“Cave Sweet Cave” post-mortem: the monster died, the game designer was born!
What went right

Playtesting.
I had the time to do two playtests. For the first one I recorded a nice animated GIF, and collected some useful about player experience, for example, my enemy sprite was seen as a child (but was not meant this way), so I replaced it with a neat soldier sprite.
Social media.
I started using Twitter, posted some warmup things. I had a good time there in development breaks, still not bothering too much. I also wrote some funny texts for my posts (“I hunger for playtesters!”).
Bosca Ceoil.
This easy to learn tool (with integrated tutorial!) made my first game music possible! I exercised with it for a couple months, so the music was good.
Unity.
I was familiar with the engine for quite a long, and for sure knew how to code for it.
Deadline.
I had to cut some corners but I was able to tweet and play others’ games while waiting for playtesters the second time. No one participated in that playtest, and I just submitted it 30 min before submission hour began and went to sleep 😛
What went wrong.

Playtesting (again).
In the first playtest some people said that they didn’t get what to do (even if it was shown in the animated GIF!). I didn’t listen to them, and realized it as a problem only post-compo… And the second playtest (some hours before deadline) was left unnoticed, zero comments.
Unity (again).
To be honest, making pixel-art games in the engine is pain! Considering to use Haxe(Flixel?) next time. I also failed to publish WebGL version of game because of the bug in Unity (confirmed by devs) which ruined the music completely.
No animations, few angles in sprites.
I had no experience in animation, and had no time to animate sprites. Also the main hero was exactly the same as I drawn him at the beginning (took about 10 minutes). And the enemy had only two sprites: alive and dead, with no angles.
Krita.
I didn’t update my tools not to break anything, but I regretted it with Krita. The version I used was a buggy one!
Few levels.
I began making levels 2-4 (total 4 in the game) at the second half of the second day. They only provide tutorials to mechanics, and not very much freedom to actually enjoy the game… I’ve returned from a business trip at last, so I think I’ll find the time to add 10+ more levels now.
Conclusion.
I had a great time participating in LD33, and playing what talented new-comers and old-timers done.
I think trying HaxeFlixel is a great idea for next time! With the tutorials it’s pretty easy to pick up, and there are many well-documented demos for game-related features.