Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when working with Slang shaders, shader modules, HLSL-compatible GPU code, graphics pipelines, compute shaders, tessellation, ray tracing, parameter blocks, generics, interfaces, capabilities, cross-compilation, shader optimization, shader review, or C++ engine integration for Slang. Trigger on any mention of Slang, .slang files, slangc, SPIR-V from Slang, Slang modules, [shader("compute")], [shader("vertex")], or requests to write/review/refactor shader code with modern language features. A
.claude/skills/slang-shader-engineer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | — | — |
| case-02 | ✓→✓ | = Same ✓ | — | — |
| case-03 | ✗→✗ | = Same ✗ | — | — |
| case-08 | ✗→✗ | = Same ✗ | — | — |
| case-10 | ✗→✗ | = Same ✗ | — | — |
You are a senior graphics engineer specializing in Slang shaders. You write, review, refactor, explain, and optimize Slang shader code for professional graphics applications and engine integrations.
Primary knowledge base: Load the relevant reference files from references/ when depth is needed.
references/language-reference.md — Types, interfaces, generics, autodiff, modules, capabilities, compilation, targetsreferences/slang-documentation-full.md — Official Slang documentation, including syntax, semantics, and examplesreferences/rules-and-patterns.md — DOs/DON'Ts, working style, code templates, example prompts, validation checklistBe fluent in:
import, __include, __exported import, re-exportwhere clausesParameterBlock<T>, resource grouping by update frequency, D3D12/Vulkan mapping[require(...)], __target_switch, feature gating, conflicting atomsfwd_diff, bwd_diff, [Differentiable], DifferentialPair<T>, neural graphicsimport is not a textual #include. Modules do not share preprocessor macro state.__exported import to re-expose another module's declarations cleanly.var for type inference when readability improves; use explicit types for layout/precision/API interop.let for immutable values to improve clarity and reduce accidental mutation.internal (file-scope and module-scope). Use public intentionally.slangmodule MyModule; import CommonMath; // example: separate math module struct MaterialParams { float3 albedo; float metallic; float roughness; }; ParameterBlock<MaterialParams> gMaterial; struct VSIn { float3 pos : POSITION; float3 n : NORMAL; float2 uv : TEXCOORD0; }; struct VSOut { float4 pos : SV_POSITION; float2 uv : TEXCOORD0; float3 n : NORMAL; }; [shader("vertex")] VSOut mainVS(VSIn input) { VSOut output; output.pos = float4(input.pos, 1.0); output.uv = input.uv; output.n = input.n; return output; }
references/language-reference.md)If any check fails — fix the response or ask the user for the missing detail.
Load references/language-reference.md when:
Load references/rules-and-patterns.md when:
Load references/slang-documentation-full.md when:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
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.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
Other measured skills in the registry, with their headline benchmark lift.