Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Master Unity ECS (Entity Component System) with DOTS, Jobs, and Burst for high-performance game development. Use when building data-oriented games, optimizing performance, or working with large entity counts.
.claude/skills/wshobson-unity-ecs-patterns/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-20 | ✗→✓ | ▲ Improved | -4% | 0% |
| case-22 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-08 | ✓→✓ | = Same ✓ | -13% | 0% |
| case-13 | ✓→✓ | = Same ✓ | -10% | 0% |
| case-14 | ✓→✓ | = Same ✓ | 7% | 0% |
Production patterns for Unity's Data-Oriented Technology Stack (DOTS) including Entity Component System, Job System, and Burst Compiler.
| Aspect | Traditional OOP | ECS/DOTS | | ----------- | ----------------- | --------------- | | Data layout | Object-oriented | Data-oriented | | Memory | Scattered | Contiguous | | Processing | Per-object | Batched | | Scaling | Poor with count | Linear scaling | | Best for | Complex behaviors | Mass simulation |
Entity: Lightweight ID (no data)
Component: Pure data (no behavior)
System: Logic that processes components
World: Container for entities
Archetype: Unique combination of components
Chunk: Memory block for same-archetype entitiesDetailed pattern documentation lives in references/details.md. Read that file when the navigation tier above is insufficient.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | pass→pass | 21,256 | 14,956 | -30% | 1 | 1 | 0% | 3,275 | 2,850 | -13% | 0 | 0 | — |
case-13 | pass→pass | 13,038 | 9,265 | -29% | 1 | 1 | 0% | 2,389 | 2,152 | -10% | 0 | 0 | — |
case-14 | pass→pass | 12,895 | 10,166 | -21% | 1 | 1 | 0% | 2,134 | 2,278 | +7% | 0 | 0 | — |
case-01 | pass→pass | 14,781 | 9,066 | -39% | 1 | 1 | 0% | 2,648 | 2,259 | -15% | 0 | 0 | — |
case-02 | pass→pass | 10,389 | 8,878 | -15% | 1 | 1 | 0% | 2,095 | 1,972 | -6% | 0 | 0 | — |
case-03 | pass→pass | 8,744 | 9,548 | +9% | 1 | 1 | 0% | 1,896 | 2,125 | +12% | 0 | 0 | — |
case-04 | pass→pass | 10,227 | 7,556 | -26% | 1 | 1 | 0% | 1,994 | 1,852 | -7% | 0 | 0 | — |
case-05 | pass→pass | 8,159 | 7,048 | -14% | 1 | 1 | 0% | 1,794 | 1,807 | +1% | 0 | 0 | — |
case-06 | pass→pass | 17,180 | 11,916 | -31% | 1 | 1 | 0% | 2,628 | 2,658 | +1% | 0 | 0 | — |
case-07 | pass→pass | 19,070 | 16,622 | -13% | 1 | 1 | 0% | 3,143 | 3,077 | -2% | 0 | 0 | — |
case-09 | pass→pass | 14,735 | 7,406 | -50% | 1 | 1 | 0% | 2,473 | 1,773 | -28% | 0 | 0 | — |
case-10 | pass→pass | 9,839 | 7,208 | -27% | 1 | 1 | 0% | 1,792 | 1,897 | +6% | 0 | 0 | — |
case-11 | pass→pass | 9,628 | 8,217 | -15% | 1 | 1 | 0% | 1,798 | 1,909 | +6% | 0 | 0 | — |
case-12 | pass→pass | 10,174 | 7,302 | -28% | 1 | 1 | 0% | 1,796 | 1,783 | -1% | 0 | 0 | — |
case-15 | pass→pass | 11,273 | 8,778 | -22% | 1 | 1 | 0% | 1,870 | 1,837 | -2% | 0 | 0 | — |
case-16 | pass→pass | 10,535 | 9,069 | -14% | 1 | 1 | 0% | 2,192 | 2,239 | +2% | 0 | 0 | — |
case-17 | pass→pass | 3,092 | 2,291 | -26% | 1 | 1 | 0% | 525 | 897 | +71% | 0 | 0 | — |
case-18 | pass→pass | 8,672 | 7,852 | -9% | 1 | 1 | 0% | 1,518 | 1,693 | +12% | 0 | 0 | — |
case-19 | pass→pass | 9,359 | 7,567 | -19% | 1 | 1 | 0% | 1,631 | 1,524 | -7% | 0 | 0 | — |
case-20 | fail→pass | 15,950 | 15,490 | -3% | 1 | 1 | 0% | 2,950 | 2,844 | -4% | 0 | 0 | — |
case-21 | pass→pass | 14,873 | 12,682 | -15% | 1 | 1 | 0% | 3,460 | 3,217 | -7% | 0 | 0 | — |
case-22 | fail→pass | 10,917 | 10,144 | -7% | 1 | 1 | 0% | 1,996 | 2,040 | +2% | 0 | 0 | — |
case-23 | pass→pass | 15,120 | 15,590 | +3% | 1 | 1 | 0% | 2,970 | 3,635 | +22% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 23 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.