Note: This is a modified repost. It was initially posted on Cohost on August 24th.

The second devlog is posted! This one dives into one of the core design concepts for Arborshire: procedurally generated puzzle/navigation dungeons. I am hoping to avoid combat entirely, though I'm still working out ways to make that possible.

The primary challenge with this (besides the technical details of the procgen itself) is what "ends" an attempt in a procgen puzzle? In a combat-focused roguelike, the player dies to enemies and has to try again, get better, and go further.

I'm exploring a few different options for Arborshire.

• A timer. I won't do this, because it just replaces combat with another source of anxiety. But, it's a very simple solution and worth mentioning.
• End when the player reaches a dead-end in the dungeon--also rewarding a narrative item.
• End when the player softlocks or fails to solve a puzzle. Most puzzles are designed to not get stuck, or reset on a whim. In Arborshire, resetting could mean resetting the entire dungeon.
• End when the player runs out of keys. There could be a limited number of keys for locked doors.
• End when the player explores some number of rooms. This also lends the idea of exchanging keys for extra rooms.

All of the above rely on one more core concept: knowledge keys. This refers to game knowledge that allows the player to take shortcuts and bypass locks on future dungeon attempts. The idea is every attempt the player would learn something new about the game mechanics that allows them to go deeper on the next attempt.

I will be developing a few prototypes to test these ideas with handcrafted dungeons first. When I'm confident in the design I can start testing ways to procedurally generate it!

[By the way, there is a handcrafted prototype available! You can try it if you join my Discord server.]