Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Unity DOTS/ECS skill for data-oriented design, jobs system, burst compiler optimization, and high-performance gameplay systems.
.claude/skills/a5c-ai-unity-ecs/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-01 | ✓→✓ | = Same ✓ | -10% | 0% |
| case-02 | ✓→✓ | = Same ✓ | 56% | 0% |
| case-03 | ✓→✓ | = Same ✓ | 3% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 3% | 0% |
Data-Oriented Technology Stack (DOTS) and Entity Component System development for Unity.
This skill provides capabilities for implementing high-performance gameplay systems using Unity's Entity Component System, Jobs System, and Burst compiler.
csharppublic struct MoveSpeed : IComponentData { public float Value; } public struct Velocity : IComponentData { public float3 Value; }
csharp[BurstCompile] public partial struct MovementSystem : ISystem { [BurstCompile] public void OnUpdate(ref SystemState state) { float deltaTime = SystemAPI.Time.DeltaTime; foreach (var (transform, velocity) in SystemAPI.Query<RefRW<LocalTransform>, RefRO<Velocity>>()) { transform.ValueRW.Position += velocity.ValueRO.Value * deltaTime; } } }
csharp[BurstCompile] public partial struct MovementJob : IJobEntity { public float DeltaTime; void Execute(ref LocalTransform transform, in Velocity velocity) { transform.Position += velocity.Value * DeltaTime; } }
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 16,461 | 21,265 | +29% | 1 | 1 | 0% | 2,546 | 2,290 | -10% | 0 | 0 | — |
case-02 | pass→pass | 5,025 | 3,029 | -40% | 1 | 1 | 0% | 672 | 1,047 | +56% | 0 | 0 | — |
case-03 | pass→pass | 13,855 | 12,081 | -13% | 1 | 1 | 0% | 2,390 | 2,464 | +3% | 0 | 0 | — |
case-04 | pass→pass | 13,382 | 7,509 | -44% | 1 | 1 | 0% | 2,008 | 2,071 | +3% | 0 | 0 | — |
case-05 | pass→pass | 7,925 | 6,582 | -17% | 1 | 1 | 0% | 1,333 | 1,764 | +32% | 0 | 0 | — |
case-06 | pass→pass | 10,355 | 7,013 | -32% | 1 | 1 | 0% | 1,524 | 1,745 | +15% | 0 | 0 | — |
case-07 | pass→pass | 5,063 | 3,872 | -24% | 1 | 1 | 0% | 687 | 1,128 | +64% | 0 | 0 | — |
case-08 | pass→pass | 5,094 | 4,186 | -18% | 1 | 1 | 0% | 795 | 1,053 | +32% | 0 | 0 | — |
case-09 | pass→pass | 4,957 | 3,922 | -21% | 1 | 1 | 0% | 774 | 1,184 | +53% | 0 | 0 | — |
case-10 | pass→pass | 6,075 | 4,216 | -31% | 1 | 1 | 0% | 836 | 1,129 | +35% | 0 | 0 | — |
case-11 | pass→pass | 3,889 | 4,261 | +10% | 1 | 1 | 0% | 662 | 1,193 | +80% | 0 | 0 | — |
case-12 | pass→pass | 3,968 | 4,288 | +8% | 1 | 1 | 0% | 622 | 1,264 | +103% | 0 | 0 | — |
case-13 | pass→pass | 6,657 | 4,150 | -38% | 1 | 1 | 0% | 940 | 1,241 | +32% | 0 | 0 | — |
case-14 | pass→pass | 14,729 | 13,620 | -8% | 1 | 1 | 0% | 2,276 | 2,726 | +20% | 0 | 0 | — |
case-15 | fail→pass | 7,889 | 5,331 | -32% | 1 | 1 | 0% | 1,335 | 1,457 | +9% | 0 | 0 | — |
case-16 | pass→pass | 20,662 | 15,266 | -26% | 1 | 1 | 0% | 2,848 | 3,163 | +11% | 0 | 0 | — |
case-17 | pass→pass | 6,927 | 3,337 | -52% | 1 | 1 | 0% | 939 | 1,005 | +7% | 0 | 0 | — |
case-18 | pass→pass | 15,821 | 14,487 | -8% | 1 | 1 | 0% | 2,272 | 2,506 | +10% | 0 | 0 | — |
case-19 | pass→pass | 7,110 | 4,758 | -33% | 1 | 1 | 0% | 997 | 1,132 | +14% | 0 | 0 | — |
case-20 | pass→pass | 3,501 | 4,154 | +19% | 1 | 1 | 0% | 430 | 1,156 | +169% | 0 | 0 | — |
case-21 | pass→pass | 6,159 | 8,577 | +39% | 1 | 1 | 0% | 990 | 1,774 | +79% | 0 | 0 | — |
case-22 | pass→pass | 4,347 | 6,329 | +46% | 1 | 1 | 0% | 750 | 1,371 | +83% | 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. 22 cases were attempted. The headline lift of +5 percentage points is the difference between those two pass rates over the 22 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.