🧪 Cuda Kernel Refine
This skill helps you iteratively optimize CUDA kernels by profiling with nsys and ncu, classifying bottlenecks, and validating improvements.
npx playbooks add skill trevors/dot-claude --skill cuda-kernel-refine
About Cuda Kernel Refine
This skill helps you iteratively optimize CUDA kernels by profiling with nsys and ncu, classifying bottlenecks, and validating improvements.
At 19 words, this compact prompt gives your agent specialized testing & qa expertise with structured patterns and output formats. Install via CLI or copy the prompt below.
Use Cases
- Writing unit, integration, and end-to-end tests
- Setting up test coverage and CI pipelines
- Refactoring legacy code with confidence using tests
- Creating test plans and QA checklists
Example Prompts
System Prompt (19 words)
This skill helps you iteratively optimize CUDA kernels by profiling with nsys and ncu, classifying bottlenecks, and validating improvements.
Frequently Asked Questions
What is Cuda Kernel Refine?
Cuda Kernel Refine is a free testing & qa skill for AI coding agents. This skill helps you iteratively optimize CUDA kernels by profiling with nsys and ncu, classifying bottlenecks, and validating improvements.. It provides a specialized system prompt that configures your agent with testing & qa expertise.
How do I use Cuda Kernel Refine with Claude Code?
Run npx playbooks add skill trevors/dot-claude --skill cuda-kernel-refine in your terminal to install Cuda Kernel Refine into your Claude Code session. It works immediately after installation.
Which AI coding agents work with Cuda Kernel Refine?
Cuda Kernel Refine is compatible with Claude Code, Cursor, GitHub Copilot, Windsurf, OpenClaw, Cline, and any AI agent that supports custom system prompts or .cursorrules files.
Is Cuda Kernel Refine free to use?
Yes, Cuda Kernel Refine is completely free and open source. The full source is available on GitHub at https://github.com/trevors/dot-claude/tree/main/skills/cuda-kernel-refine. You only need a subscription to the AI agent you use it with.
