tachyon-beep/using-simulation-tactics

Using Simulation Tactics: The Router Meta-Skill

Description

This is the PRIMARY ROUTER META-SKILL for the simulation-tactics skillpack. It teaches you how to:

1. Analyze simulation requirements - Understand what the user actually needs
2. Route to appropriate skills - Determine which of the 10 core skills apply
3. Apply skills in correct order - Use the optimal workflow for the situation
4. Combine multiple skills - Handle complex scenarios requiring several simulation types

This skill does NOT teach simulation implementation details. It teaches DECISION MAKING: which skill to use, when, and why.

When to Use This Meta-Skill

Use this meta-skill when:

• Starting ANY simulation-related game development task
• User asks about simulation but unclear which type
• Facing complex scenarios requiring multiple simulation types
• Need to determine implementation order for multi-system games
• Debugging simulation issues and unclear where to start
• Planning architecture for simulation-heavy games

How to Access Reference Sheets

IMPORTANT: All reference sheets are located in the SAME DIRECTORY as this SKILL.md file.

When this skill is loaded from: skills/using-simulation-tactics/SKILL.md

Reference sheets like physics-simulation-patterns.md are at: skills/using-simulation-tactics/physics-simulation-patterns.md

NOT at: skills/physics-simulation-patterns.md ← WRONG PATH

---

The 10 Core Skills

Before routing, understand what each skill provides:

1. simulation-vs-faking (FOUNDATIONAL)

What it teaches: The fundamental trade-off between full simulation and approximation/faking When to route: ALWAYS FIRST - determines if you even need simulation Key question: "Do I simulate this, fake it, or use a hybrid approach?"

2. physics-simulation-patterns

What it teaches: Rigid bodies, vehicles, cloth, fluids, integration methods When to route: Need realistic physics for vehicles, ragdolls, destructibles, or fluid dynamics Key question: "Does this need real-time physics simulation?"

3. ai-and-agent-simulation

What it teaches: FSM, behavior trees, utility AI, GOAP, agent behaviors When to route: Need intelligent agent behavior (enemies, NPCs, units) Key question: "Do agents need to make decisions and act autonomously?"

4. traffic-and-pathfinding

What it teaches: A*, navmesh, flow fields, traffic simulation, congestion When to route: Need agents to navigate environments or simulate traffic Key question: "Do entities need to find paths or simulate traffic flow?"

5. economic-simulation-patterns

What it teaches: Supply/demand, markets, trade networks, price discovery When to route: Need economic systems (trading, markets, resources) Key question: "Does the game involve trade, economy, or resource markets?"

6. ecosystem-simulation

What it teaches: Predator-prey dynamics, food chains, population control When to route: Need living ecosystems with wildlife populations Key question: "Do I need animals/plants that breed, eat, and die naturally?"

7. crowd-simulation

What it teaches: Boids, formations, social forces, LOD for crowds When to route: Need large groups moving together (crowds, flocks, armies) Key question: "Do I need many entities moving as a coordinated group?"

8. weather-and-time

What it teaches: Day/night cycles, weather systems, seasonal effects When to route: Need atmospheric effects or time-based gameplay Key question: "Does the game need time progression or weather?"

9. performance-optimization-for-sims

What it teaches: Profiling, spatial partitioning, LOD, time-slicing, caching When to route: Performance problems with existing simulation Key question: "Is my simulation too slow?"

10. debugging-simulation-chaos

What it teaches: Systematic debugging, desync detection, determinism, chaos prevention When to route: Simulation behaves incorrectly, chaotically, or unpredictably Key question: "Is my simulation broken, desyncing, or chaotic?"

---

CORE ROUTING FRAMEWORK

The Decision Tree

Follow this decision tree for ALL simulation tasks:

┌─────────────────────────────────────────────────────────────┐
│ STEP 1: ALWAYS START HERE                                   │
│ ═══════════════════════════════════════════════════════════ │
│ Route to: simulation-vs-faking                              │
│                                                              │
│ Questions to ask:                                            │
│ • Do I need to simulate this at all?                        │
│ • What level of detail is required?                         │
│ • What can I fake or approximate?                           │
│ • Where is the player's attention focused?                  │
│                                                              │
│ This prevents the #1 mistake: over-engineering systems     │
│ that could be faked.                                        │
└─────────────────────────────────────────────────────────────┘
                            ↓
┌─────────────────────────────────────────────────────────────┐
│ STEP 2: ROUTE TO SPECIFIC SIMULATION TYPE(S)                │
│ ═══════════════════════════════════════════════════════════ │
│ Identify which simulation domains apply:                    │
│                                                              │
│ Physics domain → physics-simulation-patterns                │
│ AI domain → ai-and-agent-simulation                         │
│ Pathfinding domain → traffic-and-pathfinding                │
│ Economy domain → economic-simulation-patterns               │
│ Ecosystem domain → ecosystem-simulation                     │
│ Crowds domain → crowd-simulation                            │
│ Atmosphere domain → weather-and-time                        │
│                                                              │
│ Multiple domains? Route to ALL applicable skills.           │
└─────────────────────────────────────────────────────────────┘
                            ↓
┌─────────────────────────────────────────────────────────────┐
│ STEP 3: IF PERFORMANCE ISSUES ARISE                         │
│ ═══════════════════════════════════════════════════════════ │
│ Route to: performance-optimization-for-sims                 │
│                                                              │
│ Triggers:                                                    │
│ • Frame rate drops below 60 FPS                             │
│ • Profiler shows simulation bottleneck                      │
│ • Agent count causes slowdown                               │
│ • Simulation gets expensive at scale                        │
│                                                              │
│ WARNING: Don't route here prematurely!                      │
│ Premature optimization wastes time.                         │
└─────────────────────────────────────────────────────────────┘
                            ↓
┌─────────────────────────────────────────────────────────────┐
│ STEP 4: IF BUGS/CHAOS OCCUR                                 │
│ ═══════════════════════════════════════════════════════════ │
│ Route to: debugging-simulation-chaos                        │
│                                                              │
│ Triggers:                                                    │
│ • Simulation behaves chaotically/unpredictably              │
│ • Multiplayer desyncs                                       │
│ • Physics explosions or NaN values                          │
│ • Agents stuck or behaving erratically                      │
│ • Systems producing nonsensical results                     │
│                                                              │
│ This is a REACTIVE skill - only use when broken.            │
└─────────────────────────────────────────────────────────────┘

Key Routing Principles

Principle 1: simulation-vs-faking is ALWAYS step 1

• Even if you "know" you need simulation, validate this assumption
• Prevents 90% of over-engineering disasters
• Takes 5 minutes, saves hours of wasted work

Principle 2: Multiple domains are common

• Most games need 2-4 simulation types
• Route to ALL applicable skills
• Order matters (see multi-skill-workflows.md)

Principle 3: Optimization comes AFTER implementation

• Don't route to performance-optimization-for-sims until you have a working simulation
• Profile first, optimize later
• Premature optimization is the root of all evil

Principle 4: Debugging is reactive, not proactive

• Only route to debugging-simulation-chaos when something is broken
• Don't use it as a preventative measure
• Fix the bug, THEN refactor to prevent recurrence

---

QUICK REFERENCE TABLE

| User Need | Primary Skill | Secondary Skills | Also Consider | |-----------|---------------|------------------|---------------| | Vehicle physics | physics-simulation-patterns | - | performance-optimization (if many vehicles) | | City traffic | traffic-and-pathfinding | simulation-vs-faking | performance-optimization (scale to 10k) | | NPC AI | ai-and-agent-simulation | simulation-vs-faking | traffic-and-pathfinding (if NPCs move) | | RTS units | ai-and-agent-simulation, traffic-and-pathfinding | crowd-simulation (formations) | performance-optimization (1000+ units) | | Trading system | economic-simulation-patterns | simulation-vs-faking | ai-and-agent-simulation (NPC traders) | | Wildlife/hunting | ecosystem-simulation | ai-and-agent-simulation | simulation-vs-faking (detail level) | | Crowds | crowd-simulation | simulation-vs-faking | performance-optimization (scale) | | Day/night | weather-and-time | simulation-vs-faking | - | | Weather effects | weather-and-time | physics-simulation-patterns (wind) | - | | Pathfinding | traffic-and-pathfinding | simulation-vs-faking | performance-optimization (many agents) | | Flocking birds | crowd-simulation | simulation-vs-faking | performance-optimization (LOD) | | Ragdolls | physics-simulation-patterns | - | debugging-simulation-chaos (stability) | | Performance issue | performance-optimization-for-sims | (original implementation skill) | debugging-simulation-chaos (if bug) | | Physics explodes | debugging-simulation-chaos | physics-simulation-patterns | - | | Ecosystem collapse | debugging-simulation-chaos | ecosystem-simulation | - | | Multiplayer desync | debugging-simulation-chaos | (any affected skills) | - |

---

DECISION FLOWCHART

Use this flowchart for quick routing decisions:

START: User describes simulation need
    ↓
[Is this DEFINITELY about simulation?]
    ├─ No → Don't use simulation-tactics at all
    └─ Yes → Continue
        ↓
[Route to: simulation-vs-faking]
    "Do I simulate, fake, or hybrid?"
        ↓
[Identify domain(s)]
    ├─ Physics? → physics-simulation-patterns
    ├─ AI/Agents? → ai-and-agent-simulation
    ├─ Pathfinding? → traffic-and-pathfinding
    ├─ Economy? → economic-simulation-patterns
    ├─ Ecosystem? → ecosystem-simulation
    ├─ Crowds? → crowd-simulation
    └─ Weather/Time? → weather-and-time
        ↓
[Is simulation ALREADY implemented?]
    ├─ No → Use identified skill(s) to implement
    └─ Yes → Continue
        ↓
[Is there a PERFORMANCE problem?]
    ├─ Yes → performance-optimization-for-sims
    └─ No → Continue
        ↓
[Is there a BUG/CHAOS problem?]
    ├─ Yes → debugging-simulation-chaos
    └─ No → Implementation complete!

---

COMMON ROUTING MISTAKES

These are the mistakes that waste the most time. Learn to recognize them.

Mistake 1: Skipping simulation-vs-faking

Symptom: Over-engineered simulation that could have been faked

Example:

• Building full ecosystem for background birds that are never scrutinized
• Simulating NPC hunger/sleep when player never notices
• Full traffic simulation for distant cars player can't interact with

Fix: ALWAYS route to simulation-vs-faking first. Ask "Will player notice if I fake this?"

Cost of mistake: Weeks of wasted work, ongoing performance burden

Mistake 2: Premature optimization

Symptom: Routing to performance-optimization-for-sims before implementation is complete

Example:

• Implementing LOD systems before having working simulation
• Using spatial partitioning before knowing if it's needed
• Caching pathfinding before pathfinding exists

Fix: Profile first, optimize later. Only route to performance-optimization-for-sims when:

• You have working simulation
• You have measured performance problem
• Profiler shows bottleneck

Cost of mistake: Wasted time optimizing code that might change, or optimizing the wrong thing

Mistake 3: Not debugging systematically

Symptom: Trying to fix bugs by changing random things, routing to implementation skills instead of debugging-simulation-chaos

Example:

• "Physics explodes, let me try different integration method" (should debug first)
• "Ecosystem collapses, let me add more food" (should debug why it collapses)
• "Pathfinding breaks, let me rewrite the algorithm" (should debug the existing code)

Fix: When simulation is broken, ALWAYS route to debugging-simulation-chaos first. Identify root cause before attempting fixes.

Cost of mistake: Bug persists, or you "fix" symptom without addressing cause

Mistake 4: Wrong skill for the domain

Symptom: Using ai-and-agent-simulation when you need traffic-and-pathfinding, etc.

Example:

• Using ai-and-agent-simulation for pathfinding (use traffic-and-pathfinding instead)
• Using physics-simulation-patterns for kinematic movement (use ai-and-agent-simulation)
• Using crowd-simulation for trading (use economic-simulation-patterns)

Fix: Understand what each skill covers. Pathfinding is NOT AI. Physics is NOT movement. Crowds are NOT flocking AI.

Cost of mistake: Learning wrong techniques for your problem

Mistake 5: Implementing in wrong order

Symptom: Building dependent system before foundation

Example:

• Implementing AI behaviors before pathfinding exists (AI can't move)
• Building economy before resource sources exist (nothing to trade)
• Adding weather effects before day/night cycle (no time progression)

Fix: Follow the dependency order in multi-skill workflows. Foundation first, then dependent systems.

Cost of mistake: Rework when foundation changes breaks dependent systems

Mistake 6: Ignoring multiplayer determinism

Symptom: Building single-player simulation without considering multiplayer needs

Example:

• Using floating-point physics for multiplayer game (desyncs)
• Random number generation without shared seed (desyncs)
• Iterating unordered collections (desyncs)

Fix: If multiplayer is planned, route to debugging-simulation-chaos early to learn determinism requirements.

Cost of mistake: Complete rewrite to fix desyncs

Mistake 7: Over-combining skills

Symptom: Trying to use every skill when only 1-2 are needed

Example:

• Simple puzzle game doesn't need ecosystem-simulation
• Turn-based game doesn't need performance-optimization-for-sims
• Static world doesn't need weather-and-time

Fix: Route to ONLY the skills you actually need. More skills = more complexity.

Cost of mistake: Wasted time learning and implementing unnecessary systems

---

Additional Guidance

For detailed examples and workflows, see:

• routing-scenarios.md - 20 concrete routing examples
• multi-skill-workflows.md - 8 game genre workflows (RTS, Survival, City Builder, etc.)
• expert-routing-guide.md - Tips, checklists, self-checks, edge cases

---

Simulation Tactics Specialist Skills Catalog

After routing, load the appropriate specialist skill for detailed guidance:

1. simulation-vs-faking.md - FOUNDATIONAL: Trade-off between full simulation and approximation/faking
2. physics-simulation-patterns.md - Rigid bodies, vehicles, cloth, fluids, integration methods
3. ai-and-agent-simulation.md - FSM, behavior trees, utility AI, GOAP, agent behaviors
4. traffic-and-pathfinding.md - A*, navmesh, flow fields, traffic simulation, congestion
5. economic-simulation-patterns.md - Supply/demand, markets, trade networks, price discovery
6. ecosystem-simulation.md - Predator-prey dynamics, food chains, population control
7. crowd-simulation.md - Boids, formations, social forces, LOD for crowds
8. weather-and-time.md - Day/night cycles, weather systems, seasonal effects
9. performance-optimization-for-sims.md - Profiling, spatial partitioning, LOD, time-slicing
10. debugging-simulation-chaos.md - Systematic debugging, desync detection, determinism

Now route confidently to the specific skills you need!