Not just a prompt collection — a structured operating system for game projects that combines:

• Rules for policy and standards
• Agents for role specialization
• Commands for repeatable entry points
• Skills for reusable execution patterns
• Contexts for phase-specific behavior
• Hooks for workflow automation
• Harness adapters for Claude, Codex, Cursor, OpenCode, and Kiro

Quickstart

git clone https://github.com/MRCalderon3D/everything-game-dev-code.git
cd everything-game-dev-code

Open the folder in your AI coding assistant (Claude Code, Cursor, Codex, OpenCode, or Kiro). The scaffold is loaded automatically from CLAUDE.md and rules/.

Then type commands in the chat:

You don't have to follow a specific order. Pick whatever command fits your current need — start a new project, generate a GDD for an existing one, run a QA review, or fix a build error.

Step-by-step guide

The guides/Dash & Collect/ folder contains a full tutorial that walks through building a game using the scaffold's commands, agents, skills, and contexts across all project phases.

Example project

The PirateInvaders/ folder contains a complete HTML game built with the /full-game command in a single pass. For real projects, we recommend going step by step.

Goals

• Keep shared game-development standards engine-neutral.
• Let Unity, Unreal, and Godot each extend the base cleanly without contaminating one another.
• Support real production work across design, engineering, content, QA, release, and live ops.
• Turn repeated solutions into reusable skills and structured workflows.
• Make the repository portable across multiple coding assistants and harnesses.

Repository Model

This scaffold is organized in layers:

• rules/ — what good looks like
• agents/ — who does the work
• commands/ — how work starts
• skills/ — how work is executed well
• contexts/ — how priorities shift by phase
• hooks/ — how workflow safeguards are enforced
• manifests/ — how subsets are installed by profile
• schemas/ — JSON validation for manifests, hooks, and plugins
• docs/templates/ — structured templates for GDD, TDD, QA plans, and other deliverables
• docs/orchestration/ — agent routing, role handoffs, and workflow sequences
• tests/ — how the scaffold verifies itself
• harness adapters — how different AI clients consume the same source of truth

Engine Isolation Policy

The repository is intentionally split into:

• rules/common/
• rules/unity/
• rules/unreal/
• rules/godot/

And equivalent skill / command / review layers where needed.

Shared documents should describe intent, ownership, and quality bars. Engine-specific files should describe implementation conventions inside that engine only.

Intended Use Cases

• New game project setup
• Multi-engine studio workflows
• Internal AI workflow standardization
• GDD and technical design maintenance
• QA and release readiness reviews
• Plugin / content / tooling governance
• Cross-discipline planning and orchestration

Supported Harnesses

• Claude Code
• Codex
• Cursor
• OpenCode
• Kiro

Each harness adapter points back to the same shared scaffold rather than becoming a second source of truth.

Current Status

The scaffold is intentionally modular. Different blocks may be added or replaced over time, but the repository should always preserve:

• flat agent and command structures
• layered rules
• grouped skills
• engine isolation
• harness portability

Principles

• Design before implementation
• Explicit ownership over implicit assumptions
• Testability over cleverness
• Documentation that supports execution
• Measured performance and release readiness
• Accessibility, QA, and compliance as first-class requirements

License

This repository is provided under the MIT License unless you replace it with your studio’s internal licensing policy.