For your timeline:
"Sight" is a lot harder than it looks. Also, any monster AI beyond the very simplest stuff is not really represented before pets. I'd insert a step between steps 4 and 5. Let's call this 4.5th step "environmental awareness": Sight for players and monsters, the ability of monsters to figure out where they want to stand and where they don't want to stand with respect to choke points and ranged abilities, monsters getting out of each others' way, monsters wandering around until they see you and then pursuing you, monsters knowing whether they can break down doors, etc.
I say to put this stuff here because you CAN make a playable game with full visibility to the entire world at all times, and you can have monsters that always know where you are (maybe when you're within a certain range), and can always beeline to you using A*. You don't need LOS or smart monsters in step 1. However, you'll want monsters to have some basic environmental awareness before you worry about pets.
That's my only real comment there. Don't get caught up on LOS and AI early on, they can be discouraging. Put them off.
Oh yeah, and random dungeons are hard too, don't worry about making them remotely great in step 1. If they are enough to test your game, then they are enough.
Where are items/equipment/inventory? I don't see weapons/armor. Put a basic system in step 1, it's important. Set some time aside to figure out exactly what you want your system to do and not do, and for the love of god, don't overcomplicate things--set your limits LOW. Remember, Nethack's and ADOM's item systems are NOT very robust in the end--Nethack just has a bunch of special cases. Don't try to let the player wield live goblins whose spiky armor gives a bonus to damage. Down that way lies madness. Trust me, I know.