I don't feel like restarting from scratch, as I find games like this get more fun at higher levels when you have more options.

Bleed isn't causing characters to die, but it does create an issue with healing: My cleric can't heal and mend wounds at the same time, while enemies can damage and inflict wounds. Also, she doesn't get enough turns. (If I start over or replay, I might choose the god that increases combat speed for her.)

Missing is a mechanic that can be dealt with. Have a mage inflict Freeze or Stun and physical attacks are guaranteed to hit. (Mage is preferred here becauise her spells work reliably, unlike the Bard song or physical attacks.) Alternatively, if you can inflict wounds (for example, with a spell), that makes enemies easier to hit.

As for your Elminage Gothic XP suggestion, the problem is that hex-editing is cheating and a more thorough playthrough wouldn't be worth that much more experience. Much faster is getting to where you can fight lesser demons quickly and then using them to learn the teleport spell. (Also, if I am going to cheat, I would rather cheat in an interesting way, like getting more EX skills or giving someone Breath +).

One more suggestion: If you are going to edit your save file in any game, do so on a separate file, not your main one, and back up your main save just in case.

Experience is invaluable in games with finite enemies. Personally, i always reload. It saves from more pain later on :)
Do you upgrade your equipment on regular basis? The higher defense - the less bleeding/wounding effects you'll receive. keep in mind, that in this game vast majority of upgrades comes from vendor.
On higher difficulty levels PP is a luxury. Wasting it to freeze enemy down by unnoticeable amount (that's until you reach high level of speed) is not viable at all. Someone needs to deal damage, if you will keep freezing enemies, they will never die. If you will try real mode, you will see what I mean ;) Easy mode characters have significantly higher PP pools and enemies have significantly less health. On low levels if your mage runs out of PP and all your party members start using "Attack" command, battle speed suddenly drops to a crawl.
I was quite enraged by having tons of misses with all the buffs, so I changed EXP and Gold into 9-digit numbers and overwrote. After that I quickly lost attachment to my party. And looks like I wasted valuable stat points into Constitution, and skill points - on useless spells.

Btw, according to game files, skill books can only be found on [skeleton] corpses and in bookshelves (and on floor/stands of course). Chests, boxes and barrels don't contain skillbooks.

By any chance, did you find any way to reroll vendor goods by some reliable way outside of spending out 48-72 in-game hours?

The idea isn't just to freeze enemies to deny them to act, but rather to watch the turn order. If you notice that your Arcane Soldier is going to attack right after your mage, you can use Shocking Touch (always hits) and then immediately use Flame Strike (always hits because the enemy is stunned). Alternatively, you can use Icy Touch followed by Frost Strike (to freeze the enemy even longer; Icy Touch + Flame Strike, while possible, will dispel the freeze).

Don't forget that physical attacks always hit Stunned/Frozen enemies. Use thst to your advantage if the situation is ripe.

Edit: I originally credited you with a quote I said. Oops!

Melee damage can't compete with magic damage, not until the very late game at least. So it is using 1.5 turns instead of 1 (max. 0.5 turns saved by freeze) and PP of 2 characters just to have 100% hit chance for 1 strike (and freeze opponent for a little longer). Easy mode has very weak enemies, so they die no matter what. Eventually you will want to kill demons, "imps" (they drop stat increasing herbs, so away with their riddles). On real mode you will start seeing this when cursed hounds will start popping in big packs. You will run out out of PP by freeze-flaming them very fast, and then it is up to your Cleric - if his/her PP will allow the battle to last whole eternity, or not.

At the point I was, I regretted not having enough skill points to raise Flames, probably learn Blizzard and Hammer, because I learnt all various useless Icy Touches. Don't forget that higher level spells also come with some stuns and what not. You see yourself, that skill points quickly start limiting you in what you can raise. I consider not investing into Song of Stunning on Bard, as it crippled him, and he can't shoot ANY enemy at all, while Song tends to be resisted 80%+ of time. On easy mode it would work great, but on other modes these choices made my party entirely harmless: they can't hit a thing, they have access to weakest spells only and continuing became very uncomfortable. Doable, but at a speed of snail.

Fractal of Mandelbrot? At first I thought new user joined our conversation :D

Of note:
My Mage has both touch spells (at level 1) and still has 15 skill points as of her next level up.
My Arcane Soldier has 13 skill points.
This is counting the 4 skill points gained for reaching level 9 (that I haven't spent yet), but not the 3 skill points I plan to buy because Gaulen needs them.

Currently, Stone Arrow tends to be more effective than Flames, but that may change later (but PP is starting to become an issue).

Remember, Icy/Shocking Touch cost less if you *don't* level them up. They won't last as long, but that may still be enough to hit the enemy with Power/Wounding/Flame/Envenomed Strike before it recovers. Remember, the turn order is important, and the game is even nice enough to tell you how many seconds before each character gets to act. Also, delaying an enemy so that it acts after your cleric can save lives.

I actually made that Mandelbrot Set picture by myself using Python with Numpy, Pillow (but not its mandelbrot filter), and Cython (to speed up the iterations). In case you are wondering about the coloration, the algorithm I used was as follows:
If the point does not escape, the color is 0 (black).
If it did escape in n iterations, check to see if I calculated the color for n iterations:
If Yes, use the already calculated color.
If No, choose a random color (from 0 to 2^24). Then remember that color for the next time.

You will be getting more efficient (damage per PP) spells later on. That will be the time when you'll regret raising low level spells :)

i so much want you try real mode, as we speak about a bit different worlds here. Even such lowly enemies as Askary Soldiers will quickly show you that some things simply won't work :)
World is always on move; newer and harder enemies have higher Speed, and Lv..1 Icy Touch makes little to no difference. And very speedy enemies (various wasps, ets.) are to be nuked as fast as possible. It also doesn't matter if enemy acts before or after Cleric as long as Cleric always gets at least 1 turn between all enemies' strikes. And if he doesn't, it is most likely because his Speed isn't what it should be.
I had program (I can try to find it if you are interested, unless it is on malware-infected CD, as I had problems with some malware at that time), which was written mostly on Assembler (not by me of course...) and which was drawing Fractal of Mandelbrot with very high efficiency, performance and precision.

Again, my point is that the idea isn't so much to deny enemies their turns but rather to stun them so that the next character's attack will hit. Also, sometimes only a little delay can make a difference, like I mentioned with the cleric.

My cleric is feeling a little slow to me even though I have boosted her speed at every level up.

For wasps, they don't have much HP, so Stone Arrow works well without much PP cost, and Blizzard might be useful once I learn it.

By the way, my making the Mandelbrot set is, to a large extent, a programming exercise for me to become more familiar with Python. If I ever get around to writing the game I have mentioned a few times, I plan on using Python. (Maybe write slow parts in C/Cython, but as I currently plan on using ASCII instead of real graphics, I don't expect this to be an issue.)

Cleric has one of lowest starting Speed, so it might be a factor too. Also, does your priest have weight penalty (equipped more gear than Weight Capacity allows)? As it slows down characters in many ways.
You didn't meet Venom and Giant Wasps yet? Lucky you :)
Python is high level language, which means quite poor performance, might be fine for ASCII game though. If you'll think about making graphical frontend for your game, I'd recommend looking into C++, if you are into making games - together with Monogame/XNA. I liked Ruby, but Ruby is high level language too.

I remember making ASCII game on Basic - randomly generated labyrinth with random encounters (with some VERY crude graphics for monsters - fight was on different screen). It was literally done on school wooden PC (with 64KB memory limitation for everything iirc), whicht was ancient even for that time (also it was some ussr computer and it didn't even like to properly work with 5-inch floppies to transfer data to MUCH more "modern" XT 086+...) So as far as it goes, it is quite easy and only requires knowledge of arrays and writing/reading from file.

Only problem which comes from graphical frontends, it is making actual good-looking graphical assets. Not everyone is good at drawing...

Here's the thing: Python may be slow when run, but it is easier and faster to write. (None of those pesky non-reproducible memory corruptions, for one thing.) If Python is too slow, there is likely one specific spot that is a bottleneck. That part can be optimized by writing it in C/Cython, and you can get most of the speed benefit without most of the work. For example, in my Mandelbrot Set program, I converted the routine that does the iterations to Cython (with appropriate typing) and the program runs much faster.

In summary, the idea is:
1. Write the code in Python. Python is easy to write and debug, so this shouldn't take that long.
2. Use a profiler (like Python's cProfile) to figure out where the bottlenecks are.
3. Optimize just those bottlenecks, rewriting in C/Cython if necessary.

By the way, I do know C++, but I think I prefer C.

I am reading about Python vs Ruby atm. Python seems quite enticing to learn, at least because it is much better documented. Problem is time and i can't call programming even as my hobby, just something to mess with in free time, which competes with playing actual games:)

Edit: heavy off-topic. What kind of compiler works good for Python 3,4? new version of Py2exe looks like highly dangerous and untested file for most antiviruses, so i don't risk with it. PyInstaller seems to not support Python 3+. Looks like cx_Freeze might be solution. What do you use personally?

I restarted from scratch. Nengorth guards aren't that much troublesome (though fights are so looong). Khornil guards were nigh impossible to beat, but I think with some farming for vendor gear, they might be doable.

Khornil guards don't disappear with death of Nengorth, right?

They disappear after you kill prince Khornil.

I don't (at least by the definition you seem to be using). I just use the Python interpreter in the Linux distribution's repos (Gentoo on desktop, Debian on laptop). One interesting thing is that Debian sid (maybe testing as well?) even includes packages for Python 3.5 (though importing numpy doesn't currently work for me on 3.5).

In any case, Cython *is* a compiler, and I occasionally use it, but the executables still depend on CPython (the main Python implementation) to run.

By looking at it, seems that Python 2.7.10 might be the version to use. Python 3.x seems to have severe problems with making stand-alone executables, nevermind compilation into true machine code.