for sword and make you more efficient at using your sword. This would mean less stamina requirements or more accurate/harder to block swings. Experience should make you better at it, not necessarily more powerful. Strength should be the only factor of how powerful your hits are.
Agility would make maybe allow your attacks to occur more quickly, but that doesn’t necessarily mean you would want to max out agility + strength and neglect the other categories. It would bring some balance.
Another example for agility would be the bow. Leveling up your bow should make it more easy to aim and have the ability to fire shots off more quickly. A bow shouldn’t magically become more powerful just because you’ve leveled up.
It can work similarly for other items. Intelligence is another interesting one. Higher or lower intelligence should be seen in conversation. I hated that in this game you could have a high reading skill, hence you’re supposed to be fairly well learned and yet you sound like a moron at times when talking to some people. Also, give the ability to maybe learn other languages (like Latin) and things of that nature.
I should also then be able to solve any and all quests by either brute, diplomacy, stealth. Quests could branch off depending on how I decide tackle them, but then ultimately lead to a similar conclusion… or maybe not? if you want cause & effect type stuff.