webgpu-wgsl-compute-shaders

What problem does it solve?

This Skill removes the guesswork and compile-time failures that happen when writing WGSL compute shaders for WebGPU, especially around mandatory workgroup sizing, invocation IDs, shared memory, and synchronization.

Core Features & Use Cases

• Correctly applies @compute and workgroup sizing for one-, two-, and three-dimensional kernels.
• Explains how to use global_invocation_id, local_invocation_id, workgroup_id, and num_workgroups without confusing workgroup size with dispatch count.
• Shows safe patterns for workgroup memory, atomic counters, workgroup barriers, storage barriers, and optional subgroup operations.
• Useful for reductions, scans, counters, image-processing kernels, and other deterministic compute workflows.

Quick Start

Use the webgpu-wgsl-compute-shaders skill to draft a WebGPU compute shader that includes the right workgroup size, bounds checks, shared-memory synchronization, and atomic or subgroup handling for your specific kernel.