EP Game Prototype — a browser videogame for the Escalation Protocol universe
Can I build a playable videogame prototype in the browser — squad combat with an escalating tension system, set in the Escalation Protocol universe — by directing AI as a non-programmer?
Building a playable game (not a paper model) is where balance problems actually show themselves. Running the outcome resolution as a Monte Carlo simulation — playing a clash out thousands of times — turns the vague question 'does this feel fair?' into measurable data. The squad-combat loop with an escalating tension meter (a nod to Ogre Battle's Chaos Frame) is the heart of it, and the part most worth tuning.
It's a playable prototype: the core loop runs. Parked at the point where the next moves are content and polish rather than proving the concept. Hard rule on this one: no third-party art, ever — in the repo or anywhere public.
What it is
A playable videogame prototype, running in the browser, set in the Escalation Protocol universe. The combat is squad-based — units resolving against each other in formations — with mechanics inspired by the classic strategy game Ogre Battle.
The mechanics worth tuning
- Squad combat — the moment-to-moment loop of units clashing.
- An escalation system — a tension meter that ramps the stakes as a battle unfolds, in the spirit of Ogre Battle's Chaos Frame.
- Outcome resolution (AREPO) — balanced with a Monte Carlo simulation: the computer plays a given encounter thousands of times and reports the spread of results, so I can tune by evidence instead of gut feel.
Why I built it
It's the most direct test of the idea that I can direct AI to build a real, interactive game as a non-programmer. The thinking is mine — the universe, the mechanics, the feel I'm after — and the build comes out of my multi-agent workspace. It's also a through-line to my origin: my thesis was about prototyping games to prove a concept. Same instinct, new tools.
The layer above the battle
The newest piece isn't combat at all — it's where the battles are chosen from.
There's now a 3D globe you deploy against: a rotating body rendered in the browser, using real lunar elevation data published by NASA for the surface. Missions are picked off the sphere. It's the same instinct as the mission-selection screens of the strategy games I grew up with, except the terrain isn't invented — it's the actual Moon, which is exactly the right note for a NASApunk universe. Nothing about the fiction has to pretend.
The globe library is vendored — a copy of it lives in the project rather than being pulled from someone else's server at page load. Slightly more disk, no dependency on a CDN staying up, and it works offline. For something I want to still open in five years, that trade is obvious.
Alongside it, the mission map got rebuilt to behave the way Ogre Battle's overworld actually behaved — popups, an area command layer, and threat that reflects real state rather than decoration. That fidelity came out of a separate study project, which is where the mechanics are being reverse-engineered properly.
What's still unsolved
Combat timing. The moment-to-moment rhythm of a clash — how long a round takes, when the information lands, where the pause falls — is the thing I can't derive from a rules document. It's felt, not specified. So it's parked on purpose while I study how the original did it, rather than guessing and calling it design.
Where it stands
The core loop is playable, the strategic layer above it now exists, and the open question is timing rather than concept. The strict asset policy stays in force: no third-party sprites or art, ever — not in the repo, not in anything public. The prototype stands on its own mechanics and visuals.