game-design-fundamentals

The foundational design discipline — the spine the rest of the pack cross-links to. Use when judging whether a mechanic is fun, auditing a system for balance, designing difficulty/flow, setting reward

SKILL.md

GitHub

Game-Design Fundamentals

The bedrock principles every other skill in the pack assumes. If a design decision feels wrong and you're not sure why, the answer is usually here.

When to use this

  • Deciding whether a mechanic, system, or choice is actually fun (not just functional)
  • Balancing — hunting dominant strategies, no-brainer options, difficulty spikes
  • Designing difficulty/flow, reward schedules, or progression
  • Diagnosing player complaints: "my choices don't matter", "it's grindy", "that death was cheap", "I'm overwhelmed"

Scope

This is the spine. Sibling skills specialize it and link back here:

  • Spatial reward cadence & exploration pull → open-world-design
  • Perceptual uniqueness / the oatmeal problem → procedural-generation, procgen-review
  • Telegraphing & fair lethality → combat-design, permadeath-and-lethality
  • Loot dominance & economy balance → rpg-systems
  • Real vs. fake branching → narrative-and-quest-design

How the pieces fit

  • GUIDE.md — the cited why: interesting decisions, flow & difficulty, motivation (SDT), feedback & agency, choice architecture. Each principle carries exemplars, sources, the named failure mode, and the procedural/headless implication.
  • CHECKLIST.md — Do/Don't + machine-checkable Test-for criteria, grouped by topic. These are written to be enforced as automated invariants in a generation loop, not just read.

The one idea to anchor on

A game is a series of interesting decisions (Sid Meier). A decision is interesting only when it has believed, persistent consequences, is made with enough information, carries a real trade-off, and has no dominant option. The single highest-leverage audit you can run: scan every system for a strictly-best strategy and delete it.

Why this matters doubly for a generator: a human team has a "this feels off" reflex that catches a broken economy or a cheap death late. An autonomous agent has none. So treat each principle here as an enforced automated invariant, not advice.

Start with GUIDE.md.