[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-jetson-package":3,"mdc-5j62d6-key":34,"related-org-nvidia-jetson-package":1178,"related-repo-nvidia-jetson-package":1335},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":23,"repoUrl":24,"updatedAt":25,"license":26,"forks":27,"topics":28,"repo":29,"sourceUrl":32,"mdContent":33},"jetson-package","configure Jetson container and runtime environments","Pick Jetson-compatible containers, vLLM runtime images, and Jetson AI Lab PyPI indexes; maps Orin SM 8.7 vs Thor SM 11.0 and JetPack-specific package choices.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,16,19,20],{"name":13,"slug":14,"type":15},"Containers","containers","tag",{"name":17,"slug":18,"type":15},"Deployment","deployment",{"name":9,"slug":8,"type":15},{"name":21,"slug":22,"type":15},"AI Infrastructure","ai-infrastructure",2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:26:08.793975","Apache-2.0",281,[],{"repoUrl":24,"stars":23,"forks":27,"topics":30,"description":31},[],"AI agent skills published by NVIDIA","https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Fjetson-package","---\nname: jetson-package\ndescription: Pick Jetson-compatible containers, vLLM runtime images, and Jetson AI Lab PyPI indexes; maps Orin SM 8.7 vs Thor SM 11.0 and JetPack-specific package choices.\nversion: 0.0.1\nlicense: \"Apache-2.0\"\nmetadata:\n  author: \"Jetson Team\"\n  tags: [jetson, package, containers]\n  languages: [bash]\n  data-classification: public\n---\n\n# Jetson Package & Environment\n\nAgents often suggest `docker pull` images or `pip install` wheels that claim **aarch64** support but were never built for Jetson’s GPU **streaming multiprocessor (SM)** targets. On Jetson, **default to NVIDIA-curated artifacts** unless the user explicitly opts out.\n\n## Purpose\n\nChoose Jetson-compatible containers and Python package indexes before installing GPU-native ML stacks. This skill prevents agents from recommending generic ARM wheels or stale container tags that do not include the right CUDA, JetPack, or SM target for the device.\n\n## When to use\n\n- \"Which Docker image \u002F container should I use on this Jetson?\"\n- \"Where do I get PyTorch \u002F vLLM \u002F CUDA wheels for Jetson?\"\n- \"`pip install` failed\" or \"wrong CUDA \u002F SM\" after installing a generic ARM wheel.\n- Before `docker run` or `pip install` for ML stacks on Orin or Thor.\n- User or agent looks for `l4t-cuda` containers on NGC — redirect to `nvcr.io\u002Fnvidia\u002Fcuda` (multi-arch).\n- \"Which PyTorch container should I use on Jetson?\" — answer depends on Thor vs Orin and JetPack version.\n\n## Canonical sources (use these first)\n\n1. **Prebuilt containers (GHCR)** — [NVIDIA-AI-IOT packages](https:\u002F\u002Fgithub.com\u002Forgs\u002FNVIDIA-AI-IOT\u002Fpackages): `llama_cpp`, `ollama`, `live-vlm-webui`, older-Orin `vllm`, and related images built for Jetson JetPack stacks. Prefer these over random `arm64` images on Docker Hub. For vLLM, use upstream `vllm\u002Fvllm-openai` on Thor and Orin JetPack 7.2 \u002F L4T r39+.\n2. **NGC CUDA \u002F PyTorch containers** — Tag selection depends on Jetson generation. Do not treat example PyTorch tag shapes as pinned recommendations; look up the current tag in the [NGC PyTorch catalog](https:\u002F\u002Fcatalog.ngc.nvidia.com\u002Forgs\u002Fnvidia\u002Fcontainers\u002Fpytorch) before giving a command.\n\n   | Jetson | CUDA base | PyTorch |\n   |--------|-----------|---------|\n   | **Thor** | `nvcr.io\u002Fnvidia\u002Fcuda:\u003Cver>-devel-ubuntu\u003Cver>` (multi-arch, arm64 included) | `nvcr.io\u002Fnvidia\u002Fpytorch:\u003Ccurrent-tag>-py3` (main multi-arch tag; verify current NGC tag) |\n   | **Orin + r36 \u002F JetPack 6** | same multi-arch CUDA base | `nvcr.io\u002Fnvidia\u002Fpytorch:\u003Ccurrent-tag>-py3-igpu` — verify the current NGC tag and use the `-igpu` suffix for Orin iGPU (SM 8.7) when NGC publishes it |\n   | **Orin + r39+ (future)** | same | likely main multi-arch tag once Orin becomes SBSA; verify when r39 ships |\n\n**`l4t-cuda` is the legacy Orin-era CUDA container line.** If a user cannot find `l4t-cuda` on NGC, redirect them to the current multi-arch `nvcr.io\u002Fnvidia\u002Fcuda` image instead of third-party images.\n3. **Python package indexes (devpi)** — [Jetson AI Lab PyPI](https:\u002F\u002Fpypi.jetson-ai-lab.io\u002F): browse the tree (for example `jp6\u002Fcu126`, `jp6\u002Fcu128`) and pick the index that matches your **JetPack \u002F CUDA userland**. Prefer these over PyPI-only wheels for GPU-native stacks.\n\n## GPU architecture reminder (why generic ARM fails)\n\n| Jetson family | CUDA compute capability | Build target | Note |\n|---------------|-------------------------|--------------|------|\n| Orin (AGX \u002F NX \u002F Nano) | **8.7** | `sm_87` | Many desktop `aarch64` wheels omit Jetson Orin kernels. |\n| Thor (T5000 \u002F T4000) | **11.0** | `sm_110` | Requires CUDA \u002F wheels \u002F containers that include Blackwell Jetson support. |\n\nA wheel or container may install on **ARM64 Linux** and still be **unusable or slow** if CUDA kernels were not compiled for your Jetson’s SM.\n\nUse CUDA build target names when discussing wheel compatibility: `sm_87` for Jetson Orin and `sm_110` for Jetson Thor. Do not infer the generation from a prompt or a hostname — run `scripts\u002Fartifact_hints.sh` and use its detected `generation`, `variant`, `l4t`, and `cuda_sm_hint` fields before recommending wheels or container tags.\n\n## GPU Python wheels on Jetson\n\nDefault PyPI wheels for GPU-native packages are usually not the right answer on Jetson, even when they claim `aarch64` support. For `onnxruntime-gpu`, PyTorch, vLLM, and similar packages, use the Jetson AI Lab package index as the canonical source and choose the subtree that matches the device's JetPack \u002F CUDA userland.\n\nFor `onnxruntime-gpu`, lead with Jetson AI Lab rather than plain PyPI:\n\n```bash\npip install --extra-index-url https:\u002F\u002Fpypi.jetson-ai-lab.io\u002Fjp6\u002Fcu126\u002F+simple\u002F onnxruntime-gpu\n```\n\nAdjust the `jp6\u002Fcu126` portion to match the detected JetPack \u002F CUDA line. Do not present `pip install onnxruntime-gpu` from default PyPI as an equivalent Jetson GPU option.\n\n## Do not fabricate device facts\n\nDo not invent SKU names, RAM sizes, JetPack versions, CUDA versions, or GPU SM targets. Quote only what `scripts\u002Fartifact_hints.sh` or the user's supplied environment reports. If a field is unavailable, omit it or say it is unknown.\n\n## Prerequisites\n\n- Run package-detection scripts on a Jetson target, not on the host workstation.\n- Network access is needed to inspect GHCR, NGC, or Jetson AI Lab package indexes.\n- Source device facts from `scripts\u002Fartifact_hints.sh`, `jetson-diagnostic`, or user-provided environment output before recommending tags or wheels.\n\n## Available Scripts\n\n| Script | Purpose | Arguments |\n|--------|---------|-----------|\n| `scripts\u002Fartifact_hints.sh` | Emits detected Jetson SKU\u002Fgeneration, CUDA SM hint, canonical package URLs, and a preferred vLLM image hint. | `--human` for a readable summary; no argument for JSON. |\n\nIf your agent runtime supports `run_script`, use it to run `scripts\u002Fartifact_hints.sh` and read the JSON output. Otherwise run the script with `bash` from the repository root.\n\n## Instructions\n\n1. Run `scripts\u002Fartifact_hints.sh` (JSON on stdout). It sources `skills\u002Fjetson-diagnostic\u002Fscripts\u002Fdetect_jetson.sh` and returns `sku`, `generation`, `product_line`, `variant`, `l4t`, a preferred **vLLM** image, `cuda_sm_hint`, and canonical URLs.\n2. For **pip**, open the devpi root in a browser, pick the **jp6** subtree that matches your CUDA line, and set `--extra-index-url` \u002F `PIP_EXTRA_INDEX_URL` — see `references\u002Fpypi-jetson-ai-lab.md`.\n3. For **containers**, see `references\u002Fghcr-images.md` and `jetson-llm-serve` for vLLM.\n\n## Limitations\n\n- This skill points to package catalogs and emits compatibility hints; it does not verify that a specific model checkpoint fits in memory.\n- NGC and GHCR tags change. Treat placeholder tag shapes such as `\u003Ccurrent-tag>-py3` as lookup instructions, not literal tags.\n- If `generation` or `cuda_sm_hint` is unknown, do not guess a container tag.\n\n## Hand off to\n\n- `jetson-llm-serve` — run upstream\u002Fnative vLLM 0.20+ on Thor and Orin JetPack 7.2 \u002F L4T r39+, or `vllm:latest-jetson-orin` on older Orin.\n- `jetson-llm-benchmark` — measure after the stack is installed.\n- `jetson-diagnostic` — if installs succeed but runtime fails, snapshot first.\n\n## Safety\n\nRead-only: points to catalogs and emits hints; does not install or pull.\n\n## Sources\n\n[NVIDIA-AI-IOT GitHub Packages](https:\u002F\u002Fgithub.com\u002Forgs\u002FNVIDIA-AI-IOT\u002Fpackages), [pypi.jetson-ai-lab.io](https:\u002F\u002Fpypi.jetson-ai-lab.io\u002F).\n",{"data":35,"body":45},{"name":4,"description":6,"version":36,"license":26,"metadata":37},"0.0.1",{"author":38,"tags":39,"languages":42,"data-classification":44},"Jetson Team",[40,41,14],"jetson","package",[43],"bash","public",{"type":46,"children":47},"root",[48,57,102,109,114,120,193,199,412,478,484,583,602,659,665,685,697,740,760,766,778,784,816,822,875,902,908,1045,1051,1090,1096,1138,1144,1149,1155,1172],{"type":49,"tag":50,"props":51,"children":53},"element","h1",{"id":52},"jetson-package-environment",[54],{"type":55,"value":56},"text","Jetson Package & Environment",{"type":49,"tag":58,"props":59,"children":60},"p",{},[61,63,70,72,78,80,86,88,93,95,100],{"type":55,"value":62},"Agents often suggest ",{"type":49,"tag":64,"props":65,"children":67},"code",{"className":66},[],[68],{"type":55,"value":69},"docker pull",{"type":55,"value":71}," images or ",{"type":49,"tag":64,"props":73,"children":75},{"className":74},[],[76],{"type":55,"value":77},"pip install",{"type":55,"value":79}," wheels that claim ",{"type":49,"tag":81,"props":82,"children":83},"strong",{},[84],{"type":55,"value":85},"aarch64",{"type":55,"value":87}," support but were never built for Jetson’s GPU ",{"type":49,"tag":81,"props":89,"children":90},{},[91],{"type":55,"value":92},"streaming multiprocessor (SM)",{"type":55,"value":94}," targets. On Jetson, ",{"type":49,"tag":81,"props":96,"children":97},{},[98],{"type":55,"value":99},"default to NVIDIA-curated artifacts",{"type":55,"value":101}," unless the user explicitly opts out.",{"type":49,"tag":103,"props":104,"children":106},"h2",{"id":105},"purpose",[107],{"type":55,"value":108},"Purpose",{"type":49,"tag":58,"props":110,"children":111},{},[112],{"type":55,"value":113},"Choose Jetson-compatible containers and Python package indexes before installing GPU-native ML stacks. This skill prevents agents from recommending generic ARM wheels or stale container tags that do not include the right CUDA, JetPack, or SM target for the device.",{"type":49,"tag":103,"props":115,"children":117},{"id":116},"when-to-use",[118],{"type":55,"value":119},"When to use",{"type":49,"tag":121,"props":122,"children":123},"ul",{},[124,130,135,147,167,188],{"type":49,"tag":125,"props":126,"children":127},"li",{},[128],{"type":55,"value":129},"\"Which Docker image \u002F container should I use on this Jetson?\"",{"type":49,"tag":125,"props":131,"children":132},{},[133],{"type":55,"value":134},"\"Where do I get PyTorch \u002F vLLM \u002F CUDA wheels for Jetson?\"",{"type":49,"tag":125,"props":136,"children":137},{},[138,140,145],{"type":55,"value":139},"\"",{"type":49,"tag":64,"props":141,"children":143},{"className":142},[],[144],{"type":55,"value":77},{"type":55,"value":146}," failed\" or \"wrong CUDA \u002F SM\" after installing a generic ARM wheel.",{"type":49,"tag":125,"props":148,"children":149},{},[150,152,158,160,165],{"type":55,"value":151},"Before ",{"type":49,"tag":64,"props":153,"children":155},{"className":154},[],[156],{"type":55,"value":157},"docker run",{"type":55,"value":159}," or ",{"type":49,"tag":64,"props":161,"children":163},{"className":162},[],[164],{"type":55,"value":77},{"type":55,"value":166}," for ML stacks on Orin or Thor.",{"type":49,"tag":125,"props":168,"children":169},{},[170,172,178,180,186],{"type":55,"value":171},"User or agent looks for ",{"type":49,"tag":64,"props":173,"children":175},{"className":174},[],[176],{"type":55,"value":177},"l4t-cuda",{"type":55,"value":179}," containers on NGC — redirect to ",{"type":49,"tag":64,"props":181,"children":183},{"className":182},[],[184],{"type":55,"value":185},"nvcr.io\u002Fnvidia\u002Fcuda",{"type":55,"value":187}," (multi-arch).",{"type":49,"tag":125,"props":189,"children":190},{},[191],{"type":55,"value":192},"\"Which PyTorch container should I use on Jetson?\" — answer depends on Thor vs Orin and JetPack version.",{"type":49,"tag":103,"props":194,"children":196},{"id":195},"canonical-sources-use-these-first",[197],{"type":55,"value":198},"Canonical sources (use these first)",{"type":49,"tag":200,"props":201,"children":202},"ol",{},[203,271],{"type":49,"tag":125,"props":204,"children":205},{},[206,211,213,222,224,230,232,238,239,245,247,253,255,261,263,269],{"type":49,"tag":81,"props":207,"children":208},{},[209],{"type":55,"value":210},"Prebuilt containers (GHCR)",{"type":55,"value":212}," — ",{"type":49,"tag":214,"props":215,"children":219},"a",{"href":216,"rel":217},"https:\u002F\u002Fgithub.com\u002Forgs\u002FNVIDIA-AI-IOT\u002Fpackages",[218],"nofollow",[220],{"type":55,"value":221},"NVIDIA-AI-IOT packages",{"type":55,"value":223},": ",{"type":49,"tag":64,"props":225,"children":227},{"className":226},[],[228],{"type":55,"value":229},"llama_cpp",{"type":55,"value":231},", ",{"type":49,"tag":64,"props":233,"children":235},{"className":234},[],[236],{"type":55,"value":237},"ollama",{"type":55,"value":231},{"type":49,"tag":64,"props":240,"children":242},{"className":241},[],[243],{"type":55,"value":244},"live-vlm-webui",{"type":55,"value":246},", older-Orin ",{"type":49,"tag":64,"props":248,"children":250},{"className":249},[],[251],{"type":55,"value":252},"vllm",{"type":55,"value":254},", and related images built for Jetson JetPack stacks. Prefer these over random ",{"type":49,"tag":64,"props":256,"children":258},{"className":257},[],[259],{"type":55,"value":260},"arm64",{"type":55,"value":262}," images on Docker Hub. For vLLM, use upstream ",{"type":49,"tag":64,"props":264,"children":266},{"className":265},[],[267],{"type":55,"value":268},"vllm\u002Fvllm-openai",{"type":55,"value":270}," on Thor and Orin JetPack 7.2 \u002F L4T r39+.",{"type":49,"tag":125,"props":272,"children":273},{},[274,279,281,288,290],{"type":49,"tag":81,"props":275,"children":276},{},[277],{"type":55,"value":278},"NGC CUDA \u002F PyTorch containers",{"type":55,"value":280}," — Tag selection depends on Jetson generation. Do not treat example PyTorch tag shapes as pinned recommendations; look up the current tag in the ",{"type":49,"tag":214,"props":282,"children":285},{"href":283,"rel":284},"https:\u002F\u002Fcatalog.ngc.nvidia.com\u002Forgs\u002Fnvidia\u002Fcontainers\u002Fpytorch",[218],[286],{"type":55,"value":287},"NGC PyTorch catalog",{"type":55,"value":289}," before giving a command.",{"type":49,"tag":291,"props":292,"children":293},"table",{},[294,318],{"type":49,"tag":295,"props":296,"children":297},"thead",{},[298],{"type":49,"tag":299,"props":300,"children":301},"tr",{},[302,308,313],{"type":49,"tag":303,"props":304,"children":305},"th",{},[306],{"type":55,"value":307},"Jetson",{"type":49,"tag":303,"props":309,"children":310},{},[311],{"type":55,"value":312},"CUDA base",{"type":49,"tag":303,"props":314,"children":315},{},[316],{"type":55,"value":317},"PyTorch",{"type":49,"tag":319,"props":320,"children":321},"tbody",{},[322,356,391],{"type":49,"tag":299,"props":323,"children":324},{},[325,334,345],{"type":49,"tag":326,"props":327,"children":328},"td",{},[329],{"type":49,"tag":81,"props":330,"children":331},{},[332],{"type":55,"value":333},"Thor",{"type":49,"tag":326,"props":335,"children":336},{},[337,343],{"type":49,"tag":64,"props":338,"children":340},{"className":339},[],[341],{"type":55,"value":342},"nvcr.io\u002Fnvidia\u002Fcuda:\u003Cver>-devel-ubuntu\u003Cver>",{"type":55,"value":344}," (multi-arch, arm64 included)",{"type":49,"tag":326,"props":346,"children":347},{},[348,354],{"type":49,"tag":64,"props":349,"children":351},{"className":350},[],[352],{"type":55,"value":353},"nvcr.io\u002Fnvidia\u002Fpytorch:\u003Ccurrent-tag>-py3",{"type":55,"value":355}," (main multi-arch tag; verify current NGC tag)",{"type":49,"tag":299,"props":357,"children":358},{},[359,367,372],{"type":49,"tag":326,"props":360,"children":361},{},[362],{"type":49,"tag":81,"props":363,"children":364},{},[365],{"type":55,"value":366},"Orin + r36 \u002F JetPack 6",{"type":49,"tag":326,"props":368,"children":369},{},[370],{"type":55,"value":371},"same multi-arch CUDA base",{"type":49,"tag":326,"props":373,"children":374},{},[375,381,383,389],{"type":49,"tag":64,"props":376,"children":378},{"className":377},[],[379],{"type":55,"value":380},"nvcr.io\u002Fnvidia\u002Fpytorch:\u003Ccurrent-tag>-py3-igpu",{"type":55,"value":382}," — verify the current NGC tag and use the ",{"type":49,"tag":64,"props":384,"children":386},{"className":385},[],[387],{"type":55,"value":388},"-igpu",{"type":55,"value":390}," suffix for Orin iGPU (SM 8.7) when NGC publishes it",{"type":49,"tag":299,"props":392,"children":393},{},[394,402,407],{"type":49,"tag":326,"props":395,"children":396},{},[397],{"type":49,"tag":81,"props":398,"children":399},{},[400],{"type":55,"value":401},"Orin + r39+ (future)",{"type":49,"tag":326,"props":403,"children":404},{},[405],{"type":55,"value":406},"same",{"type":49,"tag":326,"props":408,"children":409},{},[410],{"type":55,"value":411},"likely main multi-arch tag once Orin becomes SBSA; verify when r39 ships",{"type":49,"tag":58,"props":413,"children":414},{},[415,425,427,432,434,439,441,446,447,454,456,462,463,469,471,476],{"type":49,"tag":81,"props":416,"children":417},{},[418,423],{"type":49,"tag":64,"props":419,"children":421},{"className":420},[],[422],{"type":55,"value":177},{"type":55,"value":424}," is the legacy Orin-era CUDA container line.",{"type":55,"value":426}," If a user cannot find ",{"type":49,"tag":64,"props":428,"children":430},{"className":429},[],[431],{"type":55,"value":177},{"type":55,"value":433}," on NGC, redirect them to the current multi-arch ",{"type":49,"tag":64,"props":435,"children":437},{"className":436},[],[438],{"type":55,"value":185},{"type":55,"value":440}," image instead of third-party images.\n3. ",{"type":49,"tag":81,"props":442,"children":443},{},[444],{"type":55,"value":445},"Python package indexes (devpi)",{"type":55,"value":212},{"type":49,"tag":214,"props":448,"children":451},{"href":449,"rel":450},"https:\u002F\u002Fpypi.jetson-ai-lab.io\u002F",[218],[452],{"type":55,"value":453},"Jetson AI Lab PyPI",{"type":55,"value":455},": browse the tree (for example ",{"type":49,"tag":64,"props":457,"children":459},{"className":458},[],[460],{"type":55,"value":461},"jp6\u002Fcu126",{"type":55,"value":231},{"type":49,"tag":64,"props":464,"children":466},{"className":465},[],[467],{"type":55,"value":468},"jp6\u002Fcu128",{"type":55,"value":470},") and pick the index that matches your ",{"type":49,"tag":81,"props":472,"children":473},{},[474],{"type":55,"value":475},"JetPack \u002F CUDA userland",{"type":55,"value":477},". Prefer these over PyPI-only wheels for GPU-native stacks.",{"type":49,"tag":103,"props":479,"children":481},{"id":480},"gpu-architecture-reminder-why-generic-arm-fails",[482],{"type":55,"value":483},"GPU architecture reminder (why generic ARM fails)",{"type":49,"tag":291,"props":485,"children":486},{},[487,513],{"type":49,"tag":295,"props":488,"children":489},{},[490],{"type":49,"tag":299,"props":491,"children":492},{},[493,498,503,508],{"type":49,"tag":303,"props":494,"children":495},{},[496],{"type":55,"value":497},"Jetson family",{"type":49,"tag":303,"props":499,"children":500},{},[501],{"type":55,"value":502},"CUDA compute capability",{"type":49,"tag":303,"props":504,"children":505},{},[506],{"type":55,"value":507},"Build target",{"type":49,"tag":303,"props":509,"children":510},{},[511],{"type":55,"value":512},"Note",{"type":49,"tag":319,"props":514,"children":515},{},[516,553],{"type":49,"tag":299,"props":517,"children":518},{},[519,524,532,541],{"type":49,"tag":326,"props":520,"children":521},{},[522],{"type":55,"value":523},"Orin (AGX \u002F NX \u002F Nano)",{"type":49,"tag":326,"props":525,"children":526},{},[527],{"type":49,"tag":81,"props":528,"children":529},{},[530],{"type":55,"value":531},"8.7",{"type":49,"tag":326,"props":533,"children":534},{},[535],{"type":49,"tag":64,"props":536,"children":538},{"className":537},[],[539],{"type":55,"value":540},"sm_87",{"type":49,"tag":326,"props":542,"children":543},{},[544,546,551],{"type":55,"value":545},"Many desktop ",{"type":49,"tag":64,"props":547,"children":549},{"className":548},[],[550],{"type":55,"value":85},{"type":55,"value":552}," wheels omit Jetson Orin kernels.",{"type":49,"tag":299,"props":554,"children":555},{},[556,561,569,578],{"type":49,"tag":326,"props":557,"children":558},{},[559],{"type":55,"value":560},"Thor (T5000 \u002F T4000)",{"type":49,"tag":326,"props":562,"children":563},{},[564],{"type":49,"tag":81,"props":565,"children":566},{},[567],{"type":55,"value":568},"11.0",{"type":49,"tag":326,"props":570,"children":571},{},[572],{"type":49,"tag":64,"props":573,"children":575},{"className":574},[],[576],{"type":55,"value":577},"sm_110",{"type":49,"tag":326,"props":579,"children":580},{},[581],{"type":55,"value":582},"Requires CUDA \u002F wheels \u002F containers that include Blackwell Jetson support.",{"type":49,"tag":58,"props":584,"children":585},{},[586,588,593,595,600],{"type":55,"value":587},"A wheel or container may install on ",{"type":49,"tag":81,"props":589,"children":590},{},[591],{"type":55,"value":592},"ARM64 Linux",{"type":55,"value":594}," and still be ",{"type":49,"tag":81,"props":596,"children":597},{},[598],{"type":55,"value":599},"unusable or slow",{"type":55,"value":601}," if CUDA kernels were not compiled for your Jetson’s SM.",{"type":49,"tag":58,"props":603,"children":604},{},[605,607,612,614,619,621,627,629,635,636,642,643,649,651,657],{"type":55,"value":606},"Use CUDA build target names when discussing wheel compatibility: ",{"type":49,"tag":64,"props":608,"children":610},{"className":609},[],[611],{"type":55,"value":540},{"type":55,"value":613}," for Jetson Orin and ",{"type":49,"tag":64,"props":615,"children":617},{"className":616},[],[618],{"type":55,"value":577},{"type":55,"value":620}," for Jetson Thor. Do not infer the generation from a prompt or a hostname — run ",{"type":49,"tag":64,"props":622,"children":624},{"className":623},[],[625],{"type":55,"value":626},"scripts\u002Fartifact_hints.sh",{"type":55,"value":628}," and use its detected ",{"type":49,"tag":64,"props":630,"children":632},{"className":631},[],[633],{"type":55,"value":634},"generation",{"type":55,"value":231},{"type":49,"tag":64,"props":637,"children":639},{"className":638},[],[640],{"type":55,"value":641},"variant",{"type":55,"value":231},{"type":49,"tag":64,"props":644,"children":646},{"className":645},[],[647],{"type":55,"value":648},"l4t",{"type":55,"value":650},", and ",{"type":49,"tag":64,"props":652,"children":654},{"className":653},[],[655],{"type":55,"value":656},"cuda_sm_hint",{"type":55,"value":658}," fields before recommending wheels or container tags.",{"type":49,"tag":103,"props":660,"children":662},{"id":661},"gpu-python-wheels-on-jetson",[663],{"type":55,"value":664},"GPU Python wheels on Jetson",{"type":49,"tag":58,"props":666,"children":667},{},[668,670,675,677,683],{"type":55,"value":669},"Default PyPI wheels for GPU-native packages are usually not the right answer on Jetson, even when they claim ",{"type":49,"tag":64,"props":671,"children":673},{"className":672},[],[674],{"type":55,"value":85},{"type":55,"value":676}," support. For ",{"type":49,"tag":64,"props":678,"children":680},{"className":679},[],[681],{"type":55,"value":682},"onnxruntime-gpu",{"type":55,"value":684},", PyTorch, vLLM, and similar packages, use the Jetson AI Lab package index as the canonical source and choose the subtree that matches the device's JetPack \u002F CUDA userland.",{"type":49,"tag":58,"props":686,"children":687},{},[688,690,695],{"type":55,"value":689},"For ",{"type":49,"tag":64,"props":691,"children":693},{"className":692},[],[694],{"type":55,"value":682},{"type":55,"value":696},", lead with Jetson AI Lab rather than plain PyPI:",{"type":49,"tag":698,"props":699,"children":703},"pre",{"className":700,"code":701,"language":43,"meta":702,"style":702},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","pip install --extra-index-url https:\u002F\u002Fpypi.jetson-ai-lab.io\u002Fjp6\u002Fcu126\u002F+simple\u002F onnxruntime-gpu\n","",[704],{"type":49,"tag":64,"props":705,"children":706},{"__ignoreMap":702},[707],{"type":49,"tag":708,"props":709,"children":712},"span",{"class":710,"line":711},"line",1,[713,719,725,730,735],{"type":49,"tag":708,"props":714,"children":716},{"style":715},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[717],{"type":55,"value":718},"pip",{"type":49,"tag":708,"props":720,"children":722},{"style":721},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[723],{"type":55,"value":724}," install",{"type":49,"tag":708,"props":726,"children":727},{"style":721},[728],{"type":55,"value":729}," --extra-index-url",{"type":49,"tag":708,"props":731,"children":732},{"style":721},[733],{"type":55,"value":734}," https:\u002F\u002Fpypi.jetson-ai-lab.io\u002Fjp6\u002Fcu126\u002F+simple\u002F",{"type":49,"tag":708,"props":736,"children":737},{"style":721},[738],{"type":55,"value":739}," onnxruntime-gpu\n",{"type":49,"tag":58,"props":741,"children":742},{},[743,745,750,752,758],{"type":55,"value":744},"Adjust the ",{"type":49,"tag":64,"props":746,"children":748},{"className":747},[],[749],{"type":55,"value":461},{"type":55,"value":751}," portion to match the detected JetPack \u002F CUDA line. Do not present ",{"type":49,"tag":64,"props":753,"children":755},{"className":754},[],[756],{"type":55,"value":757},"pip install onnxruntime-gpu",{"type":55,"value":759}," from default PyPI as an equivalent Jetson GPU option.",{"type":49,"tag":103,"props":761,"children":763},{"id":762},"do-not-fabricate-device-facts",[764],{"type":55,"value":765},"Do not fabricate device facts",{"type":49,"tag":58,"props":767,"children":768},{},[769,771,776],{"type":55,"value":770},"Do not invent SKU names, RAM sizes, JetPack versions, CUDA versions, or GPU SM targets. Quote only what ",{"type":49,"tag":64,"props":772,"children":774},{"className":773},[],[775],{"type":55,"value":626},{"type":55,"value":777}," or the user's supplied environment reports. If a field is unavailable, omit it or say it is unknown.",{"type":49,"tag":103,"props":779,"children":781},{"id":780},"prerequisites",[782],{"type":55,"value":783},"Prerequisites",{"type":49,"tag":121,"props":785,"children":786},{},[787,792,797],{"type":49,"tag":125,"props":788,"children":789},{},[790],{"type":55,"value":791},"Run package-detection scripts on a Jetson target, not on the host workstation.",{"type":49,"tag":125,"props":793,"children":794},{},[795],{"type":55,"value":796},"Network access is needed to inspect GHCR, NGC, or Jetson AI Lab package indexes.",{"type":49,"tag":125,"props":798,"children":799},{},[800,802,807,808,814],{"type":55,"value":801},"Source device facts from ",{"type":49,"tag":64,"props":803,"children":805},{"className":804},[],[806],{"type":55,"value":626},{"type":55,"value":231},{"type":49,"tag":64,"props":809,"children":811},{"className":810},[],[812],{"type":55,"value":813},"jetson-diagnostic",{"type":55,"value":815},", or user-provided environment output before recommending tags or wheels.",{"type":49,"tag":103,"props":817,"children":819},{"id":818},"available-scripts",[820],{"type":55,"value":821},"Available Scripts",{"type":49,"tag":291,"props":823,"children":824},{},[825,845],{"type":49,"tag":295,"props":826,"children":827},{},[828],{"type":49,"tag":299,"props":829,"children":830},{},[831,836,840],{"type":49,"tag":303,"props":832,"children":833},{},[834],{"type":55,"value":835},"Script",{"type":49,"tag":303,"props":837,"children":838},{},[839],{"type":55,"value":108},{"type":49,"tag":303,"props":841,"children":842},{},[843],{"type":55,"value":844},"Arguments",{"type":49,"tag":319,"props":846,"children":847},{},[848],{"type":49,"tag":299,"props":849,"children":850},{},[851,859,864],{"type":49,"tag":326,"props":852,"children":853},{},[854],{"type":49,"tag":64,"props":855,"children":857},{"className":856},[],[858],{"type":55,"value":626},{"type":49,"tag":326,"props":860,"children":861},{},[862],{"type":55,"value":863},"Emits detected Jetson SKU\u002Fgeneration, CUDA SM hint, canonical package URLs, and a preferred vLLM image hint.",{"type":49,"tag":326,"props":865,"children":866},{},[867,873],{"type":49,"tag":64,"props":868,"children":870},{"className":869},[],[871],{"type":55,"value":872},"--human",{"type":55,"value":874}," for a readable summary; no argument for JSON.",{"type":49,"tag":58,"props":876,"children":877},{},[878,880,886,888,893,895,900],{"type":55,"value":879},"If your agent runtime supports ",{"type":49,"tag":64,"props":881,"children":883},{"className":882},[],[884],{"type":55,"value":885},"run_script",{"type":55,"value":887},", use it to run ",{"type":49,"tag":64,"props":889,"children":891},{"className":890},[],[892],{"type":55,"value":626},{"type":55,"value":894}," and read the JSON output. Otherwise run the script with ",{"type":49,"tag":64,"props":896,"children":898},{"className":897},[],[899],{"type":55,"value":43},{"type":55,"value":901}," from the repository root.",{"type":49,"tag":103,"props":903,"children":905},{"id":904},"instructions",[906],{"type":55,"value":907},"Instructions",{"type":49,"tag":200,"props":909,"children":910},{},[911,978,1019],{"type":49,"tag":125,"props":912,"children":913},{},[914,916,921,923,929,931,937,938,943,944,950,951,956,957,962,964,969,971,976],{"type":55,"value":915},"Run ",{"type":49,"tag":64,"props":917,"children":919},{"className":918},[],[920],{"type":55,"value":626},{"type":55,"value":922}," (JSON on stdout). It sources ",{"type":49,"tag":64,"props":924,"children":926},{"className":925},[],[927],{"type":55,"value":928},"skills\u002Fjetson-diagnostic\u002Fscripts\u002Fdetect_jetson.sh",{"type":55,"value":930}," and returns ",{"type":49,"tag":64,"props":932,"children":934},{"className":933},[],[935],{"type":55,"value":936},"sku",{"type":55,"value":231},{"type":49,"tag":64,"props":939,"children":941},{"className":940},[],[942],{"type":55,"value":634},{"type":55,"value":231},{"type":49,"tag":64,"props":945,"children":947},{"className":946},[],[948],{"type":55,"value":949},"product_line",{"type":55,"value":231},{"type":49,"tag":64,"props":952,"children":954},{"className":953},[],[955],{"type":55,"value":641},{"type":55,"value":231},{"type":49,"tag":64,"props":958,"children":960},{"className":959},[],[961],{"type":55,"value":648},{"type":55,"value":963},", a preferred ",{"type":49,"tag":81,"props":965,"children":966},{},[967],{"type":55,"value":968},"vLLM",{"type":55,"value":970}," image, ",{"type":49,"tag":64,"props":972,"children":974},{"className":973},[],[975],{"type":55,"value":656},{"type":55,"value":977},", and canonical URLs.",{"type":49,"tag":125,"props":979,"children":980},{},[981,982,986,988,993,995,1001,1003,1009,1011,1017],{"type":55,"value":689},{"type":49,"tag":81,"props":983,"children":984},{},[985],{"type":55,"value":718},{"type":55,"value":987},", open the devpi root in a browser, pick the ",{"type":49,"tag":81,"props":989,"children":990},{},[991],{"type":55,"value":992},"jp6",{"type":55,"value":994}," subtree that matches your CUDA line, and set ",{"type":49,"tag":64,"props":996,"children":998},{"className":997},[],[999],{"type":55,"value":1000},"--extra-index-url",{"type":55,"value":1002}," \u002F ",{"type":49,"tag":64,"props":1004,"children":1006},{"className":1005},[],[1007],{"type":55,"value":1008},"PIP_EXTRA_INDEX_URL",{"type":55,"value":1010}," — see ",{"type":49,"tag":64,"props":1012,"children":1014},{"className":1013},[],[1015],{"type":55,"value":1016},"references\u002Fpypi-jetson-ai-lab.md",{"type":55,"value":1018},".",{"type":49,"tag":125,"props":1020,"children":1021},{},[1022,1023,1027,1029,1035,1037,1043],{"type":55,"value":689},{"type":49,"tag":81,"props":1024,"children":1025},{},[1026],{"type":55,"value":14},{"type":55,"value":1028},", see ",{"type":49,"tag":64,"props":1030,"children":1032},{"className":1031},[],[1033],{"type":55,"value":1034},"references\u002Fghcr-images.md",{"type":55,"value":1036}," and ",{"type":49,"tag":64,"props":1038,"children":1040},{"className":1039},[],[1041],{"type":55,"value":1042},"jetson-llm-serve",{"type":55,"value":1044}," for vLLM.",{"type":49,"tag":103,"props":1046,"children":1048},{"id":1047},"limitations",[1049],{"type":55,"value":1050},"Limitations",{"type":49,"tag":121,"props":1052,"children":1053},{},[1054,1059,1072],{"type":49,"tag":125,"props":1055,"children":1056},{},[1057],{"type":55,"value":1058},"This skill points to package catalogs and emits compatibility hints; it does not verify that a specific model checkpoint fits in memory.",{"type":49,"tag":125,"props":1060,"children":1061},{},[1062,1064,1070],{"type":55,"value":1063},"NGC and GHCR tags change. Treat placeholder tag shapes such as ",{"type":49,"tag":64,"props":1065,"children":1067},{"className":1066},[],[1068],{"type":55,"value":1069},"\u003Ccurrent-tag>-py3",{"type":55,"value":1071}," as lookup instructions, not literal tags.",{"type":49,"tag":125,"props":1073,"children":1074},{},[1075,1077,1082,1083,1088],{"type":55,"value":1076},"If ",{"type":49,"tag":64,"props":1078,"children":1080},{"className":1079},[],[1081],{"type":55,"value":634},{"type":55,"value":159},{"type":49,"tag":64,"props":1084,"children":1086},{"className":1085},[],[1087],{"type":55,"value":656},{"type":55,"value":1089}," is unknown, do not guess a container tag.",{"type":49,"tag":103,"props":1091,"children":1093},{"id":1092},"hand-off-to",[1094],{"type":55,"value":1095},"Hand off to",{"type":49,"tag":121,"props":1097,"children":1098},{},[1099,1117,1128],{"type":49,"tag":125,"props":1100,"children":1101},{},[1102,1107,1109,1115],{"type":49,"tag":64,"props":1103,"children":1105},{"className":1104},[],[1106],{"type":55,"value":1042},{"type":55,"value":1108}," — run upstream\u002Fnative vLLM 0.20+ on Thor and Orin JetPack 7.2 \u002F L4T r39+, or ",{"type":49,"tag":64,"props":1110,"children":1112},{"className":1111},[],[1113],{"type":55,"value":1114},"vllm:latest-jetson-orin",{"type":55,"value":1116}," on older Orin.",{"type":49,"tag":125,"props":1118,"children":1119},{},[1120,1126],{"type":49,"tag":64,"props":1121,"children":1123},{"className":1122},[],[1124],{"type":55,"value":1125},"jetson-llm-benchmark",{"type":55,"value":1127}," — measure after the stack is installed.",{"type":49,"tag":125,"props":1129,"children":1130},{},[1131,1136],{"type":49,"tag":64,"props":1132,"children":1134},{"className":1133},[],[1135],{"type":55,"value":813},{"type":55,"value":1137}," — if installs succeed but runtime fails, snapshot first.",{"type":49,"tag":103,"props":1139,"children":1141},{"id":1140},"safety",[1142],{"type":55,"value":1143},"Safety",{"type":49,"tag":58,"props":1145,"children":1146},{},[1147],{"type":55,"value":1148},"Read-only: points to catalogs and emits hints; does not install or pull.",{"type":49,"tag":103,"props":1150,"children":1152},{"id":1151},"sources",[1153],{"type":55,"value":1154},"Sources",{"type":49,"tag":58,"props":1156,"children":1157},{},[1158,1164,1165,1171],{"type":49,"tag":214,"props":1159,"children":1161},{"href":216,"rel":1160},[218],[1162],{"type":55,"value":1163},"NVIDIA-AI-IOT GitHub Packages",{"type":55,"value":231},{"type":49,"tag":214,"props":1166,"children":1168},{"href":449,"rel":1167},[218],[1169],{"type":55,"value":1170},"pypi.jetson-ai-lab.io",{"type":55,"value":1018},{"type":49,"tag":1173,"props":1174,"children":1175},"style",{},[1176],{"type":55,"value":1177},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":1179,"total":1334},[1180,1198,1212,1223,1235,1249,1262,1276,1289,1300,1314,1323],{"slug":1181,"name":1181,"fn":1182,"description":1183,"org":1184,"tags":1185,"stars":1195,"repoUrl":1196,"updatedAt":1197},"nemoclaw-user-guide","retrieve NemoClaw documentation and configuration","Guides human users' AI agents to the NemoClaw docs MCP server and canonical Fern documentation in Markdown form. Use when users ask how to install, configure, operate, troubleshoot, secure, or learn NemoClaw with an AI coding assistant. Trigger keywords - nemoclaw docs, use nemoclaw with ai agent, nemoclaw mcp docs, nemoclaw install help, nemoclaw quickstart, nemoclaw markdown docs, llms.txt, agent skills.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1186,1189,1192],{"name":1187,"slug":1188,"type":15},"Documentation","documentation",{"name":1190,"slug":1191,"type":15},"MCP","mcp",{"name":1193,"slug":1194,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":1199,"name":1199,"fn":1200,"description":1201,"org":1202,"tags":1203,"stars":1209,"repoUrl":1210,"updatedAt":1211},"mcore-build-and-dependency","manage Megatron-LM development environments","Container-based dev environment setup and dependency management for Megatron-LM. Covers acquiring and launching the CI container, uv package management, and updating uv.lock.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1204,1205,1206],{"name":13,"slug":14,"type":15},{"name":17,"slug":18,"type":15},{"name":1207,"slug":1208,"type":15},"Python","python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":1213,"name":1213,"fn":1214,"description":1215,"org":1216,"tags":1217,"stars":1209,"repoUrl":1210,"updatedAt":1222},"mcore-bump-base-image","update NVIDIA PyTorch base images","Bump the NVIDIA PyTorch base image (`nvcr.io\u002Fnvidia\u002Fpytorch:YY.MM-py3`) used by Megatron-LM CI. Covers the two pin sites (GitHub CI in `docker\u002F.ngc_version.dev` and GitLab CI in `.gitlab\u002Fstages\u002F01.build.yml`), the post-bump CI loop (re-run functional tests, refresh golden values, mark broken tests), and the gotchas that bit PRs",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1218,1221],{"name":1219,"slug":1220,"type":15},"CI\u002FCD","ci-cd",{"name":17,"slug":18,"type":15},"2026-07-14T05:25:59.97109",{"slug":1224,"name":1224,"fn":1225,"description":1226,"org":1227,"tags":1228,"stars":1209,"repoUrl":1210,"updatedAt":1234},"mcore-cicd","manage CI\u002FCD pipelines for Megatron-LM","CI\u002FCD reference for Megatron-LM. Covers CI pipeline structure, PR scope labels, triggering internal GitLab CI (which force-pushes the current branch to a pull-request\u002FBRANCH ref — always dry-run and verify the destination first; never run against shared or protected branches), and CI failure investigation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1229,1230,1231],{"name":1219,"slug":1220,"type":15},{"name":17,"slug":18,"type":15},{"name":1232,"slug":1233,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":1236,"name":1236,"fn":1237,"description":1238,"org":1239,"tags":1240,"stars":1209,"repoUrl":1210,"updatedAt":1248},"mcore-create-issue","investigate CI failures and create issues","Investigate a failing GitHub Actions run or job and create a GitHub issue for the failure.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1241,1244,1245],{"name":1242,"slug":1243,"type":15},"Debugging","debugging",{"name":1232,"slug":1233,"type":15},{"name":1246,"slug":1247,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":1250,"name":1250,"fn":1251,"description":1252,"org":1253,"tags":1254,"stars":1209,"repoUrl":1210,"updatedAt":1261},"mcore-linting-and-formatting","lint and format Megatron-LM code","Linting and formatting for Megatron-LM. Covers running autoformat.sh, tools (ruff, black, isort, pylint, mypy), and code style rules.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1255,1258],{"name":1256,"slug":1257,"type":15},"Best Practices","best-practices",{"name":1259,"slug":1260,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":1263,"name":1263,"fn":1264,"description":1265,"org":1266,"tags":1267,"stars":1209,"repoUrl":1210,"updatedAt":1275},"mcore-migrate-gpt-to-hybrid","migrate Megatron-LM models to HybridModel","Migration guide for moving Megatron Core GPTModel checkpoints, model providers, training commands, and layer mappings to HybridModel.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1268,1271,1274],{"name":1269,"slug":1270,"type":15},"Machine Learning","machine-learning",{"name":1272,"slug":1273,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":1277,"name":1277,"fn":1278,"description":1279,"org":1280,"tags":1281,"stars":1209,"repoUrl":1210,"updatedAt":1288},"mcore-onboard-gb200-1node-tests","onboard functional tests for GB200","Onboard 1-node GitHub MR functional tests for GB200 from existing mr-scoped 2-node tests.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1282,1285],{"name":1283,"slug":1284,"type":15},"QA","qa",{"name":1286,"slug":1287,"type":15},"Testing","testing","2026-07-14T05:25:53.673039",{"slug":1290,"name":1290,"fn":1291,"description":1292,"org":1293,"tags":1294,"stars":1209,"repoUrl":1210,"updatedAt":1299},"mcore-run-on-slurm","launch distributed training jobs on SLURM","How to launch distributed Megatron-LM training jobs on a SLURM cluster. Covers a minimal sbatch skeleton, environment-variable setup for torch.distributed.run, CUDA_DEVICE_MAX_CONNECTIONS rules across hardware and parallelism modes, container conventions, monitoring, and per-rank failure diagnosis.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1295,1296],{"name":17,"slug":18,"type":15},{"name":1297,"slug":1298,"type":15},"Infrastructure","infrastructure","2026-07-14T05:25:49.362534",{"slug":1301,"name":1301,"fn":1302,"description":1303,"org":1304,"tags":1305,"stars":1209,"repoUrl":1210,"updatedAt":1313},"mcore-split-pr","split pull requests to reduce review load","Split a PR into multiple PRs to reduce the number of required CODEOWNERS reviewer groups.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1306,1309,1310],{"name":1307,"slug":1308,"type":15},"Code Review","code-review",{"name":1232,"slug":1233,"type":15},{"name":1311,"slug":1312,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":1315,"name":1315,"fn":1316,"description":1317,"org":1318,"tags":1319,"stars":1209,"repoUrl":1210,"updatedAt":1322},"mcore-testing","run and manage Megatron-LM tests","Test system for Megatron-LM. Covers test layout, recipe YAML structure, adding and running unit and functional tests, golden values, marker filters, and CI parity.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1320,1321],{"name":1283,"slug":1284,"type":15},{"name":1286,"slug":1287,"type":15},"2026-07-14T05:25:54.928983",{"slug":1324,"name":1324,"fn":1325,"description":1326,"org":1327,"tags":1328,"stars":1209,"repoUrl":1210,"updatedAt":1333},"nightly-sync","manage nightly main-to-dev sync workflows","Domain knowledge for the nightly main-to-dev sync workflow. Covers merge strategy, CI architecture, failure investigation, and known issues.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1329,1332],{"name":1330,"slug":1331,"type":15},"Automation","automation",{"name":1219,"slug":1220,"type":15},"2026-07-30T05:29:03.275638",496,{"items":1336,"total":1432},[1337,1354,1364,1378,1388,1403,1418],{"slug":1338,"name":1338,"fn":1339,"description":1340,"org":1341,"tags":1342,"stars":23,"repoUrl":24,"updatedAt":1353},"accelerated-computing-cudf","accelerate data processing with cuDF","Official NVIDIA-authored guidance for NVIDIA cuDF GPU DataFrames, pandas acceleration, dask-cuDF, ETL, joins, groupby, CSV\u002FParquet I\u002FO, nullable semantics, and multi-GPU DataFrame workloads.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1343,1346,1349,1350],{"name":1344,"slug":1345,"type":15},"Data Analysis","data-analysis",{"name":1347,"slug":1348,"type":15},"Data Engineering","data-engineering",{"name":9,"slug":8,"type":15},{"name":1351,"slug":1352,"type":15},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":1355,"name":1355,"fn":1356,"description":1357,"org":1358,"tags":1359,"stars":23,"repoUrl":24,"updatedAt":1363},"aiq-deploy","deploy and manage NVIDIA AI-Q infrastructure","Use when asked to install, deploy, run, validate, troubleshoot, or stop NVIDIA AI-Q Blueprint infrastructure.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1360,1361,1362],{"name":17,"slug":18,"type":15},{"name":1297,"slug":1298,"type":15},{"name":9,"slug":8,"type":15},"2026-07-14T05:29:06.667109",{"slug":1365,"name":1365,"fn":1366,"description":1367,"org":1368,"tags":1369,"stars":23,"repoUrl":24,"updatedAt":1377},"aiq-research","conduct deep research with AI-Q","Use when asked to run deep research or AI-Q research through a reachable NVIDIA AI-Q Blueprint backend.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1370,1373,1374],{"name":1371,"slug":1372,"type":15},"Agents","agents",{"name":9,"slug":8,"type":15},{"name":1375,"slug":1376,"type":15},"Research","research","2026-07-14T05:28:06.816956",{"slug":1379,"name":1379,"fn":1380,"description":1381,"org":1382,"tags":1383,"stars":23,"repoUrl":24,"updatedAt":1387},"amc-run-sample-calibration","run AMC sample dataset calibration","Run end-to-end calibration on the shipped sample dataset (sdg_08_2_sample_data_010926.zip) against a running AMC microservice. Use when user says 'test sample dataset', 'run sample calibration', 'verify AMC install', or 'launch and test'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1384,1385,1386],{"name":1344,"slug":1345,"type":15},{"name":9,"slug":8,"type":15},{"name":1286,"slug":1287,"type":15},"2026-07-17T05:29:03.913266",{"slug":1389,"name":1389,"fn":1390,"description":1391,"org":1392,"tags":1393,"stars":23,"repoUrl":24,"updatedAt":1402},"amc-run-video-calibration","calibrate video datasets with AutoMagicCalib","Calibrate a new dataset from pre-recorded video files via the AutoMagicCalib REST API. Use when user has local MP4s and says 'calibrate my videos', 'run AMC on these videos', or similar. For RTSP\u002Flive streams, use amc-run-rtsp-calibration instead.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1394,1395,1398,1399],{"name":1330,"slug":1331,"type":15},{"name":1396,"slug":1397,"type":15},"Imaging","imaging",{"name":9,"slug":8,"type":15},{"name":1400,"slug":1401,"type":15},"Video","video","2026-07-17T05:28:53.905004",{"slug":1404,"name":1404,"fn":1405,"description":1406,"org":1407,"tags":1408,"stars":23,"repoUrl":24,"updatedAt":1417},"amc-setup-calibration-stack","deploy AutoMagicCalib microservice with Docker","Launch AutoMagicCalib microservice and web UI from NGC release images via Docker Compose. Use when user says 'deploy auto calibration', 'launch auto calibration', 'launch AMC', 'start MS+UI', or 'set up auto-magic-calib'. Requires NGC API key.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1409,1410,1413,1414],{"name":17,"slug":18,"type":15},{"name":1411,"slug":1412,"type":15},"Docker","docker",{"name":9,"slug":8,"type":15},{"name":1415,"slug":1416,"type":15},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":1419,"name":1419,"fn":1420,"description":1421,"org":1422,"tags":1423,"stars":23,"repoUrl":24,"updatedAt":1431},"cudaq-guide","develop quantum applications with CUDA-Q","CUDA-Q onboarding guide for installation, test programs, GPU simulation, QPU hardware, and quantum applications.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1424,1425,1428],{"name":9,"slug":8,"type":15},{"name":1426,"slug":1427,"type":15},"Quantum Computing","quantum-computing",{"name":1429,"slug":1430,"type":15},"Simulation","simulation","2026-07-14T05:26:58.898253",305]