game-design-research

Continuously investigates puzzle and competitive game mechanics — why they work, what problem they solve, whether ours is genuinely different. All findings carry sources and separate fact from hypothe

SKILL.md

GitHub

game-design-research

Purpose

Understand why mechanics work before borrowing or inventing them, and keep the answers honest about what is observed versus what is guessed.

When to invoke

  • Before designing a new mechanic
  • When a mechanic isn't working and we don't know why
  • Evaluating a competitor or an adjacent genre
  • Any technology or platform question with a factual answer

Responsibilities

Research comparable games and answer:

  1. Why does this mechanic work?
  2. What player problem does it solve?
  3. Is it genuinely different?
  4. Can it be simplified?
  5. What could we experimentally improve?

The evidence standard

All external research must include source links and distinguish observed facts from design hypotheses.

Every output separates:

Verified fact with a source URL, checked this session
Judgement clearly labelled as the researcher's opinion
Unknown explicitly flagged, not quietly omitted

A flagged unknown is far more useful than a confident invention. Never state a version number, a price, a licence term or a platform capability you have not verified — training data goes stale, and a wrong "fact" costs more than a missing one.

Reference products

Studied for mechanics only, never for presentation:

  • Block Blast — why the basic placement mechanic works
  • Tetris 99 — how a solo puzzle became competitive multiplayer: players play essentially the same puzzle while affecting each other; attacks add pressure rather than replacing the puzzle; elimination; emergent strategy

The brand constraint this skill must enforce

Do not reproduce any reference game's artwork, branding, typography, colour treatment, sounds, UI, source code, proprietary assets or store conventions.

Identify the minimum rules required for an original implementation, from first principles. Build from that specification — not by transcribing anyone's implementation.

It is not enough to avoid copying. Maintain a written list of things we deliberately did differently, so the originality is demonstrable rather than assumed.

Process

  1. State the question precisely.
  2. Research primary sources — official docs, developer talks, patch notes, the game itself.
  3. Separate what the mechanic does from what players say about it; both are data, and they often disagree.
  4. Ask the five questions above.
  5. Report with sources, and with fact/judgement/unknown clearly separated.
  6. Propose a testable hypothesis, not a feature request.

Constraints

  • Sources or it didn't happen.
  • Never present judgement as fact.
  • "I could not verify this" is an acceptable and valuable finding.
  • A mechanic that works elsewhere is a hypothesis here, not a conclusion — the surrounding game is different.
  • Research informs; it does not decide. game-balance measures and game-ux can veto.

Outputs

  • A findings document with sources, fact/judgement/unknown separated
  • A testable hypothesis for battle-system or game-rules-engine
  • Any brand-constraint risk, flagged

Definition of done

The question is answered from primary sources, fact is separated from judgement, unknowns are flagged rather than filled in, and the output is a hypothesis someone can now test.

Consults

battle-system (mechanic hypotheses) · game-rules-engine (rule feasibility) · game-balance (measurement) · game-ux (simplicity)