Best GPU and accelerator computing skills for Claude Code, Codex and other agents.

Trains or fine-tunes language and vision models with TRL or Unsloth on Hugging Face Jobs cloud GPUs, then converts the results to GGUF.

huggingface/skills

Runs an acceptance-style evaluation of an ML GPU cluster: environment dump, matmul FLOPS, all-reduce, disk benchmarks and a dated markdown report.

stas00/ml-engineering

A skill your agent uses when working with Paddle 3.0 compiler full pipeline: SOT (Symbolic Opcode Translator) for bytecode-level dy2st graph capture, PIR (Paddle IR) for SSA-based intermediate…

PaddlePaddle/Paddle

Debug hanging issues in SGLang distributed inference (TP/PP/DP/EP).

sgl-project/sglang

Explains why editing CUTLASS fused-MHA headers in ONNX Runtime can leave stale CUDA kernels after an incremental build, and how to force and verify a real rebuild.

microsoft/onnxruntime

Choose 32-bit vs 64-bit index math in PyTorch CUDA kernels. An agent skill from pytorch/pytorch.

pytorch/pytorch

Runs evaluations of Hugging Face Hub models on local hardware with inspect-ai or lighteval, and helps choose between vLLM, Transformers and accelerate backends.

huggingface/skills

Adds Liger Kernel support for a new HuggingFace Transformers model, or modifies existing monkey-patching.

linkedin/Liger-Kernel

Operates the Inspire ML platform through its local `inspire` CLI: picking account, workspace and resources, launching notebooks, jobs and services, then cleaning up.

realZillionX/InspireSkill

Develops production-ready Triton kernels for Liger Kernel. An agent skill from linkedin/Liger-Kernel.

linkedin/Liger-Kernel

Workflow for porting an existing Triton kernel in fla/ops/ to Gluon (triton.experimental.gluon) to gain explicit control over tensor layouts, shared memory, async data movement (cp.async / TMA), MMA…

fla-org/flash-linear-attention

Optimizes the performance of existing Liger Kernel Triton kernels.

linkedin/Liger-Kernel

Plans memory headroom, works through out-of-memory failures and watches temperature and power during long ML training jobs on NVIDIA DGX Spark.

wshobson/agents

Reads SGLang or vLLM startup logs to show where GPU memory went and estimates how many concurrent requests fit at common token lengths.

BBuf/AI-Infra-Auto-Driven-SKILLS

Adapts and debugs Hugging Face or local models to run on vLLM with Ascend NPU, validates them by serving, and delivers the result as one signed commit.

vllm-project/vllm-ascend

Profiles, benchmarks and tunes AI training workloads on Moore Threads MUSA GPUs with a measurement-first process that keeps model behavior unchanged.

open-infra-skills/infra-skills

Profile AI inference workloads (vLLM, SGLang, TensorRT-LLM, PyTorch, any GPU application) with the Graphsignal profiler and read the results from its local /signals JSON endpoint.

graphsignal/graphsignal

CUDA kernel development, debugging, and performance optimization for Claude Code.

technillogue/ptx-isa-markdown

Iteratively optimize a CUDA/CUTLASS/Triton kernel only when strict on-device compilation, correctness, timing, and NCU evidence gates pass.

KernelFlow-ops/cuda-optimized-skill

Preflight checks and diagnosis for ten known failure modes of ML training on NVIDIA DGX Spark's GB10, spanning launch errors, memory, thermals, bandwidth and precision.

wshobson/agents

Diagnose failed, hung, slow, OOM, NaN, illegal-memory-access, NCCL, compilation, rollout, or training runs in AReno.

inclusionAI/AReno

Trains and fine-tunes object detection, image classification and SAM or SAM2 segmentation models on Hugging Face Jobs cloud GPUs and saves the results to the Hub.

huggingface/skills

Analyzes Torch Profiler traces from SGLang, vLLM and TensorRT-LLM servers into kernel attribution, overlap and fusion tables.

BBuf/AI-Infra-Auto-Driven-SKILLS

Write Metal/MPS kernels for PyTorch operators. An agent skill from pytorch/pytorch.

pytorch/pytorch

Guides building on the 0G Compute Network, a decentralized GPU marketplace for AI inference and fine-tuning, with SDK patterns and CLI commands.

internet-court/internet-court-skill

Shows how to organize PyTorch training with Lightning's LightningModule and Trainer, covering validation, DDP, callbacks and learning-rate scheduling.

Orchestra-Research/AI-Research-SKILLs

A skill your agent uses when writing, debugging, porting, reviewing, or optimizing CUDA C++ or PTX kernels; investigating CUDA Runtime or Driver API behavior; profiling kernels with Nsight Systems…

vipshop/cache-dit

Runs the ONNX Runtime transformers Python tests against a GPU wheel and proves the cuDNN flash attention path was used rather than a silent fallback.

microsoft/onnxruntime

A skill your agent uses when adding a new optimized kernel or operator to veomni/ops/.

ByteDance-Seed/VeOmni

Profile CUDA kernels with Nsight Compute on B200 / sm100. An agent skill from mit-han-lab/ncu-report-skill.

mit-han-lab/ncu-report-skill

Quick-reference help for fine-tuning language models with Axolotl, covering YAML configs, FSDP, context parallelism, compressed saves and dataset formats.

Orchestra-Research/AI-Research-SKILLs

Builds an operator-level compute template for an LLM and estimates FLOPs and MFU for a serving shape, with tensor shapes and parallelism what-if checks.

BBuf/AI-Infra-Auto-Driven-SKILLS

GPU optimization workflow using uipc.profile, uipc.profile.nsight, and Nsight Compute CLI.

spiriMirror/libuipc

Debug PyTorch 2 compiler stack failures including Dynamo graph breaks, Inductor codegen errors, AOTAutograd crashes, and accuracy mismatches.

pytorch/pytorch

Creates one-node GitHub merge-request variants of existing two-node GB200 functional tests in Megatron-LM, adjusting parallelism settings to fit four GPUs.

NVIDIA/Megatron-LM

Serverless GPU cloud platform for running ML workloads. An agent skill from Orchestra-Research/AI-Research-SKILLs.

Orchestra-Research/AI-Research-SKILLs

GPU-accelerated data curation for LLM training. An agent skill from Orchestra-Research/AI-Research-SKILLs.

Orchestra-Research/AI-Research-SKILLs

Optimizes and serves LLMs on NVIDIA GPUs with TensorRT-LLM, covering quantization, in-flight batching, multi-GPU parallelism and the trtllm-serve command.

Orchestra-Research/AI-Research-SKILLs

Write or review Triton kernels for vLLM, with practical guidance for generated-code inspection, launch grids, indexing, specialization, tuning, and representative performance validation.

guqiong96/Lvllm

Step-by-step tutorial for adding a new lightweight JIT CUDA kernel to sglang's jitkernel module

guqiong96/Lsglang

Shows how to build and run ONNX Runtime WebGPU provider tests on Linux with no GPU, using the Mesa lavapipe software Vulkan adapter, and where that approach falls short.

microsoft/onnxruntime

Runs ML training and batch jobs across clouds with SkyPilot, using spot instances, automatic region selection and managed recovery to cut GPU cost.

Orchestra-Research/AI-Research-SKILLs

Adds distributed and mixed-precision training to a PyTorch script with a few Accelerate lines, then launches it on one GPU, many GPUs or DeepSpeed and FSDP setups.

Orchestra-Research/AI-Research-SKILLs

基于 helpfile 工程经验，协助 agent 配置 CV 主机和边缘设备环境、编译视觉与推理依赖、接入摄像头视频并打包部署服务。适用于 Ubuntu、CentOS、Windows、macOS、Jetson、树莓派和 RK3399 的 CV 工程实施与故障排查，以及相关移动端配套工具；模型训练和纯算法设计不属于本技能主线。

LMIXR/CV_Deployment_skill

Run jcm on a Kubernetes GPU cluster — generate benchmark and production Job manifests, pick a comparable GPU, survive eviction, collect results.

climate-analytics-lab/jax-gcm

Shows how to launch distributed Megatron-LM training on a SLURM cluster: sbatch skeleton, torch.distributed.run setup, CUDA_DEVICE_MAX_CONNECTIONS rules and failure diagnosis.

NVIDIA/Megatron-LM

dstack is an open-source control plane for GPU provisioning and orchestration across GPU clouds, Kubernetes, and on-prem clusters.

dstackai/dstack

Loads large language models in 8-bit or 4-bit with bitsandbytes so they fit smaller GPUs, and sets up QLoRA fine-tuning on a 4-bit base model.

Orchestra-Research/AI-Research-SKILLs