Skip to content

mxcl/game-studio-simulator

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

hero

You have a game idea.

We have the agents.

Prompt it, watch it come to life, iterate like crazy, and actually
finish something for once in your pathetic, human life.

Prompt. Play. Publish.

Game Studio Simulator (GSS) is an AI-powered game development studio in a single app. You direct; specialized agents execute, and you play the result in a live, playable canvas. The editor is the game.

https://gamestudiosimulator.com

Project Status

Pre-release.

Game Studio Simulator (GSS)

Game Studio Simulator is an AI-directed system for constructing games through modular agents. A Director Claw coordinates specialized skills that generate well-defined modules which can be composed into a working game.

The initial target is 2D platformers, but the architecture is designed to expand to other genres.

The system emphasizes:

  • strict module boundaries
  • explicit APIs between components
  • event-driven composition
  • agent-generated artifacts that remain understandable and testable

Architecture Overview

Core Concept

GSS operates through a Director model.

A single orchestrating agent, the Claw, plans the structure of the game and invokes specialized skills to generate the components required to build it.

The Director never writes code directly. Instead it assembles independently generated modules.

User Prompt
     ↓
Director Claw
     ↓
Skill Invocation
     ↓
Module Generation
     ↓
Dependency Resolution
     ↓
Game Assembly

Module System

Every artifact produced by the system is packaged as a Module.

Modules are self-contained units that expose a strict API and declare their dependencies.

module/
  module.json
  code/
  assets/
  tests/
  readme.md

The module.json file defines the contract the Director uses to compose systems.

Example:

{
  "name": "double_jump",
  "type": "mechanic",
  "version": "1.0",
  "api": {
    "events_in": ["player_jump"],
    "events_out": ["double_jump_triggered"],
    "state": ["remaining_jumps"]
  },
  "dependencies": ["player_controller"]
}

Modules are designed to be:

  • composable
  • inspectable
  • replaceable
  • testable

Event System

All modules communicate through events.

Modules do not call each other directly. Instead they emit and subscribe to events in the game runtime.

Input → Entity → Mechanic → Physics → Render

Example:

input_jump
   ↓
player_controller
   ↓
player_jump
   ↓
wall_jump_module
   ↓
velocity_change
   ↓
physics_system

This architecture allows the Director to rearrange systems without rewriting code.


Skill System

Skills are specialized generators responsible for producing modules.

Each skill focuses on a narrow domain.

Typical skills include:

Mechanic Designer

Creates gameplay mechanics such as abilities, enemy behaviors, and interactive objects.

Entity Architect

Defines game entities such as players, enemies, collectibles, and world objects.

Physics Generator

Produces motion and collision systems appropriate for the platformer environment.

Level Generator

Constructs level layouts, spawn points, triggers, and progression structures.

Asset Generator

Creates sprites, animation sheets, tilesets, and UI elements.

QA Simulation

Runs automated simulations to validate mechanics, difficulty, and playability.


Director Claw

The Director Claw is responsible for orchestrating the system.

Responsibilities include:

  • interpreting the user prompt
  • planning the module graph
  • invoking skills
  • validating module APIs
  • resolving dependencies
  • assembling the runtime configuration

The Director operates on a game plan graph rather than raw code.

GamePlan
 ├─ Modules
 ├─ Event Graph
 ├─ Asset Registry
 └─ Level Layout

Validation

Before modules are accepted into the game plan they are validated.

Validation checks include:

  • API compatibility
  • dependency resolution
  • event schema correctness
  • test presence
  • runtime integration safety

This allows the Director to reason about modules without needing to interpret the code itself.


Design Goals

GSS is designed to enable:

  • agent-driven game construction
  • composable gameplay systems
  • transparent architecture
  • iterative AI collaboration
  • extensible genre support

The system favors many small modules over large monolithic codebases, making it suitable for automated generation and modification.

About

You have the game idea. We have the agents.

Resources

Stars

Watchers

Forks

Releases

No releases published

Contributors