Wizard Waves
A wave-survival game for iPhone where a voxel wizard casts elemental spells by drawing glyphs with your thumb.
Abandoned Side Project
Wizard Waves was a wave-survival game for iPhone, starring a lone voxel wizard in an open arena, with enemy mages pouring in from every side. Your left thumb moves the wizard, and your right thumb draws glyphs to cast spells: a triangle for a firebolt, a chevron for an ice lance, a wavy line for a water jet, and so on. Fire beats ice, ice beats water, and water beats fire. Survive 3 waves, beat the archmage at the end of the area, and you pick up a new spell before moving on.
Draw a glyph, hold it overhead, flick it at someone
This project was mostly an experiment in process. Before any code was written, I put together a detailed plan, a game design doc, and a set of ground rules for the AI agents doing the work, including a definition of "done" for every milestone. Then I let a team of agents loose on it. They split each milestone into slices, built them in parallel, merged the results, and ran reviews against the spec before moving on.
Chain Frost jumping between enemies
The first build worked surprisingly well. In about a day, the agents went from an empty folder to a game that was playable end to end: waves of enemies, boss fights, 14 spells (including a few secret ones you can only find by experimenting), potions, a spellbook, and sound effects and music synthesized entirely in code. No third-party libraries, and hundreds of automated tests.
Every area ends with an archmage
Once I deployed it to my phone, things got complicated. Although everything technically worked, it didn't feel right. The glyph recognizer was way too rigid, so I made it more forgiving and tied drawing accuracy to the spell's power. Aiming was the bigger problem: over about 10 days I tried auto-aim, drawing glyphs over enemies to cast in their direction, dragging and flicking, and finally holding each spell as a glowing ball over the wizard's head until you flick it at something. None of them quite clicked the way I wanted, and the game just wasn't fun.
The last casting scheme I tried: the spell hangs overhead until you flick it
Ultimately, my takeaway from this project is that sometimes ideas are more fun in your head than in your hands. Agents can build something astonishingly fast, but they can't tell you whether it's fun. For that, you have to actually use it. Maybe one day I'll return to the gesture-based spell concept, but for now the wizard is going on an extended vacation.