Skip to content
#2728 Overall#321 in CodingVerified SKILL.md

Doca Gpunetio

Use this skill when the user is doing hands-on DOCA GPUNetIO programming — wiring a CUDA kernel on an NVIDIA GPU to a doca-eth queue via doca_gpu_eth_rxq / doca_gpu_eth_txq, standing up the per-CUDA-device doca_gpu context, designing the persistent CUDA kernel that drains the GPU-visible queue, running the dual capability check (DOCA cap-query plus cudaGetDeviceProperties), registering cudaMalloc pools via doca_buf_arr_create_*, or debugging DOCA_ERROR_* returns from the GPUNetIO API. Trigger even when the user does not explicitly mention "DOCA GPUNetIO" or "persistent kernel" — typical implicit phrasings include "CUDA kernel reading packets directly from the NIC", "GPU-initiated networking on BlueField", "DOCA_ERROR_DRIVER on doca_gpu_create", "nvidia_peermem not loaded", "kernel-per-packet is too slow", or "which GPU supports GPU-side packet I/O". Refuse and route elsewhere for general CUDA programming, DOCA Ethernet queue bring-up, DOCA DPA, or DOCA install — those belong to other skills.

Claude CodeCodexSKILL.md Official publisher
0Total downloads
0Active installs
3,046Community stars
Momentum rank

Overview

What It Does

Doca Gpunetio packages a focused coding & development workflow for an AI agent. Use this skill when the user is doing hands-on DOCA GPUNetIO programming — wiring a CUDA kernel on an NVIDIA GPU to a doca-eth queue via doca_gpu_eth_rxq / doca_gpu_eth_txq, standing up the per-CUDA-device doca_gpu context, designing the persistent CUDA kernel that drains the GPU-visible queue, running the dual capability check (DOCA cap-query plus cudaGetDeviceProperties), registering cudaMalloc pools via doca_buf_arr_create_*, or debugging DOCA_ERROR_* returns from the GPUNetIO API. Trigger even when the user does not explicitly mention "DOCA GPUNetIO" or "persistent kernel" — typical implicit phrasings include "CUDA kernel reading packets directly from the NIC", "GPU-initiated networking on BlueField", "DOCA_ERROR_DRIVER on doca_gpu_create", "nvidia_peermem not loaded", "kernel-per-packet is too slow", or "which GPU supports GPU-side packet I/O". Refuse and route elsewhere for general CUDA programming, DOCA Ethernet queue bring-up, DOCA DPA, or DOCA install — those belong to other skills. It is best suited to users who can review the resulting actions and provide only the accounts, files, or command access needed for the task. It is not a substitute for human approval on destructive, financial, security-sensitive, or public-facing actions.

Task ideas

Popular Use Cases

  • Review and improve a code change
  • Automate a repeatable development task
  • Investigate failures and prepare a fix
  • Work with repositories and developer tooling

Installation

Install this Agent Skill

Claude Code

npx skills add https://github.com/NVIDIA/skills --skill doca-gpunetio

Commands derived from the public GitHub SKILL.md record. Checked 2026-09-02. Review the source before running them.

Before you start

Requirements

Claude CodeRequired / review
CodexRequired / review
Public SKILL.md sourceRequired / review
Review instructions and requested permissions before installationRequired / review
Paid serviceCheck source
Supported systemCheck source

Popularity context

Why It’s Popular

Doca Gpunetio is a verified Agent Skill from NVIDIA with a public SKILL.md, compatible with Claude Code, Codex.

Alternatives

Similar Skills