---

name: game-design description: This skill should be used when the user asks to "design a game mechanic", "evaluate gameplay feel", "tune game systems", "review player experience", "debug why something feels wrong", "balance combat", "design progression", or when working on any player-facing game feature. Provides a constraint system for evaluating mechanics with focus on player experience over feature completion.

Game Design Framework

Purpose: Constraint system for evaluating and implementing game mechanics with focus on player experience over feature completion.

Core principle: Mechanics are code. Gameplay is the player's experience of that code. The goal is not to implement features, but to implement Relevance.

---

Quick Reference: The 5-Component Filter

Before implementing or critiquing ANY game feature, evaluate against:

Component	Core Question	Quick Check
Clarity	Can the player predict what will happen?	Telegraph exists before resolution
Motivation	Does the player care about the outcome?	Outcome affects persistent state
Response	Do player inputs matter?	Actions can be buffered/cancelled meaningfully
Satisfaction	Does success feel earned?	Multiple feedback channels fire (visual + audio minimum)
Fit	Does it match the game's identity?	Weight, timing, audio match entity type

Conflict priority: Response > Clarity > Satisfaction > Fit > Motivation

For detailed evaluation rubrics, consult references/5-component-rubric.md.

---

Operating Protocol

1. Before Implementation

1. Identify active domain(s) from references/domain-guide.md
2. Evaluate against the 5-Component Filter
3. Complete the State Machine Checklist if the feature involves player state changes
4. Check the Numbers Policy before proposing any values

2. Numbers Policy (Mandatory)

When proposing ANY numeric value (timing windows, costs, speeds, damage, etc.), choose ONE:

Option A — Source-backed:

• Cite a verifiable reference (GDC talk, postmortem, published analysis)
• Example: "Coyote time of 80-150ms (Source: Maddy Thorson's Celeste postmortem)"

Option B — Starting value with test plan:

• Label explicitly as "Starting value"
• Include: micro test plan, pass/fail metric, adjustment direction if it fails
• Example: "Starting value: 120ms. Test: Can players make intended jumps 9/10 times? If fail rate >20%, increase by 30ms increments."

Never claim "industry standard" or "common practice" without a source.

3. Assumption Labeling

When critical information is missing, state explicitly:

ASSUMPTION: [what you're assuming]
IMPACT: [why it matters to the design]
IF WRONG: [failure mode]
VALIDATE: [how to check quickly]

4. Research Triggers

Search before proposing when:

• About to claim "best practice" or "standard approach"
• Balance/economy values need benchmarks
• Accessibility requirements apply
• Comparative references needed from similar games

If search unavailable, convert to "Assumption + Test Plan" format.

---

State Machine Checklist

For ANY feature that changes player state (movement abilities, combat actions, status effects):

Property	Must Define
Entry conditions	What states can transition INTO this?
Exit conditions	What ends this state? (timer, input, external event)
Interruptibility	What can cancel this? (damage, player input, other abilities)
Chained actions	What states can this transition TO?
Resource cost	What is consumed on entry? On sustain?
Edge cases	Behavior on: slopes, ceilings, moving platforms, during hitstun, at resource zero

---

Debugging Protocol

When told "it feels wrong/boring/clunky," diagnose in order:

Symptom	Check First	Before Tuning Numbers
"I didn't know that would happen"	Clarity	Add telegraph, audio cue, UI indicator
"I don't care"	Motivation	Connect to progression, increase stakes
"It feels laggy"	Response	Add buffering, allow cancels, reduce lockouts
"It feels weak"	Satisfaction	Add feedback channels (minimum 2)
"It doesn't fit"	Fit	Adjust timing, weight, audio texture

Rule: Do not tune damage/timing numbers until Clarity and Response are verified as not the root cause.

---

Playtest Requirements

Every significant feature must include scenarios for:

1. New player test: Can they infer the rules without being told?
2. Stress test: Spam inputs, boundary conditions, edge cases
3. Skill test: Can mastery improve outcomes meaningfully?
4. Abuse test: Can this be exploited to skip content or trivialize risk?
5. Readability test: Can an observer understand what happened and why?

---

Red Flags (Stop and Clarify)

• State machine transitions are undefined ("works from any state")
• Multiplayer authority is unspecified
• Economy/currency feature has no balance targets
• Camera behavior during action is undefined
• Feature scope is actually 3+ features in disguise

---

Definition of Done

• 5-Component Filter evaluated and documented
• State Machine Checklist completed (if applicable)
• Edge cases enumerated and handled
• Minimum 2 feedback channels for significant actions
• Playtest script written and smoke-tested
• Numbers justified per Numbers Policy

---

Output Structure

When proposing or critiquing a feature:

1. Player Goal & Context — What is the player trying to do and why?
2. System Rules — Core behavior, failure conditions, edge cases
3. 5-Component Evaluation — Which components are strong/weak?
4. Risks & Abuse Cases — What could break or be exploited?
5. Playtest Scenarios — How to validate quickly
6. Tuning Priority — What to adjust first if it doesn't feel right

---

Reference Files

For detailed guidance:

• references/5-component-rubric.md - Full evaluation rubrics with signals, rules, knobs, acceptance tests
• references/domain-guide.md - Combat, movement, camera, audio, UI/UX, progression, persistence domains
• references/templates.md - Edge case enumeration, debugging flow, playtest scripts