[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7BKA49BBJYHGVR6FHC6KVF8":3,"skill-same:01M7BKA49BBJYHGVR6FHC6KVF8":89},{"skillNo":4,"name":5,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":37,"commitSha":38,"skillDirectory":39,"forksCount":13,"repositoryHtmlUrl":40,"repositoryHomepageUrl":41,"ownerHtmlUrl":42,"ownerBlogUrl":6,"ownerDescription":43,"updatedAt":44,"install":45,"resources":58},"01M7BKA49BBJYHGVR6FHC6KVF8","tao-setup-nvidia-gpu-host",null,"YYC-Cube\u002FYYC3-AI-Agent-Archive","YYC-Cube","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F235006543?v=4","01M79J2PH4SWD2X0825X6PN748","yyc3","01M79K9J2WD44BBY4662XKNZRB",2,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Fskills-hub\u002Fai-ml\u002Fnvidia-skills\u002Fskills\u002Ftao-setup-nvidia-gpu-host\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: tao-setup-nvidia-gpu-host\ndescription: >-\ncategory: development-code\nversion: 1.0.0\n  Host setup for TAO GPU backends. Checks and, after user approval, installs\n  NVIDIA driver branch 580, CUDA Toolkit 13.0, and NVIDIA Container Toolkit\n  1.19.0 for Docker\u002Flocal-Docker and Kubernetes GPU worker hosts. The\n  `--check-only` path works on any Linux distribution; `--install` automates\n  debian-family (Ubuntu\u002FDebian\u002FPop!_OS\u002FMint\u002FZorin\u002FRaspbian), rhel-family\n  (Fedora\u002FRHEL\u002FRocky\u002FAlmaLinux), and suse-family (openSUSE\u002FSLES) hosts, and\n  prints actionable manual-install steps for everything else.\nlicense: Apache-2.0\ncompatibility: Runs `--check-only` on any Linux distribution. `--install` automates Ubuntu 22.04\u002F24.04 + Debian 12 (apt), Fedora + RHEL\u002FRocky\u002FAlmaLinux 9\u002F10 (dnf), and openSUSE Leap \u002F SLES 15 (zypper). Requires sudo\u002Froot, internet access to NVIDIA package repositories (and download.docker.com on rhel-family), and an x86_64 or aarch64 (sbsa) host. Other distributions (Arch, Alpine, Gentoo, NixOS, …) get a clear error that names the version targets and the NVIDIA install-guide URL.\nmetadata:\n  author: NVIDIA Corporation\n  version: \"0.1.0\"\nallowed-tools: Read Bash\ntags:\n- setup\n- nvidia\n- cuda\n- docker\n- kubernetes\n---\n\n# NVIDIA GPU Host Setup\n\nUse this setup skill before TAO workflows run on the `docker`, `local-docker`,\nor `kubernetes` backend. It standardizes the host GPU runtime on:\n\n- NVIDIA driver branch `580` (open kernel module preferred)\n- CUDA Toolkit package `cuda-toolkit-13-0`\n- NVIDIA Container Toolkit `1.19.0`\n- Docker engine — only installed for `docker` \u002F `local-docker` backends and\n  only when Docker is missing. The package picked depends on the distro\n  family (`docker.io` on Debian-family by default, `moby-engine` \u002F\n  `docker-ce` from `download.docker.com` on RHEL-family, `docker` on\n  SUSE-family). Pass `--skip-docker-install` to opt out.\n\nThe check is safe and read-only by default — it works on any Linux\ndistribution because it only probes `nvidia-smi`, the CUDA toolkit path,\nthe installed container-toolkit package version (via `dpkg`\u002F`rpm`\u002Fthe\n`nvidia-ctk` binary version), and the Docker daemon's NVIDIA runtime.\n\nInstallation must be explicitly authorized by the user and rerun with\n`--install`. The install path is automated for these distro families:\n\n| Family | Tested distros | Manager | Notes |\n|---|---|---|---|\n| debian | Ubuntu 22.04 \u002F 24.04, Debian 12 (and derivatives Pop!_OS, Mint, Zorin, Raspbian, KDE Neon, etc. via `UBUNTU_CODENAME` \u002F `VERSION_CODENAME`) | `apt-get` | Adds NVIDIA `cuda-keyring` + Container Toolkit `.list`. Docker via `docker.io` (override `$DOCKER_PACKAGE_DEBIAN`). |\n| rhel | Fedora 39+, RHEL \u002F Rocky \u002F AlmaLinux 9 and 10 | `dnf` (or `yum`) | Adds NVIDIA `cuda-\u003Cdistro>.repo` + Container Toolkit `.repo`. Docker via Fedora `moby-engine` when available, otherwise `docker-ce` from `download.docker.com`. |\n| suse | openSUSE Leap 15, SLES 15 | `zypper` | Adds the same NVIDIA `.repo` files. Docker via the distribution `docker` package. |\n| other (Arch, Alpine, Gentoo, NixOS, FreeBSD, …) | n\u002Fa | n\u002Fa | `--install` exits with a clear error listing the version targets and the NVIDIA install-guide URLs. Install manually, then rerun `--check-only`. |\n\n## Quick Start\n\nFrom the skill bank root:\n\n```bash\n# Check the local Docker backend host.\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend docker --check-only\n\n# Install or repair after user approval (prompts for confirmation; see the note below for non-interactive runs).\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend docker --install\n\n# Check a Kubernetes GPU worker host.\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend kubernetes --check-only\n```\n\n> ⚠️ **Note — running non-interactively (agent \u002F skill runs):** a skill run has\n> no terminal, so the installer's `Continue? [y\u002FN]` confirmation cannot be\n> answered. After running `--check-only` to preview what is missing and getting\n> the user's explicit approval, append the assume-yes flag (`--yes`) to the\n> `--install` command so it proceeds without a prompt. That auto-confirms\n> installation of system packages (NVIDIA driver branch 580, CUDA Toolkit 13.0,\n> NVIDIA Container Toolkit, and — for Docker backends — Docker) and modifies the\n> host: it adds NVIDIA package repositories, may restart Docker, and adds the\n> invoking user to the `docker` group, so only do this on a host you control and\n> have the privileges to change. When a person runs `--install` directly at a\n> terminal, the script instead prompts with the exact package list before making\n> any changes.\n\nIn an installed plugin copy that exposes `skills\u002F`, use:\n\n```bash\nbash skills\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend docker --check-only\n```\n\n## Workflow Contract\n\nDocker and Kubernetes workflows must run the check before submitting GPU work:\n\n```bash\nSETUP_SCRIPT=\"${TAO_SKILL_BANK_ROOT:-$PWD}\u002Fskills\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh\"\n[ -x \"$SETUP_SCRIPT\" ] || SETUP_SCRIPT=\"${TAO_SKILL_BANK_ROOT:-$PWD}\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh\"\n\nbash \"$SETUP_SCRIPT\" --backend docker --check-only || {\n  echo \"MISSING: TAO GPU host runtime is not ready.\"\n  echo \"After user approval, run (append --yes for non-interactive agent runs):\"\n  echo \"  bash \\\"$SETUP_SCRIPT\\\" --backend docker --install\"\n  exit 1\n}\n```\n\nNever install silently. If the check fails, explain what is missing, ask the\nuser to authorize the fix, then run the install command and rerun the check.\n\n## What The Installer Does\n\nThe installer dispatches on the detected distribution family. On every\nsupported family it adds NVIDIA's CUDA and Container Toolkit repositories\n(if missing), installs the pinned runtime packages, optionally installs\nDocker, wires the NVIDIA Docker runtime, and adds the invoking user to\nthe `docker` group.\n\nCommon steps (all families):\n\n1. Adds NVIDIA's CUDA repository if missing (apt `cuda-keyring` deb,\n   `cuda-\u003Cdistro>.repo` for dnf\u002Fzypper).\n2. Adds NVIDIA's Container Toolkit repository if missing (`.list` for apt,\n   `.repo` for dnf\u002Fzypper).\n3. Installs the matching kernel header \u002F devel package for the running\n   kernel.\n4. Installs the driver branch 580 packages, `cuda-toolkit-13-0`, and the\n   Container Toolkit pinned to `1.19.0` (the dpkg-suffixed `1.19.0-1` is\n   the same upstream version expressed for apt).\n5. For Docker backends and when Docker is missing, installs Docker\n   (override \u002F opt-out flags below), enables\u002Fstarts the daemon, then runs\n   `nvidia-ctk runtime configure --runtime=docker` and restarts Docker\n   when `systemctl` is available.\n6. Adds the invoking user (`$SUDO_USER` if available, else `$USER`) to the\n   `docker` group so subsequent shells can run `docker` without `sudo` —\n   opt out with `--skip-docker-group`. **The new group membership does not\n   take effect in the current shell**: log out and back in, or run\n   `newgrp docker` in each new shell.\n7. Attempts `modprobe nvidia` so verification can pass before reboot.\n\nFamily-specific package selections:\n\n| Step | debian-family | rhel-family | suse-family |\n|---|---|---|---|\n| Kernel headers | `linux-headers-$(uname -r)` | `kernel-devel-$(uname -r)`, `kernel-headers-$(uname -r)` | `kernel-default-devel` |\n| Driver | `nvidia-driver-pinning-580`, `nvidia-open-580` (override: `$NVIDIA_DRIVER_PACKAGE_DEBIAN`) | `nvidia-driver-cuda`, `kmod-nvidia-open-dkms` (override: `$NVIDIA_DRIVER_PACKAGE_RHEL`, `$NVIDIA_DRIVER_KMOD_RHEL`) | `nvidia-open-driver-G06-signed-kmp-default` (override: `$NVIDIA_DRIVER_PACKAGE_SUSE`) |\n| CUDA toolkit | `cuda-toolkit-13-0` | `cuda-toolkit-13-0` | `cuda-toolkit-13-0` |\n| Container Toolkit | `nvidia-container-toolkit=1.19.0-1` + base\u002Ftools\u002Flibs | `nvidia-container-toolkit-1.19.0` + base\u002Ftools\u002Flibs | same as rhel |\n| Docker | `docker.io` (override: `$DOCKER_PACKAGE_DEBIAN`) | `moby-engine`+`moby-cli` on Fedora when available, else `docker-ce docker-ce-cli containerd.io` from `download.docker.com` | `docker` |\n\n## Verification\n\nAfter installation, verify:\n\n```bash\nnvidia-smi\n\u002Fusr\u002Flocal\u002Fcuda-13.0\u002Fbin\u002Fnvcc --version\ndocker info --format '{{json .Runtimes}}' | grep nvidia\nsudo docker run --rm --runtime=nvidia --gpus all ubuntu nvidia-smi\n```\n\nExpected `nvidia-smi` output includes driver `580.x` and CUDA Version `13.0`.\nExpected `nvcc` output includes `release 13.0`.\n\n## Kubernetes Notes\n\nFor self-managed Kubernetes clusters, run the host installer on every GPU\nworker node or bake the same package set into the node image before installing\nthe NVIDIA GPU Operator or device plugin.\n\nThe workflow check also warns if `kubectl` is available but the cluster reports\nno `nvidia.com\u002Fgpu` allocatable capacity. In that case, install\u002Fconfigure the\nNVIDIA GPU Operator after the worker host runtime is ready:\n\n```bash\nhelm repo add nvidia https:\u002F\u002Fhelm.ngc.nvidia.com\u002Fnvidia\nhelm repo update\nhelm install --wait gpu-operator -n gpu-operator --create-namespace nvidia\u002Fgpu-operator\n```\n\nManaged Kubernetes providers may own driver installation through node images or\nGPU Operator policy. Do not overwrite a provider-managed GPU node without user\napproval and a rollback plan.\n\n## Failure Modes\n\n**Unsupported distribution family**: `--install` automates debian-, rhel-,\nand suse-family hosts. On Arch, Alpine, Gentoo, NixOS, FreeBSD, or anything\nwithout `\u002Fetc\u002Fos-release` (e.g. macOS), the script exits with a clear error\nthat lists the four version targets and the upstream NVIDIA install-guide\nURLs:\n\n- `https:\u002F\u002Fdocs.nvidia.com\u002Fcuda\u002Fcuda-installation-guide-linux\u002F`\n- `https:\u002F\u002Fdocs.nvidia.com\u002Fdatacenter\u002Fcloud-native\u002Fcontainer-toolkit\u002Flatest\u002Finstall-guide.html`\n- `https:\u002F\u002Fdocs.docker.com\u002Fengine\u002Finstall\u002F`\n\nInstall those four pieces using your distribution's package manager and\nrerun the script with `--check-only` to verify. The check is universally\nportable — it only queries the binaries \u002F package databases — so once the\nruntime is in place the workflow contract is satisfied regardless of the\nunderlying distro.\n\n**Unsupported Ubuntu\u002FDebian derivative**: When `ID` is e.g. `pop`, `mint`,\n`zorin`, `raspbian`, or another debian-family derivative, the script maps\nthe host onto the upstream Ubuntu\u002FDebian CUDA repo via `UBUNTU_CODENAME` \u002F\n`VERSION_CODENAME` (`focal`\u002F`jammy`\u002F`noble` → Ubuntu 20.04\u002F22.04\u002F24.04;\n`bullseye`\u002F`bookworm`\u002F`trixie` → Debian 11\u002F12\u002F12). If the host's codename\ndoesn't match a known upstream release, `--install` exits with the same\nmanual-install guidance described above.\n\n**Docker not installed**: `--check-only` reports `MISSING: Docker is not\ninstalled` and prints the exact rerun command appropriate to the detected\ndistro family. The default `--install` path installs Docker (`docker.io` \u002F\n`moby-engine` \u002F `docker-ce` \u002F `docker` depending on family), enables\u002Fstarts\nthe daemon, configures the NVIDIA runtime, and adds the invoking user to\nthe `docker` group. If you prefer to manage Docker yourself, install it\nbefore rerunning the script or pass `--skip-docker-install`.\n\n**Docker installed but `docker run` still needs sudo**: The script adds the\ninvoking user to the `docker` group, but Linux only refreshes group\nmembership on a new login session. Log out and back in, or run\n`newgrp docker` in each new shell, until the new membership is active.\n\n**Docker runtime still missing**: Restart Docker, then rerun\n`nvidia-ctk runtime configure --runtime=docker`.\n\n**Driver branch detected != 580**: The driver-branch pin is exact on\ndebian-family (`nvidia-open-580`). On rhel-\u002Fsuse-family the script\ninstalls the latest open driver shipped in NVIDIA's CUDA 13.0 repo for\nthe detected distro, which is always ≥ 580. If your host needs a stricter\npin, set `$NVIDIA_DRIVER_PACKAGE_RHEL` \u002F `$NVIDIA_DRIVER_KMOD_RHEL` \u002F\n`$NVIDIA_DRIVER_PACKAGE_SUSE` to the exact package names you want before\nrunning `--install`.\n\n**Driver installed but `nvidia-smi` fails**: Load the module with\n`sudo modprobe nvidia` or reboot. Secure Boot may require MOK enrollment on\nsystems where it is enabled.\n\n**Kubernetes still has no GPU capacity**: Confirm the driver works on each GPU\nnode with `nvidia-smi`, then check the GPU Operator\u002Fdevice plugin pods and node\nlabels.\n","# NVIDIA GPU Host Setup\n\nUse this setup skill before TAO workflows run on the `docker`, `local-docker`,\nor `kubernetes` backend. It standardizes the host GPU runtime on:\n\n- NVIDIA driver branch `580` (open kernel module preferred)\n- CUDA Toolkit package `cuda-toolkit-13-0`\n- NVIDIA Container Toolkit `1.19.0`\n- Docker engine — only installed for `docker` \u002F `local-docker` backends and\n  only when Docker is missing. The package picked depends on the distro\n  family (`docker.io` on Debian-family by default, `moby-engine` \u002F\n  `docker-ce` from `download.docker.com` on RHEL-family, `docker` on\n  SUSE-family). Pass `--skip-docker-install` to opt out.\n\nThe check is safe and read-only by default — it works on any Linux\ndistribution because it only probes `nvidia-smi`, the CUDA toolkit path,\nthe installed container-toolkit package version (via `dpkg`\u002F`rpm`\u002Fthe\n`nvidia-ctk` binary version), and the Docker daemon's NVIDIA runtime.\n\nInstallation must be explicitly authorized by the user and rerun with\n`--install`. The install path is automated for these distro families:\n\n| Family | Tested distros | Manager | Notes |\n|---|---|---|---|\n| debian | Ubuntu 22.04 \u002F 24.04, Debian 12 (and derivatives Pop!_OS, Mint, Zorin, Raspbian, KDE Neon, etc. via `UBUNTU_CODENAME` \u002F `VERSION_CODENAME`) | `apt-get` | Adds NVIDIA `cuda-keyring` + Container Toolkit `.list`. Docker via `docker.io` (override `$DOCKER_PACKAGE_DEBIAN`). |\n| rhel | Fedora 39+, RHEL \u002F Rocky \u002F AlmaLinux 9 and 10 | `dnf` (or `yum`) | Adds NVIDIA `cuda-\u003Cdistro>.repo` + Container Toolkit `.repo`. Docker via Fedora `moby-engine` when available, otherwise `docker-ce` from `download.docker.com`. |\n| suse | openSUSE Leap 15, SLES 15 | `zypper` | Adds the same NVIDIA `.repo` files. Docker via the distribution `docker` package. |\n| other (Arch, Alpine, Gentoo, NixOS, FreeBSD, …) | n\u002Fa | n\u002Fa | `--install` exits with a clear error listing the version targets and the NVIDIA install-guide URLs. Install manually, then rerun `--check-only`. |\n\n## Quick Start\n\nFrom the skill bank root:\n\n```bash\n# Check the local Docker backend host.\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend docker --check-only\n\n# Install or repair after user approval (prompts for confirmation; see the note below for non-interactive runs).\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend docker --install\n\n# Check a Kubernetes GPU worker host.\nbash skills\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend kubernetes --check-only\n```\n\n> ⚠️ **Note — running non-interactively (agent \u002F skill runs):** a skill run has\n> no terminal, so the installer's `Continue? [y\u002FN]` confirmation cannot be\n> answered. After running `--check-only` to preview what is missing and getting\n> the user's explicit approval, append the assume-yes flag (`--yes`) to the\n> `--install` command so it proceeds without a prompt. That auto-confirms\n> installation of system packages (NVIDIA driver branch 580, CUDA Toolkit 13.0,\n> NVIDIA Container Toolkit, and — for Docker backends — Docker) and modifies the\n> host: it adds NVIDIA package repositories, may restart Docker, and adds the\n> invoking user to the `docker` group, so only do this on a host you control and\n> have the privileges to change. When a person runs `--install` directly at a\n> terminal, the script instead prompts with the exact package list before making\n> any changes.\n\nIn an installed plugin copy that exposes `skills\u002F`, use:\n\n```bash\nbash skills\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh --backend docker --check-only\n```\n\n## Workflow Contract\n\nDocker and Kubernetes workflows must run the check before submitting GPU work:\n\n```bash\nSETUP_SCRIPT=\"${TAO_SKILL_BANK_ROOT:-$PWD}\u002Fskills\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh\"\n[ -x \"$SETUP_SCRIPT\" ] || SETUP_SCRIPT=\"${TAO_SKILL_BANK_ROOT:-$PWD}\u002Fplatform\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh\"\n\nbash \"$SETUP_SCRIPT\" --backend docker --check-only || {\n  echo \"MISSING: TAO GPU host runtime is not ready.\"\n  echo \"After user approval, run (append --yes for non-interactive agent runs):\"\n  echo \"  bash \\\"$SETUP_SCRIPT\\\" --backend docker --install\"\n  exit 1\n}\n```\n\nNever install silently. If the check fails, explain what is missing, ask the\nuser to authorize the fix, then run the install command and rerun the check.\n\n## What The Installer Does\n\nThe installer dispatches on the detected distribution family. On every\nsupported family it adds NVIDIA's CUDA and Container Toolkit repositories\n(if missing), installs the pinned runtime packages, optionally installs\nDocker, wires the NVIDIA Docker runtime, and adds the invoking user to\nthe `docker` group.\n\nCommon steps (all families):\n\n1. Adds NVIDIA's CUDA repository if missing (apt `cuda-keyring` deb,\n   `cuda-\u003Cdistro>.repo` for dnf\u002Fzypper).\n2. Adds NVIDIA's Container Toolkit repository if missing (`.list` for apt,\n   `.repo` for dnf\u002Fzypper).\n3. Installs the matching kernel header \u002F devel package for the running\n   kernel.\n4. Installs the driver branch 580 packages, `cuda-toolkit-13-0`, and the\n   Container Toolkit pinned to `1.19.0` (the dpkg-suffixed `1.19.0-1` is\n   the same upstream version expressed for apt).\n5. For Docker backends and when Docker is missing, installs Docker\n   (override \u002F opt-out flags below), enables\u002Fstarts the daemon, then runs\n   `nvidia-ctk runtime configure --runtime=docker` and restarts Docker\n   when `systemctl` is available.\n6. Adds the invoking user (`$SUDO_USER` if available, else `$USER`) to the\n   `docker` group so subsequent shells can run `docker` without `sudo` —\n   opt out with `--skip-docker-group`. **The new group membership does not\n   take effect in the current shell**: log out and back in, or run\n   `newgrp docker` in each new shell.\n7. Attempts `modprobe nvidia` so verification can pass before reboot.\n\nFamily-specific package selections:\n\n| Step | debian-family | rhel-family | suse-family |\n|---|---|---|---|\n| Kernel headers | `linux-headers-$(uname -r)` | `kernel-devel-$(uname -r)`, `kernel-headers-$(uname -r)` | `kernel-default-devel` |\n| Driver | `nvidia-driver-pinning-580`, `nvidia-open-580` (override: `$NVIDIA_DRIVER_PACKAGE_DEBIAN`) | `nvidia-driver-cuda`, `kmod-nvidia-open-dkms` (override: `$NVIDIA_DRIVER_PACKAGE_RHEL`, `$NVIDIA_DRIVER_KMOD_RHEL`) | `nvidia-open-driver-G06-signed-kmp-default` (override: `$NVIDIA_DRIVER_PACKAGE_SUSE`) |\n| CUDA toolkit | `cuda-toolkit-13-0` | `cuda-toolkit-13-0` | `cuda-toolkit-13-0` |\n| Container Toolkit | `nvidia-container-toolkit=1.19.0-1` + base\u002Ftools\u002Flibs | `nvidia-container-toolkit-1.19.0` + base\u002Ftools\u002Flibs | same as rhel |\n| Docker | `docker.io` (override: `$DOCKER_PACKAGE_DEBIAN`) | `moby-engine`+`moby-cli` on Fedora when available, else `docker-ce docker-ce-cli containerd.io` from `download.docker.com` | `docker` |\n\n## Verification\n\nAfter installation, verify:\n\n```bash\nnvidia-smi\n\u002Fusr\u002Flocal\u002Fcuda-13.0\u002Fbin\u002Fnvcc --version\ndocker info --format '{{json .Runtimes}}' | grep nvidia\nsudo docker run --rm --runtime=nvidia --gpus all ubuntu nvidia-smi\n```\n\nExpected `nvidia-smi` output includes driver `580.x` and CUDA Version `13.0`.\nExpected `nvcc` output includes `release 13.0`.\n\n## Kubernetes Notes\n\nFor self-managed Kubernetes clusters, run the host installer on every GPU\nworker node or bake the same package set into the node image before installing\nthe NVIDIA GPU Operator or device plugin.\n\nThe workflow check also warns if `kubectl` is available but the cluster reports\nno `nvidia.com\u002Fgpu` allocatable capacity. In that case, install\u002Fconfigure the\nNVIDIA GPU Operator after the worker host runtime is ready:\n\n```bash\nhelm repo add nvidia https:\u002F\u002Fhelm.ngc.nvidia.com\u002Fnvidia\nhelm repo update\nhelm install --wait gpu-operator -n gpu-operator --create-namespace nvidia\u002Fgpu-operator\n```\n\nManaged Kubernetes providers may own driver installation through node images or\nGPU Operator policy. Do not overwrite a provider-managed GPU node without user\napproval and a rollback plan.\n\n## Failure Modes\n\n**Unsupported distribution family**: `--install` automates debian-, rhel-,\nand suse-family hosts. On Arch, Alpine, Gentoo, NixOS, FreeBSD, or anything\nwithout `\u002Fetc\u002Fos-release` (e.g. macOS), the script exits with a clear error\nthat lists the four version targets and the upstream NVIDIA install-guide\nURLs:\n\n- `https:\u002F\u002Fdocs.nvidia.com\u002Fcuda\u002Fcuda-installation-guide-linux\u002F`\n- `https:\u002F\u002Fdocs.nvidia.com\u002Fdatacenter\u002Fcloud-native\u002Fcontainer-toolkit\u002Flatest\u002Finstall-guide.html`\n- `https:\u002F\u002Fdocs.docker.com\u002Fengine\u002Finstall\u002F`\n\nInstall those four pieces using your distribution's package manager and\nrerun the script with `--check-only` to verify. The check is universally\nportable — it only queries the binaries \u002F package databases — so once the\nruntime is in place the workflow contract is satisfied regardless of the\nunderlying distro.\n\n**Unsupported Ubuntu\u002FDebian derivative**: When `ID` is e.g. `pop`, `mint`,\n`zorin`, `raspbian`, or another debian-family derivative, the script maps\nthe host onto the upstream Ubuntu\u002FDebian CUDA repo via `UBUNTU_CODENAME` \u002F\n`VERSION_CODENAME` (`focal`\u002F`jammy`\u002F`noble` → Ubuntu 20.04\u002F22.04\u002F24.04;\n`bullseye`\u002F`bookworm`\u002F`trixie` → Debian 11\u002F12\u002F12). If the host's codename\ndoesn't match a known upstream release, `--install` exits with the same\nmanual-install guidance described above.\n\n**Docker not installed**: `--check-only` reports `MISSING: Docker is not\ninstalled` and prints the exact rerun command appropriate to the detected\ndistro family. The default `--install` path installs Docker (`docker.io` \u002F\n`moby-engine` \u002F `docker-ce` \u002F `docker` depending on family), enables\u002Fstarts\nthe daemon, configures the NVIDIA runtime, and adds the invoking user to\nthe `docker` group. If you prefer to manage Docker yourself, install it\nbefore rerunning the script or pass `--skip-docker-install`.\n\n**Docker installed but `docker run` still needs sudo**: The script adds the\ninvoking user to the `docker` group, but Linux only refreshes group\nmembership on a new login session. Log out and back in, or run\n`newgrp docker` in each new shell, until the new membership is active.\n\n**Docker runtime still missing**: Restart Docker, then rerun\n`nvidia-ctk runtime configure --runtime=docker`.\n\n**Driver branch detected != 580**: The driver-branch pin is exact on\ndebian-family (`nvidia-open-580`). On rhel-\u002Fsuse-family the script\ninstalls the latest open driver shipped in NVIDIA's CUDA 13.0 repo for\nthe detected distro, which is always ≥ 580. If your host needs a stricter\npin, set `$NVIDIA_DRIVER_PACKAGE_RHEL` \u002F `$NVIDIA_DRIVER_KMOD_RHEL` \u002F\n`$NVIDIA_DRIVER_PACKAGE_SUSE` to the exact package names you want before\nrunning `--install`.\n\n**Driver installed but `nvidia-smi` fails**: Load the module with\n`sudo modprobe nvidia` or reboot. Secure Boot may require MOK enrollment on\nsystems where it is enabled.\n\n**Kubernetes still has no GPU capacity**: Confirm the driver works on each GPU\nnode with `nvidia-smi`, then check the GPU Operator\u002Fdevice plugin pods and node\nlabels.\n",{"name":5,"tags":23,"license":14,"version":29,"category":30,"metadata":31,"description":34,"allowed-tools":35,"compatibility":36},[24,25,26,27,28],"setup","nvidia","cuda","docker","kubernetes","1.0.0 Host setup for TAO GPU backends. Checks and, after user approval, installs NVIDIA driver branch 580, CUDA Toolkit 13.0, and NVIDIA Container Toolkit 1.19.0 for Docker\u002Flocal-Docker and Kubernetes GPU worker hosts. The `--check-only` path works on any Linux distribution; `--install` automates debian-family (Ubuntu\u002FDebian\u002FPop!_OS\u002FMint\u002FZorin\u002FRaspbian), rhel-family (Fedora\u002FRHEL\u002FRocky\u002FAlmaLinux), and suse-family (openSUSE\u002FSLES) hosts, and prints actionable manual-install steps for everything else.","development-code",{"author":32,"version":33},"NVIDIA Corporation","0.1.0","","Read Bash","Runs `--check-only` on any Linux distribution. `--install` automates Ubuntu 22.04\u002F24.04 + Debian 12 (apt), Fedora + RHEL\u002FRocky\u002FAlmaLinux 9\u002F10 (dnf), and openSUSE Leap \u002F SLES 15 (zypper). Requires sudo\u002Froot, internet access to NVIDIA package repositories (and download.docker.com on rhel-family), and an x86_64 or aarch64 (sbsa) host. Other distributions (Arch, Alpine, Gentoo, NixOS, …) get a clear error that names the version targets and the NVIDIA install-guide URL.","main","d591030ef9fb1bbcebed3454aa5c2dcb8d608f69","skills-hub\u002Fai-ml\u002Fnvidia-skills\u002Fskills\u002Ftao-setup-nvidia-gpu-host","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive","https:\u002F\u002Fai-agent.yyc3.vip\u002F","https:\u002F\u002Fgithub.com\u002FYYC-Cube","万象归元于云枢 | 深栈智启新纪元","2026-11-13 22:26:14",{"source":7,"skillName":5,"commands":46,"prompt":56,"zipAvailable":57},[47,50,53],{"tool":48,"command":49},"npx","npx skills add YYC-Cube\u002FYYC3-AI-Agent-Archive --skill tao-setup-nvidia-gpu-host",{"tool":51,"command":52},"bunx","bunx skills add YYC-Cube\u002FYYC3-AI-Agent-Archive --skill tao-setup-nvidia-gpu-host",{"tool":54,"command":55},"pnpm","pnpm dlx skills add YYC-Cube\u002FYYC3-AI-Agent-Archive --skill tao-setup-nvidia-gpu-host","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Fskills-hub\u002Fai-ml\u002Fnvidia-skills\u002Fskills\u002Ftao-setup-nvidia-gpu-host\u002FSKILL.md",true,[59,65,69,73,79,83],{"path":60,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":63,"sourceHtmlUrl":64},"BENCHMARK.md","REFERENCE","参考文档",4310,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Fskills-hub\u002Fai-ml\u002Fnvidia-skills\u002Fskills\u002Ftao-setup-nvidia-gpu-host\u002FBENCHMARK.md",{"path":66,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":67,"sourceHtmlUrl":68},"evals\u002Fevals.json",834,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Fskills-hub\u002Fai-ml\u002Fnvidia-skills\u002Fskills\u002Ftao-setup-nvidia-gpu-host\u002Fevals\u002Fevals.json",{"path":70,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Fskill_info.yaml",818,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Fskills-hub\u002Fai-ml\u002Fnvidia-skills\u002Fskills\u002Ftao-setup-nvidia-gpu-host\u002Freferences\u002Fskill_info.yaml",{"path":74,"resourceType":75,"resourceTypeDesc":76,"sizeBytes":77,"sourceHtmlUrl":78},"scripts\u002Fsetup-nvidia-gpu-host.sh","SCRIPT","脚本",26306,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Fskills-hub\u002Fai-ml\u002Fnvidia-skills\u002Fskills\u002Ftao-setup-nvidia-gpu-host\u002Fscripts\u002Fsetup-nvidia-gpu-host.sh",{"path":80,"resourceType":61,"resourceTypeDesc":62,"sizeBytes":81,"sourceHtmlUrl":82},"skill-card.md",4034,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Fskills-hub\u002Fai-ml\u002Fnvidia-skills\u002Fskills\u002Ftao-setup-nvidia-gpu-host\u002Fskill-card.md",{"path":84,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":87,"sourceHtmlUrl":88},"skill.oms.sig","OTHER","其他",5081,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Fskills-hub\u002Fai-ml\u002Fnvidia-skills\u002Fskills\u002Ftao-setup-nvidia-gpu-host\u002Fskill.oms.sig",[90,98,105,112,119,127],{"skillNo":91,"name":92,"nameZh":6,"description":93,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":94,"licenseFileUrl":6,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":18},"01M7BTB92RE8MT5YE706QBNDTQ","yingmi-skill","当用户需要查询基金、策略、公告、财经资讯，做资产配置、组合诊断、风险回测、现金流分析，或生成图表、PDF 时，优先使用本 Skill 获取真实数据与可执行能力。","MIT","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Ftools-hub\u002Fcode-ide\u002Fskills-marketplace\u002Fskills\u002Fyingmi-skill\u002FSKILL.md",[],[],{"skillNo":99,"name":100,"nameZh":6,"description":101,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":94,"licenseFileUrl":6,"sourceHtmlUrl":102,"categories":103,"keywords":104,"publishedAt":18},"01M7BT7ZYR091HPJYEFNWVTEJJ","web-access","所有联网操作必须通过此 skill 处理，包括：搜索、网页抓取、登录后操作、网络交互等。","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Ftools-hub\u002Fcode-ide\u002Fskills-marketplace\u002Fskills\u002Fweb-access\u002FSKILL.md",[],[],{"skillNo":106,"name":107,"nameZh":6,"description":108,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":94,"licenseFileUrl":6,"sourceHtmlUrl":109,"categories":110,"keywords":111,"publishedAt":18},"01M7BT4NKS0DNP8A53M1BGGZR2","tencentmap-map-assistant","腾讯位置服务·地图助手 Skill，一句自然语言调用腾讯地图全套能力，无需开发者账号、开箱即用。提供 AI 旅游攻略、地点搜索（含评分\u002F人均\u002F营业时间）、关键词提示、路线规划（驾车\u002F步行\u002F公交\u002F骑行）、地址解析与逆解析、行政区划、IP 定位、距离计算、天气查询，并可将行程或多 POI 渲染成网页地图。涉及找地点、规划路线、旅游行程、查天气、坐标转换等出行场景时使用。","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Ftools-hub\u002Fcode-ide\u002Fskills-marketplace\u002Fskills\u002Ftencentmap-map-assistant\u002FSKILL.md",[],[],{"skillNo":113,"name":114,"nameZh":6,"description":115,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":94,"licenseFileUrl":6,"sourceHtmlUrl":116,"categories":117,"keywords":118,"publishedAt":18},"01M7BSWKV1T5XYYRGV4W4JP88C","skill-scanner","Scan any agent skill for security risks before you install or use it. Powered by Tencent Zhuque Lab A.I.G (AI-Infra-Guard). 100% local static analysis — no file contents or credentials leave your device. Compatible with CodeBuddy, Cursor, Windsurf, Claude Code, OpenClaw and more. Triggers on: 这个 skill 安全吗, skill 安全扫描, 检查 skill 安全, audit skill, scan skill, check skill safety, analyze skill, inspect skill, verify skill, skill security, skill supply chain. Do NOT trigger for general agent usage, full system health checks, project debugging, or normal development.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Ftools-hub\u002Fcode-ide\u002Fskills-marketplace\u002Fskills\u002Fskill-scanner\u002FSKILL.md",[],[],{"skillNo":120,"name":121,"nameZh":6,"description":122,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":123,"sourceHtmlUrl":124,"categories":125,"keywords":126,"publishedAt":18},"01M7BSWGTCFW7F123S4BM4XY6W","skill-creator","Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends agent capabilities with specialized knowledge, workflows, or tool integrations.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Ftools-hub\u002Fcode-ide\u002Fskills-marketplace\u002Fskills\u002Fskill-creator\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Ftools-hub\u002Fcode-ide\u002Fskills-marketplace\u002Fskills\u002Fskill-creator\u002FSKILL.md",[],[],{"skillNo":128,"name":129,"nameZh":6,"description":130,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":94,"licenseFileUrl":6,"sourceHtmlUrl":131,"categories":132,"keywords":133,"publishedAt":18},"01M7BSVV4DH8JK2ZMQSCW25VCD","shader-dev","Comprehensive GLSL shader techniques for creating stunning visual effects — ray marching, SDF modeling, fluid simulation, particle systems, procedural generation, lighting, post-processing, and more.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-AI-Agent-Archive\u002Fblob\u002Fmain\u002Ftools-hub\u002Fcode-ide\u002Fskills-marketplace\u002Fskills\u002Fshader-dev\u002FSKILL.md",[],[]]