AI & LLM Engineering · CUDA · For developers

Replay-first debug flow for SGLang serving problems. An agent skill from sgl-project/sglang.

sgl-project/sglang

Debug AOTInductor (AOTI) errors and crashes. An agent skill from pytorch/pytorch.

pytorch/pytorch

A skill your agent uses when needing to compile, rebuild, or install Paddle from source after code changes.

PaddlePaddle/Paddle

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

Guide for adding Triton-backed CUDA operators to FastLLM. An agent skill from ztxz16/fastllm.

ztxz16/fastllm

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

LeetCUDA 中文技术书（584 页，XeLaTeX 源）按需查阅 skill——写、优化、调试或 review CUDA C++/PTX kernel 时的权威参考路由层。当任务涉及：GPU 架构/Roofline/ occupancy、向量化与 coalescing、warp/block reduce、softmax（online/LSE merge）、…

xlite-dev/LeetCUDA

Builds ONNX Runtime from source with its build scripts, explaining the update, build and test phases, key flags and where the build output lands.

microsoft/onnxruntime

Biohub ESMFold2 / ESMFold2-Fast all-atom co-folding (Candido et al.

JimLiu/science-skills

Condensed debugging method and tool recipes for Unix, Python and PyTorch programs: crashes, hangs, segfaults, wrong output, CUDA OOM, NaN values and slowness.

stas00/the-art-of-debugging

Lets a Hermes agent run Autocontext scenarios, inspect Hermes curator state, export reusable knowledge and prepare local MLX or CUDA training data through the autoctx CLI.

greyhaven-ai/autocontext

Builds ExecuTorch from source: the Python package, C++ runtime, model runners, Android and iOS cross-compilation and backend-specific builds, with environment checks.

pytorch/executorch

Build, debug, and interpret vLLM GPU kernel microbenchmarks for CUDA, Triton, and CuteDSL, including CUPTI timing, correctness checks, generated-code inspection, multi-GPU measurements, and SOL…

guqiong96/Lvllm

Diagnose OpenAI-compatible model-serving failures from symptoms, endpoint reports, explicit configuration files, or logs while preserving evidence status and requiring confirm/refute checks.

Blackwellboy/model-serving-minefield

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

A skill your agent uses for ANY bug, error, crash, wrong output, loss divergence, gradient explosion, test failure, CUDA error, distributed training hang, checkpoint load failure, or unexpected…

ByteDance-Seed/VeOmni

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

technillogue/ptx-isa-markdown

在 Paddle 代码库中定位问题并输出高质量调试报告的专用技能。当遇到以下场景时优先使用：(1) Paddle 框架 bug 调试，(2) 算子实现问题排查，(3) 训练脚本异常诊断，(4) 分布式训练故障定位，(5) CUDA/GPU 相关错误处理，(6) 需要生成结构化调试报告。

PaddlePaddle/Paddle

Finds llama.cpp-compatible GGUF models on the Hugging Face Hub, picks a quantization for your hardware and launches them with llama-cli or llama-server.

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

Drive a single-operator optimization campaign per .agents/guidance/OPTIMIZATIONGUIDELINES.md, with a deterministically enforced definition-of-done and versioned MR summary.

CVCUDA/CV-CUDA

Ensures new Attention Gym eager, Triton, CuTeDSL, and external-library implementations are torch.compile-friendly and correctly registered.

meta-pytorch/attention-gym

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

pytorch/pytorch

Drive an agentic loop that iteratively optimizes a GPU kernel for maximum speedup.

TongmingLAIC/AKO4ALL

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

PaddlePaddle (飞桨) C++ 算子开发指南。提供从 YAML 配置、InferMeta 函数、Kernel 实现、Python API 封装、单元测试到编译验证的完整算子开发流程指导。在以下场景使用此 skill：(1) 为 Paddle 框架新增 C++ 算子 (2) 修改或调试已有 Paddle 算子 (3) 编写算子的 YAML…

PaddlePaddle/Paddle

A skill your agent uses when optimizing or generating a CUDA megakernel for a HuggingFace Llama-family model with AutoMegaKernel (AMK), drives the correctness-gated propose - eval - keep/revert loop…

RightNow-AI/AutoMegaKernel

Diagnose and fix Cosmos3 environment, installation, and runtime errors.

NVIDIA/cosmos-framework

Fixes tool environment setup failures in proto-tools, either just for the current machine (eject the tool's standalone dir, patch it, and point PROTO<TOOLKITSTANDALONEDIR at it; works for any…

evo-design/proto-tools

Write, debug, and optimize CUTLASS, CuTe, and CuTeDSL GPU kernels from local upstream source, examples, and headers.

slowlyC/agent-gpu-skills

Train a custom VideoHighlighter action or object model from a few videos the user provides — cut into samples, sort with CLIP, review contact sheets, build, train, install only if better.

Aseiel/VideoHighlighter

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

spiriMirror/libuipc

Set up the NVIDIA "Build an Agent" DevX workshop as a working JupyterLab environment from INSIDE a locked-down OpenShell/NemoClaw sandbox, and hand the user the token URL + access commands.

brevdev/workshop-build-an-agent

A skill your agent uses when writing, modifying, porting, or optimizing CuTe DSL GPU kernels in Python; reading CuTe DSL API reference material; integrating a CuTe DSL kernel into a project; or…

vipshop/cache-dit

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

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

guqiong96/Lsglang

Register, extend, or audit instructions in the table-driven T.ptx dialect (python/tvm/backend/cuda/ptx/table.py), and move the table to a newer PTX ISA version.

mlc-ai/relax

Scaffold new Celeritas source and test files with the required copyright header and register them in CMake.

celeritas-project/celeritas

Integrate a new text-to-speech model into vLLM-Omni from HuggingFace reference implementation through production-ready serving with streaming and CUDA graph acceleration.

vllm-project/vllm-omni

Guided workflow for adding a new model architecture to llama.cpp.

JakeATX/llamAmpere

A skill your agent uses when writing, debugging, porting, reviewing, or optimizing CUTLASS or CuTe C++ kernels and templates; navigating CUTLASS examples, collectives, epilogues, pipelines, GEMM…

vipshop/cache-dit

A skill your agent uses for performance profiling and optimization.

ByteDance-Seed/VeOmni

Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware.

Orchestra-Research/AI-Research-SKILLs

Step-by-step playbook for migrating a pytorch/pytorch .github/workflows/.yml from EC2 to OSDC (ARC) runners — covers both dial-up and 100% opt-in patterns, with the inputs that must be plumbed…

pytorch/test-infra

A skill your agent uses when something is wrong: Search() hangs, all evaluations return INVALIDSCORE, scores aren't improving, every config returns the same number, ptxas errors fill the log, CV% is…

NVIDIA/CompileIQ

Covers the rules for writing Gradio Spaces on ZeroGPU hardware: the @spaces.GPU decorator, duration and quota tuning, process isolation and CUDA build limits.

huggingface/skills