Dev Endeavors: The Story of Dicey Dungeon
From Dice Rolls to Web App
So, it started during a Shadowdark solo session. I was using Axebane's Deck of Many Dungeons, and it was working great, but I kept hitting the same limitation — you only get what's printed on each card. Some have room descriptions, some don't. Some have monsters or quest ideas, some have a treasure roll-table instead. After a while, you start seeing the same room combinations and you still have to fill in the gaps yourself. And laying on another supplement on top to give each dungeon room more depth was slow and didn't feel right. I kept coming back to the same wish...
"What if one roll could generate an entire dungeon room?"
I knew what I really wished I could use — all 7 of the standard set of dice every D&D player has. There's something satisfying about rolling them all at once, watching them scatter across the table — but we never really get to do it! If I could have each die control a different aspect of the room I'd be pretty happy.
"Why not just use existing generators?"
I tried — I looked on DriveThruRPG and searched online for something similar to what i wanted, but didn't find any that used all 7 dice, if i missed any please let me know. Also, the supplements I own have you flipping between pages of roll tables or turning multiple cards or roll specialized dice, and that was still tedious and really stopped the game flow. I wanted something more dynamic. So, I started off with the the d100 and d10. They could handle dimensions (room width & length), and maybe the d4 for hallway length before a given room. I just needed to define the remaining dice rolls and I'd have it — One roll, one complete room. My eyes lit up when i realized it might be possible. At least that's what it felt like — I'm not sure, I was alone. But then I really had to ask myself...
"Wait, am I just making this more complicated again?"
That realization hit when I was trying to organize everything in the Obsidian note I had and needed to add a table (not a strong feature of Obsidian). Seven dice meant potentially referencing seven different roll tables for every single room. I loved the physical act of rolling all the dice at once, but the documentation was becoming exactly what I was trying to avoid. That's when the idea of a web app started forming. What if I could have both? Something that let me roll physical dice and quickly reference all tables in one place, or just click a button and get instant results?
I tried moving it to a google spreadsheet, thinking that might help organize things, but I was spending more time manually updating the tables when my ideas would change. I started calculating just how many possible room combinations existed — I knew I needed a better solution, and maybe a whole new approach.
I had a friend who'd been using Claude Projects to help him author and manage his game dev documentation and Unity code in Github. He suggested I do something similar. That really got me thinking about organizing the project better.
"So, how exactly did this evolve into a webapp?"
Originally, I just wanted help just documenting and expanding the rules I already had — something similar to Four Against Darkness, where you have a clear process and endpoint for dungeon generation. I took my Obsidian doc (already dubbed "Dicey Dungeon" because who doesn't love a pun?) and shared it with ChatGPT since I already had an account. I'd been using Obsidian with markdown to document everything up until then, but I wanted something more shareable that could integrate tables easier. When I asked ChatGPT to format the document output in HTML it came back with an interesting suggestion: "Since we're using HTML anyway, want to add a roll button for the dice?" I laughed - I didn't really consider that. But sure enough, within minutes it had a preview of the document with a simple dice roller at the top. Could this end up being more than just a document — could it become a solo-rpg tool?
"So, you gave up writing the instructions and made the webapp?"
Oh no, I still completed the instructions and continued to update them as the project progressed. But, over the next week, as I referenced it alongside my analog play-testing, I started seeing more possibilities. The first feature I had added was a way to customize the roll tables. After all, that need was what caused the pivot in the first place. Also, not everyone wants the same results for their dungeons. Next came the room visualization (html5 canvas) — because being told the result is a 4 x 6 room with 2 exits/doors is nice, but seeing it visually represented was a lot easier to immediately draw onto your grid paper. Each time I thought "this would be nice to have," I'd describe the feature to Claude or GPT, and it would help me implement it.
"How much back and forth did it take to get to the current version?"
A lot, honestly — from adding local storage so your customizations stick around, to building export features for those marathon dungeon-building sessions. When I wanted to make the rooms look better, and added a theme system last. Each feature started as a casual "wouldn't it be cool if..." conversation with an AI, and somehow turned into working code.
What started as simple dice rolls turned into this whole digital toolset, that still keeps that core idea: grab your dice and roll up a dungeon. If you're curious about this dice-rolling, AI-assisted development, you can check it out live version below or dig through the repo. Maybe even add your own chaos to the mix — it's open source and just for fun, after all. Thanks for checking out this project! If you’d like to give Dicey Dungeon a try, click the link below.
Stay nerdy,
Karuu