wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
A guide to optimizing GPU operations that apply the same calculation independently to each value, such as addition, multiplication, activation functions, and broadcasts.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
A set of complete examples showing how Triton CUDA kernels work inside PyTorch, including vector addition, matrix multiplication, layer normalization, and softmax.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
A guide to choosing the grid and block settings that control how a CUDA GPU kernel runs in parallel.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
A guide to optimizing CUDA GPU code for matrix multiplication, including batched multiplication and fully connected layers.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
A guide to improving how CUDA GPU kernels read and arrange data in memory, including shared memory, coalesced access, and avoiding bank conflicts.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
A general reference for optimizing and debugging Triton CUDA GPU kernels, including API limits and platform restrictions.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
A set of standard Triton patterns and templates for three common GPU tasks: element-by-element operations, reductions, and matrix multiplication.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
A guide to writing CUDA GPU code that combines many values into results such as sums, averages, maximums, and minimums. It also covers softmax, layer normalization, and log-softmax.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
Iterative GPU kernel optimization orchestrator for CUDA/CUTLASS/CuTe DSL/Triton kernels. Use for measured, one-change-at-a-time optimization loops with correctness, NCU profiling, KBS evidence, hypothesis discipline, hard iteration gates, final benchmarking, and a traceable report.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
Standalone kernel profiling skill for cuda-cpp, cute-dsl, cutlass, and triton implementations. Checks CUDA/PyTorch/Triton/CuTe DSL/CUTLASS/NCU/nsight-python readiness, optionally locks GPU clocks, validates correctness, collects Nsight Compute metrics with nsight-python, produces envcheck.md, correctness.md…
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
Optimize existing Triton kernels for NVIDIA TileIR backend on Blackwell GPUs (sm100+). Adds TileIR-specific autotune configs: occupancy, numctas, TMA descriptors. Covers kernel classification (dot-related, norm-like, elementwise, reduction), type-specific transformations, and PTX-vs-TileIR benchmarking. Triggered by…
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
ONLY for OpenAI Triton (@triton.jit) kernel development. NEVER use for CUDA C++ kernels, TileIR, or profiling tools (ncu, nsys). The user's request must involve Triton explicitly. Covers Triton-specific patterns: fused elementwise, reductions (softmax, LayerNorm, RMSNorm), tiled GEMM with triton.autotune, and flash…
wenyi-li/awesome-agent-kernel-skills
Skill Claude Code
A two-stage checker for GPU operator code. It first checks syntax, compilation, imports, and DSL rules, then compares generated output with a framework implementation using Torch or MindSpore on CUDA, Ascend, or CPU.
wenyi-li/awesome-agent-kernel-skills
Skill Claude Code
Use when the user asks about optimizing NVIDIA Blackwell (SM100, B200) or Hopper (SM90, H100) GPU kernels — tcgen05/TMEM/CLC/NVFP4/2-SM cooperative, warp specialization, FlashAttention-4, DeepGEMM, FlashMLA, MoE, grouped GEMM, CuTe-DSL/PTX/Triton on Blackwell, or wants concrete PR references from…
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
A workflow for generating, checking, and optimizing GPU operators across Triton, CUDA C, C++, and TileLang.
wenyi-li/awesome-agent-kernel-skills
Skill Claude Code
Unified GPU kernel operator generation and optimization skill. Automatically detects the target repository type (FlagGems, vLLM, or general Python/Triton) and dispatches to the appropriate specialized sub-skill. Includes operator generation, MCP-based iterative optimization, and feedback submission sub-skills. Use…
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
Framework-independent LLM serving benchmark skill for comparing SGLang, vLLM, TensorRT-LLM, or another serving framework. Use when a user wants to find the best deployment command for one model across multiple serving frameworks under the same workload, GPU budget, and latency SLA.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
Unified LLM torch-profiler triage skill for sglang, vllm, and TensorRT-LLM. Use it to inspect an existing trace.json(.gz) or profile directory, or to drive live profiling against a running server and return one three-table report with kernel, overlap-opportunity, and fuse-pattern tables.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
MI300/CDNA3 architecture guide for HIP/Triton optimization—MFMA variants, dual register files, data formats, sparsity, LDS/GWS, and best practices.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
CDNA3/MI300 HIP programming insights—chiplet/cache model, Infinity Cache, memory coherency, matrix cores, sparsity, and best practices.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
MI300 HIP programming differences vs NVIDIA—wavefront vs warp, memory hierarchy, MFMA usage, occupancy, and profiling pitfalls.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
Automated NCU (Nsight Compute) profiling workflow with full metrics collection and persistent storage.
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
Profiles a CUDA kernel with NCU and analyzes the resulting .ncu-rep report to diagnose performance bottlenecks and generate optimization recommendations. Use when the user provides a .cu file or a .ncu-rep file and asks for performance analysis, NCU profiling, or bottleneck diagnosis. If given a .cu file, runs NCU via…
wenyi-li/awesome-agent-kernel-skills
Skill Claude CodeCodex
A workflow for profiling CUDA kernels on NVIDIA B200 GPUs with Nsight Compute and using the results to diagnose performance problems.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: