Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Web browser game development principles. Framework selection, WebGPU, optimization, PWA.
.claude/skills/lingxling-web-games/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -61% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -46% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | -8% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -13% | 0% |
> Framework selection and browser-specific principles.
What type of game?
│
├── 2D Game
│ ├── Full game engine features? → Phaser
│ └── Raw rendering power? → PixiJS
│
├── 3D Game
│ ├── Full engine (physics, XR)? → Babylon.js
│ └── Rendering focused? → Three.js
│
└── Hybrid / Canvas
└── Custom → Raw Canvas/WebGL| Framework | Type | Best For | |-----------|------|----------| | Phaser 4 | 2D | Full game features | | PixiJS 8 | 2D | Rendering, UI | | Three.js | 3D | Visualizations, lightweight | | Babylon.js 7 | 3D | Full engine, XR |
| Browser | Support | |---------|---------| | Chrome | ✅ Since v113 | | Edge | ✅ Since v113 | | Firefox | ✅ Since v131 | | Safari | ✅ Since 18.0 | | Total | ~73% global |
navigator.gpu| Constraint | Strategy | |------------|----------| | No local file access | Asset bundling, CDN | | Tab throttling | Pause when hidden | | Mobile data limits | Compress assets | | Audio autoplay | Require user interaction |
| Type | Format | |------|--------| | Textures | KTX2 + Basis Universal | | Audio | WebM/Opus (fallback: MP3) | | 3D Models | glTF + Draco/Meshopt |
| Phase | Load | |-------|------| | Startup | Core assets, <2MB | | Gameplay | Stream on demand | | Background | Prefetch next level |
| ❌ Don't | ✅ Do | |----------|-------| | Load all assets upfront | Progressive loading | | Ignore tab visibility | Pause when hidden | | Block on audio load | Lazy load audio | | Skip compression | Compress everything | | Assume fast connection | Handle slow networks |
> Remember: Browser is the most accessible platform. Respect its constraints.
This skill is applicable to execute the workflow or actions described in the overview.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-19 | fail→fail | 33,180 | 36,907 | +11% | 1 | 1 | 0% | 5,471 | 6,216 | +14% | 0 | 0 | — |
case-01 | fail→fail | 47,248 | 25,004 | -47% | 1 | 1 | 0% | 6,772 | 5,307 | -22% | 0 | 0 | — |
case-02 | pass→pass | 17,671 | 13,145 | -26% | 1 | 1 | 0% | 2,801 | 2,569 | -8% | 0 | 0 | — |
case-03 | fail→pass | 24,032 | 11,927 | -50% | 1 | 1 | 0% | 2,529 | 2,162 | -15% | 0 | 0 | — |
case-04 | pass→pass | 16,784 | 11,187 | -33% | 1 | 1 | 0% | 2,683 | 2,343 | -13% | 0 | 0 | — |
case-05 | pass→pass | 13,121 | 6,082 | -54% | 1 | 1 | 0% | 2,088 | 1,814 | -13% | 0 | 0 | — |
case-06 | pass→pass | 19,299 | 18,929 | -2% | 1 | 1 | 0% | 2,753 | 3,788 | +38% | 0 | 0 | — |
case-07 | fail→pass | 19,207 | 2,220 | -88% | 1 | 1 | 0% | 2,928 | 1,141 | -61% | 0 | 0 | — |
case-08 | pass→pass | 9,217 | 2,109 | -77% | 1 | 1 | 0% | 1,581 | 1,189 | -25% | 0 | 0 | — |
case-09 | pass→pass | 19,743 | 16,289 | -17% | 1 | 1 | 0% | 2,714 | 3,450 | +27% | 0 | 0 | — |
case-20 | fail→fail | 24,295 | 17,436 | -28% | 1 | 1 | 0% | 3,671 | 4,363 | +19% | 0 | 0 | — |
case-10 | fail→fail | 16,378 | 17,934 | +10% | 1 | 1 | 0% | 2,770 | 3,790 | +37% | 0 | 0 | — |
case-11 | fail→pass | 22,797 | 6,168 | -73% | 1 | 1 | 0% | 3,423 | 1,861 | -46% | 0 | 0 | — |
case-12 | pass→pass | 18,633 | 3,051 | -84% | 1 | 1 | 0% | 2,602 | 1,239 | -52% | 0 | 0 | — |
case-13 | pass→pass | 15,808 | 7,003 | -56% | 1 | 1 | 0% | 2,714 | 2,115 | -22% | 0 | 0 | — |
case-14 | pass→pass | 24,345 | 15,173 | -38% | 1 | 1 | 0% | 3,406 | 3,714 | +9% | 0 | 0 | — |
case-15 | pass→pass | 18,912 | 17,733 | -6% | 1 | 1 | 0% | 2,843 | 4,201 | +48% | 0 | 0 | — |
case-16 | pass→pass | 16,071 | 10,300 | -36% | 1 | 1 | 0% | 2,476 | 2,716 | +10% | 0 | 0 | — |
case-17 | pass→pass | 8,404 | 7,011 | -17% | 1 | 1 | 0% | 1,321 | 2,043 | +55% | 0 | 0 | — |
case-18 | fail→fail | 24,617 | 19,980 | -19% | 1 | 1 | 0% | 3,507 | 3,768 | +7% | 0 | 0 | — |
case-21 | pass→pass | 22,826 | 13,863 | -39% | 1 | 1 | 0% | 3,125 | 3,256 | +4% | 0 | 0 | — |
case-22 | pass→pass | 21,616 | 20,196 | -7% | 1 | 1 | 0% | 2,815 | 3,683 | +31% | 0 | 0 | — |
case-23 | pass→pass | 15,397 | 3,523 | -77% | 1 | 1 | 0% | 2,192 | 1,454 | -34% | 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 +13 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.