[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-nv-segment-ct":3,"mdc-hqy46f-key":31,"related-org-nvidia-nv-segment-ct":1492,"related-repo-nvidia-nv-segment-ct":1652},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":20,"repoUrl":21,"updatedAt":22,"license":23,"forks":24,"topics":25,"repo":26,"sourceUrl":29,"mdContent":30},"nv-segment-ct","run VISTA3D segmentation on CT volumes","Used for running NV-Segment-CT VISTA3D on CT NIfTI volumes and recording label-map evidence.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,16,17],{"name":13,"slug":14,"type":15},"Healthcare","healthcare","tag",{"name":9,"slug":8,"type":15},{"name":18,"slug":19,"type":15},"Imaging","imaging",2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:26:33.879276","Apache-2.0",281,[],{"repoUrl":21,"stars":20,"forks":24,"topics":27,"description":28},[],"AI agent skills published by NVIDIA","https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Fnv-segment-ct","---\nname: nv-segment-ct\ndescription: Used for running NV-Segment-CT VISTA3D on CT NIfTI volumes and recording label-map evidence.\nlicense: Apache-2.0\nallowed-tools: Bash\nmetadata:\n  author: NVIDIA MedTech Team\n  tags:\n    - MedTech\n    - CT\n    - segmentation\n---\n\n# NV-Segment-CT\n\n## Purpose\n- Used for running NV-Segment-CT VISTA3D on CT NIfTI volumes and recording label-map evidence. Not for clinical interpretation.\n- Use the wrapper exactly as documented; do not replace the upstream entrypoint with a handwritten implementation.\n- Manifest I\u002FO: inputs are `ct_volume`; outputs are `label_map` and `result_json`.\n\n## Instructions\n- Read `skill_manifest.yaml` before changing arguments, side effects, or validation gates.\n- Run `scripts\u002Frun_vista3d.py` through the documented command below; keep outputs under a caller-provided run directory.\n- If a host agent exposes `run_script`, use `run_script(\"scripts\u002Frun_vista3d.py\", args=[...])`; otherwise run the Bash\u002FPython command shown below.\n- Create the documented Python 3.10 virtual environment and invoke its binaries directly; do not install model dependencies into the caller's active environment.\n- Check the emitted JSON and paired verifier guidance before treating the run as evidence.\n\n## Available Scripts\n| Script | Purpose | Arguments |\n|---|---|---|\n| `scripts\u002Frun_vista3d.py` | Primary entrypoint declared by skill_manifest.yaml. | `PATH_TO_CT.nii.gz [--output-dir OUT_DIR] [--label-prompts IDS]` |\n\n## Prerequisites\n- Runtime requirements: Python 3.10 with `venv` support and GPU\u002FCUDA when declared by the manifest. Model packages come from the pinned upstream requirements file; only wrapper-specific packages are added locally.\n- Side effects: creates an isolated environment under\n  `~\u002F.cache\u002Fnvidia-skills\u002Fvenvs\u002Fnv-segment-ct-f9f5f51\u002F`, writes the downloaded\n  bundle under `skills\u002Fnv-segment-ct\u002Fbundle\u002F`, may cache model assets under\n  `~\u002F.cache\u002Fhuggingface\u002F`, and may contact `https:\u002F\u002Fhuggingface.co` and\n  `https:\u002F\u002Fraw.githubusercontent.com` during first setup; the optional spleen\n  fixture fetcher downloads MSD09 from\n  `https:\u002F\u002Fmsd-for-monai.s3-us-west-2.amazonaws.com`.\n- Run commands from the repository root unless an existing section below says otherwise.\n\n## Limitations\n- This is a thin wrapper. Inference, preprocessing, and postprocessing are delegated entirely to the official `hugging_face_pipeline.HuggingFacePipelineHelper` in bundle\u002F. Do not modify code under bundle\u002F.\n- `transformers==4.46.3` is the wrapper compatibility overlay tested with the upstream requirements' Torch 2.0.1; newer Transformers releases can disable that older Torch backend.\n- The pinned upstream requirements include Torch 2.0.1. Use only the pinned NVIDIA model assets; do not load untrusted checkpoints in this legacy reproduction environment.\n- Device auto-detected (cuda if available, else cpu); `--device` flag overrides.\n- Output may be schema-valid but semantically empty (e.g. label prompts that do not match the input anatomy). Sanity gates assert at least one foreground voxel per requested anatomy.\n- Not for clinical deployment, clinical interpretation, autonomous diagnosis, regulatory submission.\n\n## Troubleshooting\n| Error | Cause | Fix |\n|---|---|---|\n| `ensurepip is not available` while creating the environment | The host Python installation omitted its OS `venv` package. | Install the matching Python 3.10 venv support package or create the same isolated environment with `virtualenv -p python3.10`. |\n| Missing dependency or import error | Runtime package drift from `skill_manifest.yaml`. | Install the packages declared in the manifest or use the documented setup command. |\n| Empty or schema-invalid output | Wrong input path, unsupported modality, or upstream failure. | Re-run with a known fixture and inspect the wrapper JSON plus stderr. |\n| Validation gate failure | Output violated a declared engineering invariant. | Keep the failed evidence pack and use the gate message to repair inputs or wrapper code. |\n\nWraps the upstream `nvidia\u002FNV-Segment-CT` helper. The wrapper does not\nreimplement VISTA3D inference.\n\n\n## Exact Runnable Surface\n\nFor CT segmentation user runs, use this repo-root wrapper path exactly:\n\n```bash\n\"$NV_SEGMENT_CT_VENV\u002Fbin\u002Fpython\" skills\u002Fnv-segment-ct\u002Fscripts\u002Frun_vista3d.py PATH_TO_CT.nii.gz --label-prompts \"1,3,5,14\" --output-dir OUT_DIR\n```\n\nDo not invent `infer.py`, `Medical AI Skills run`, `python -m nv_segment_ct`, or anatomy-name-only flags. For spleen, liver, right kidney, and left kidney, the required VISTA3D label IDs are exactly `1,3,5,14`.\n\n## Preconditions\n\nThe skill assumes a Python 3.10 interpreter with `venv` support. Its documented\ncommand creates a dedicated environment and installs the model dependencies\nfrom `NV-Segment-CT\u002Frequirements.txt` at the immutable NVIDIA-Medtech commit\n`f9f5f51b589e5dc9c23c453cf5138398e4084056`. The Hugging Face bundle itself\ndoes not ship a `requirements.txt`.\n\nTwo one-time downloads (the documented command does the first one; the\nfixture fetch is a separate step you run when bootstrapping):\n\n```bash\n# Spleen example fixture from Decathlon MSD09 (~1.5 GB tar, ~11 MB\n# fixture extracted into skills\u002Fnv-segment-ct\u002Ffixtures\u002Fspleen_03.nii.gz):\npython skills\u002Fnv-segment-ct\u002Ffixtures\u002Ffetch_spleen_fixture.py\n```\n\nBoth downloads (the bundle below, and the fixture) are gitignored\n(Medical AI Skills policy: no medical data or model weights in git). The fetch\nscript is idempotent and caches the tar under\n`.workbench_data\u002Fdatasets\u002F` so re-runs are no-ops.\n\nRuntime needs an NVIDIA GPU with CUDA. CPU fallback is supported but slow.\n\n## Usage\n\nFrom the skills repository root, run the complete bootstrap. Invoke the virtual\nenvironment's binaries directly so the caller's active environment is not\nmodified:\n\n```bash\nexport NV_SEGMENT_CT_VENV=\"${NV_SEGMENT_CT_VENV:-$HOME\u002F.cache\u002Fnvidia-skills\u002Fvenvs\u002Fnv-segment-ct-f9f5f51}\"\nexport NV_SEGMENT_CT_REQUIREMENTS=\"${NV_SEGMENT_CT_REQUIREMENTS:-https:\u002F\u002Fraw.githubusercontent.com\u002FNVIDIA-Medtech\u002FNV-Segment-CTMR\u002Ff9f5f51b589e5dc9c23c453cf5138398e4084056\u002FNV-Segment-CT\u002Frequirements.txt}\"\n\nif [ ! -x \"$NV_SEGMENT_CT_VENV\u002Fbin\u002Fpython\" ]; then\n  python3.10 -m venv \"$NV_SEGMENT_CT_VENV\"\nfi\n\n\"$NV_SEGMENT_CT_VENV\u002Fbin\u002Fpython\" -m pip install \\\n  -r \"$NV_SEGMENT_CT_REQUIREMENTS\" \\\n  \"transformers==4.46.3\" \\\n  \"typer>=0.9\"\n\n\"$NV_SEGMENT_CT_VENV\u002Fbin\u002Fhf\" download nvidia\u002FNV-Segment-CT \\\n  --revision afb51518689f71e6abb367ee6301b2cd0225c66a \\\n  --local-dir skills\u002Fnv-segment-ct\u002Fbundle\u002F\n\n\"$NV_SEGMENT_CT_VENV\u002Fbin\u002Fpython\" skills\u002Fnv-segment-ct\u002Fscripts\u002Frun_vista3d.py PATH_TO_CT.nii.gz \\\n  --label-prompts \"1,3,5,14\" \\\n  --output-dir vista3d_outputs\n```\n\nWhen the user names anatomies, translate them to VISTA3D class IDs before\nrunning. For the common abdominal CT request:\n\n| Anatomy | VISTA3D class ID |\n|---|---:|\n| liver | 1 |\n| spleen | 3 |\n| right kidney | 5 |\n| left kidney | 14 |\n\nFor \"segment the spleen, liver, right kidney, and left kidney\", the correct\n`--label-prompts` value is exactly `\"1,3,5,14\"`. Do not substitute kidney\nIDs from another label dictionary; the wrapper validates the requested label\nset and will mark the run invalid if the emitted mask contains labels outside\nthe requested set.\n\nThe install and download steps are load-bearing. The pinned upstream file owns\nthe model environment, while Transformers and Typer support this thin wrapper.\n`hf download` pulls the ~832 MB model bundle into\n`skills\u002Fnv-segment-ct\u002Fbundle\u002F`; subsequent calls reuse the caches.\n\n`label-prompts` are VISTA3D class IDs. The evidence output records input\ngeometry, output mask path, observed label IDs, unexpected labels,\nper-class voxel counts, per-class physical volumes computed from the output\nmask header spacing, runtime, model identity, and fixed code-derived artifact\nchecks such as mask shape, affine match, label set, foreground count, and\nclass-volume bounds.\n\nPass `--ground-truth PATH` to record a reference label-map path under\n`input.ground_truth_path`. The skill does not compute Dice; that is the\npaired verifier's job.\n\nAnatomy plausibility (per-class volume bounds, fragmentation, bilateral\nsymmetry, liver larger than spleen) and optional per-class Dice\u002FIoU against\nthe recorded ground truth are checked by\n`verifiers\u002Fct_segmentation_quality_v1`.\n\nNot for clinical interpretation, production deployment, or non-CT modalities.\n",{"data":32,"body":40},{"name":4,"description":6,"license":23,"allowed-tools":33,"metadata":34},"Bash",{"author":35,"tags":36},"NVIDIA MedTech Team",[37,38,39],"MedTech","CT","segmentation",{"type":41,"children":42},"root",[43,51,58,103,109,169,175,232,238,311,317,372,378,503,517,523,528,603,637,643,678,683,722,735,740,746,751,1316,1321,1396,1417,1437,1448,1469,1481,1486],{"type":44,"tag":45,"props":46,"children":47},"element","h1",{"id":4},[48],{"type":49,"value":50},"text","NV-Segment-CT",{"type":44,"tag":52,"props":53,"children":55},"h2",{"id":54},"purpose",[56],{"type":49,"value":57},"Purpose",{"type":44,"tag":59,"props":60,"children":61},"ul",{},[62,68,73],{"type":44,"tag":63,"props":64,"children":65},"li",{},[66],{"type":49,"value":67},"Used for running NV-Segment-CT VISTA3D on CT NIfTI volumes and recording label-map evidence. Not for clinical interpretation.",{"type":44,"tag":63,"props":69,"children":70},{},[71],{"type":49,"value":72},"Use the wrapper exactly as documented; do not replace the upstream entrypoint with a handwritten implementation.",{"type":44,"tag":63,"props":74,"children":75},{},[76,78,85,87,93,95,101],{"type":49,"value":77},"Manifest I\u002FO: inputs are ",{"type":44,"tag":79,"props":80,"children":82},"code",{"className":81},[],[83],{"type":49,"value":84},"ct_volume",{"type":49,"value":86},"; outputs are ",{"type":44,"tag":79,"props":88,"children":90},{"className":89},[],[91],{"type":49,"value":92},"label_map",{"type":49,"value":94}," and ",{"type":44,"tag":79,"props":96,"children":98},{"className":97},[],[99],{"type":49,"value":100},"result_json",{"type":49,"value":102},".",{"type":44,"tag":52,"props":104,"children":106},{"id":105},"instructions",[107],{"type":49,"value":108},"Instructions",{"type":44,"tag":59,"props":110,"children":111},{},[112,125,138,159,164],{"type":44,"tag":63,"props":113,"children":114},{},[115,117,123],{"type":49,"value":116},"Read ",{"type":44,"tag":79,"props":118,"children":120},{"className":119},[],[121],{"type":49,"value":122},"skill_manifest.yaml",{"type":49,"value":124}," before changing arguments, side effects, or validation gates.",{"type":44,"tag":63,"props":126,"children":127},{},[128,130,136],{"type":49,"value":129},"Run ",{"type":44,"tag":79,"props":131,"children":133},{"className":132},[],[134],{"type":49,"value":135},"scripts\u002Frun_vista3d.py",{"type":49,"value":137}," through the documented command below; keep outputs under a caller-provided run directory.",{"type":44,"tag":63,"props":139,"children":140},{},[141,143,149,151,157],{"type":49,"value":142},"If a host agent exposes ",{"type":44,"tag":79,"props":144,"children":146},{"className":145},[],[147],{"type":49,"value":148},"run_script",{"type":49,"value":150},", use ",{"type":44,"tag":79,"props":152,"children":154},{"className":153},[],[155],{"type":49,"value":156},"run_script(\"scripts\u002Frun_vista3d.py\", args=[...])",{"type":49,"value":158},"; otherwise run the Bash\u002FPython command shown below.",{"type":44,"tag":63,"props":160,"children":161},{},[162],{"type":49,"value":163},"Create the documented Python 3.10 virtual environment and invoke its binaries directly; do not install model dependencies into the caller's active environment.",{"type":44,"tag":63,"props":165,"children":166},{},[167],{"type":49,"value":168},"Check the emitted JSON and paired verifier guidance before treating the run as evidence.",{"type":44,"tag":52,"props":170,"children":172},{"id":171},"available-scripts",[173],{"type":49,"value":174},"Available Scripts",{"type":44,"tag":176,"props":177,"children":178},"table",{},[179,202],{"type":44,"tag":180,"props":181,"children":182},"thead",{},[183],{"type":44,"tag":184,"props":185,"children":186},"tr",{},[187,193,197],{"type":44,"tag":188,"props":189,"children":190},"th",{},[191],{"type":49,"value":192},"Script",{"type":44,"tag":188,"props":194,"children":195},{},[196],{"type":49,"value":57},{"type":44,"tag":188,"props":198,"children":199},{},[200],{"type":49,"value":201},"Arguments",{"type":44,"tag":203,"props":204,"children":205},"tbody",{},[206],{"type":44,"tag":184,"props":207,"children":208},{},[209,218,223],{"type":44,"tag":210,"props":211,"children":212},"td",{},[213],{"type":44,"tag":79,"props":214,"children":216},{"className":215},[],[217],{"type":49,"value":135},{"type":44,"tag":210,"props":219,"children":220},{},[221],{"type":49,"value":222},"Primary entrypoint declared by skill_manifest.yaml.",{"type":44,"tag":210,"props":224,"children":225},{},[226],{"type":44,"tag":79,"props":227,"children":229},{"className":228},[],[230],{"type":49,"value":231},"PATH_TO_CT.nii.gz [--output-dir OUT_DIR] [--label-prompts IDS]",{"type":44,"tag":52,"props":233,"children":235},{"id":234},"prerequisites",[236],{"type":49,"value":237},"Prerequisites",{"type":44,"tag":59,"props":239,"children":240},{},[241,254,306],{"type":44,"tag":63,"props":242,"children":243},{},[244,246,252],{"type":49,"value":245},"Runtime requirements: Python 3.10 with ",{"type":44,"tag":79,"props":247,"children":249},{"className":248},[],[250],{"type":49,"value":251},"venv",{"type":49,"value":253}," support and GPU\u002FCUDA when declared by the manifest. Model packages come from the pinned upstream requirements file; only wrapper-specific packages are added locally.",{"type":44,"tag":63,"props":255,"children":256},{},[257,259,265,267,273,275,281,283,289,291,297,299,305],{"type":49,"value":258},"Side effects: creates an isolated environment under\n",{"type":44,"tag":79,"props":260,"children":262},{"className":261},[],[263],{"type":49,"value":264},"~\u002F.cache\u002Fnvidia-skills\u002Fvenvs\u002Fnv-segment-ct-f9f5f51\u002F",{"type":49,"value":266},", writes the downloaded\nbundle under ",{"type":44,"tag":79,"props":268,"children":270},{"className":269},[],[271],{"type":49,"value":272},"skills\u002Fnv-segment-ct\u002Fbundle\u002F",{"type":49,"value":274},", may cache model assets under\n",{"type":44,"tag":79,"props":276,"children":278},{"className":277},[],[279],{"type":49,"value":280},"~\u002F.cache\u002Fhuggingface\u002F",{"type":49,"value":282},", and may contact ",{"type":44,"tag":79,"props":284,"children":286},{"className":285},[],[287],{"type":49,"value":288},"https:\u002F\u002Fhuggingface.co",{"type":49,"value":290}," and\n",{"type":44,"tag":79,"props":292,"children":294},{"className":293},[],[295],{"type":49,"value":296},"https:\u002F\u002Fraw.githubusercontent.com",{"type":49,"value":298}," during first setup; the optional spleen\nfixture fetcher downloads MSD09 from\n",{"type":44,"tag":79,"props":300,"children":302},{"className":301},[],[303],{"type":49,"value":304},"https:\u002F\u002Fmsd-for-monai.s3-us-west-2.amazonaws.com",{"type":49,"value":102},{"type":44,"tag":63,"props":307,"children":308},{},[309],{"type":49,"value":310},"Run commands from the repository root unless an existing section below says otherwise.",{"type":44,"tag":52,"props":312,"children":314},{"id":313},"limitations",[315],{"type":49,"value":316},"Limitations",{"type":44,"tag":59,"props":318,"children":319},{},[320,333,344,349,362,367],{"type":44,"tag":63,"props":321,"children":322},{},[323,325,331],{"type":49,"value":324},"This is a thin wrapper. Inference, preprocessing, and postprocessing are delegated entirely to the official ",{"type":44,"tag":79,"props":326,"children":328},{"className":327},[],[329],{"type":49,"value":330},"hugging_face_pipeline.HuggingFacePipelineHelper",{"type":49,"value":332}," in bundle\u002F. Do not modify code under bundle\u002F.",{"type":44,"tag":63,"props":334,"children":335},{},[336,342],{"type":44,"tag":79,"props":337,"children":339},{"className":338},[],[340],{"type":49,"value":341},"transformers==4.46.3",{"type":49,"value":343}," is the wrapper compatibility overlay tested with the upstream requirements' Torch 2.0.1; newer Transformers releases can disable that older Torch backend.",{"type":44,"tag":63,"props":345,"children":346},{},[347],{"type":49,"value":348},"The pinned upstream requirements include Torch 2.0.1. Use only the pinned NVIDIA model assets; do not load untrusted checkpoints in this legacy reproduction environment.",{"type":44,"tag":63,"props":350,"children":351},{},[352,354,360],{"type":49,"value":353},"Device auto-detected (cuda if available, else cpu); ",{"type":44,"tag":79,"props":355,"children":357},{"className":356},[],[358],{"type":49,"value":359},"--device",{"type":49,"value":361}," flag overrides.",{"type":44,"tag":63,"props":363,"children":364},{},[365],{"type":49,"value":366},"Output may be schema-valid but semantically empty (e.g. label prompts that do not match the input anatomy). Sanity gates assert at least one foreground voxel per requested anatomy.",{"type":44,"tag":63,"props":368,"children":369},{},[370],{"type":49,"value":371},"Not for clinical deployment, clinical interpretation, autonomous diagnosis, regulatory submission.",{"type":44,"tag":52,"props":373,"children":375},{"id":374},"troubleshooting",[376],{"type":49,"value":377},"Troubleshooting",{"type":44,"tag":176,"props":379,"children":380},{},[381,402],{"type":44,"tag":180,"props":382,"children":383},{},[384],{"type":44,"tag":184,"props":385,"children":386},{},[387,392,397],{"type":44,"tag":188,"props":388,"children":389},{},[390],{"type":49,"value":391},"Error",{"type":44,"tag":188,"props":393,"children":394},{},[395],{"type":49,"value":396},"Cause",{"type":44,"tag":188,"props":398,"children":399},{},[400],{"type":49,"value":401},"Fix",{"type":44,"tag":203,"props":403,"children":404},{},[405,443,467,485],{"type":44,"tag":184,"props":406,"children":407},{},[408,419,431],{"type":44,"tag":210,"props":409,"children":410},{},[411,417],{"type":44,"tag":79,"props":412,"children":414},{"className":413},[],[415],{"type":49,"value":416},"ensurepip is not available",{"type":49,"value":418}," while creating the environment",{"type":44,"tag":210,"props":420,"children":421},{},[422,424,429],{"type":49,"value":423},"The host Python installation omitted its OS ",{"type":44,"tag":79,"props":425,"children":427},{"className":426},[],[428],{"type":49,"value":251},{"type":49,"value":430}," package.",{"type":44,"tag":210,"props":432,"children":433},{},[434,436,442],{"type":49,"value":435},"Install the matching Python 3.10 venv support package or create the same isolated environment with ",{"type":44,"tag":79,"props":437,"children":439},{"className":438},[],[440],{"type":49,"value":441},"virtualenv -p python3.10",{"type":49,"value":102},{"type":44,"tag":184,"props":444,"children":445},{},[446,451,462],{"type":44,"tag":210,"props":447,"children":448},{},[449],{"type":49,"value":450},"Missing dependency or import error",{"type":44,"tag":210,"props":452,"children":453},{},[454,456,461],{"type":49,"value":455},"Runtime package drift from ",{"type":44,"tag":79,"props":457,"children":459},{"className":458},[],[460],{"type":49,"value":122},{"type":49,"value":102},{"type":44,"tag":210,"props":463,"children":464},{},[465],{"type":49,"value":466},"Install the packages declared in the manifest or use the documented setup command.",{"type":44,"tag":184,"props":468,"children":469},{},[470,475,480],{"type":44,"tag":210,"props":471,"children":472},{},[473],{"type":49,"value":474},"Empty or schema-invalid output",{"type":44,"tag":210,"props":476,"children":477},{},[478],{"type":49,"value":479},"Wrong input path, unsupported modality, or upstream failure.",{"type":44,"tag":210,"props":481,"children":482},{},[483],{"type":49,"value":484},"Re-run with a known fixture and inspect the wrapper JSON plus stderr.",{"type":44,"tag":184,"props":486,"children":487},{},[488,493,498],{"type":44,"tag":210,"props":489,"children":490},{},[491],{"type":49,"value":492},"Validation gate failure",{"type":44,"tag":210,"props":494,"children":495},{},[496],{"type":49,"value":497},"Output violated a declared engineering invariant.",{"type":44,"tag":210,"props":499,"children":500},{},[501],{"type":49,"value":502},"Keep the failed evidence pack and use the gate message to repair inputs or wrapper code.",{"type":44,"tag":504,"props":505,"children":506},"p",{},[507,509,515],{"type":49,"value":508},"Wraps the upstream ",{"type":44,"tag":79,"props":510,"children":512},{"className":511},[],[513],{"type":49,"value":514},"nvidia\u002FNV-Segment-CT",{"type":49,"value":516}," helper. The wrapper does not\nreimplement VISTA3D inference.",{"type":44,"tag":52,"props":518,"children":520},{"id":519},"exact-runnable-surface",[521],{"type":49,"value":522},"Exact Runnable Surface",{"type":44,"tag":504,"props":524,"children":525},{},[526],{"type":49,"value":527},"For CT segmentation user runs, use this repo-root wrapper path exactly:",{"type":44,"tag":529,"props":530,"children":535},"pre",{"className":531,"code":532,"language":533,"meta":534,"style":534},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","\"$NV_SEGMENT_CT_VENV\u002Fbin\u002Fpython\" skills\u002Fnv-segment-ct\u002Fscripts\u002Frun_vista3d.py PATH_TO_CT.nii.gz --label-prompts \"1,3,5,14\" --output-dir OUT_DIR\n","bash","",[536],{"type":44,"tag":79,"props":537,"children":538},{"__ignoreMap":534},[539],{"type":44,"tag":540,"props":541,"children":544},"span",{"class":542,"line":543},"line",1,[545,551,557,562,568,573,578,584,589,593,598],{"type":44,"tag":540,"props":546,"children":548},{"style":547},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[549],{"type":49,"value":550},"\"",{"type":44,"tag":540,"props":552,"children":554},{"style":553},"--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8",[555],{"type":49,"value":556},"$NV_SEGMENT_CT_VENV",{"type":44,"tag":540,"props":558,"children":559},{"style":547},[560],{"type":49,"value":561},"\u002Fbin\u002Fpython\"",{"type":44,"tag":540,"props":563,"children":565},{"style":564},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[566],{"type":49,"value":567}," skills\u002Fnv-segment-ct\u002Fscripts\u002Frun_vista3d.py",{"type":44,"tag":540,"props":569,"children":570},{"style":564},[571],{"type":49,"value":572}," PATH_TO_CT.nii.gz",{"type":44,"tag":540,"props":574,"children":575},{"style":564},[576],{"type":49,"value":577}," --label-prompts",{"type":44,"tag":540,"props":579,"children":581},{"style":580},"--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF",[582],{"type":49,"value":583}," \"",{"type":44,"tag":540,"props":585,"children":586},{"style":564},[587],{"type":49,"value":588},"1,3,5,14",{"type":44,"tag":540,"props":590,"children":591},{"style":580},[592],{"type":49,"value":550},{"type":44,"tag":540,"props":594,"children":595},{"style":564},[596],{"type":49,"value":597}," --output-dir",{"type":44,"tag":540,"props":599,"children":600},{"style":564},[601],{"type":49,"value":602}," OUT_DIR\n",{"type":44,"tag":504,"props":604,"children":605},{},[606,608,614,616,622,623,629,631,636],{"type":49,"value":607},"Do not invent ",{"type":44,"tag":79,"props":609,"children":611},{"className":610},[],[612],{"type":49,"value":613},"infer.py",{"type":49,"value":615},", ",{"type":44,"tag":79,"props":617,"children":619},{"className":618},[],[620],{"type":49,"value":621},"Medical AI Skills run",{"type":49,"value":615},{"type":44,"tag":79,"props":624,"children":626},{"className":625},[],[627],{"type":49,"value":628},"python -m nv_segment_ct",{"type":49,"value":630},", or anatomy-name-only flags. For spleen, liver, right kidney, and left kidney, the required VISTA3D label IDs are exactly ",{"type":44,"tag":79,"props":632,"children":634},{"className":633},[],[635],{"type":49,"value":588},{"type":49,"value":102},{"type":44,"tag":52,"props":638,"children":640},{"id":639},"preconditions",[641],{"type":49,"value":642},"Preconditions",{"type":44,"tag":504,"props":644,"children":645},{},[646,648,653,655,661,663,669,671,677],{"type":49,"value":647},"The skill assumes a Python 3.10 interpreter with ",{"type":44,"tag":79,"props":649,"children":651},{"className":650},[],[652],{"type":49,"value":251},{"type":49,"value":654}," support. Its documented\ncommand creates a dedicated environment and installs the model dependencies\nfrom ",{"type":44,"tag":79,"props":656,"children":658},{"className":657},[],[659],{"type":49,"value":660},"NV-Segment-CT\u002Frequirements.txt",{"type":49,"value":662}," at the immutable NVIDIA-Medtech commit\n",{"type":44,"tag":79,"props":664,"children":666},{"className":665},[],[667],{"type":49,"value":668},"f9f5f51b589e5dc9c23c453cf5138398e4084056",{"type":49,"value":670},". The Hugging Face bundle itself\ndoes not ship a ",{"type":44,"tag":79,"props":672,"children":674},{"className":673},[],[675],{"type":49,"value":676},"requirements.txt",{"type":49,"value":102},{"type":44,"tag":504,"props":679,"children":680},{},[681],{"type":49,"value":682},"Two one-time downloads (the documented command does the first one; the\nfixture fetch is a separate step you run when bootstrapping):",{"type":44,"tag":529,"props":684,"children":686},{"className":531,"code":685,"language":533,"meta":534,"style":534},"# Spleen example fixture from Decathlon MSD09 (~1.5 GB tar, ~11 MB\n# fixture extracted into skills\u002Fnv-segment-ct\u002Ffixtures\u002Fspleen_03.nii.gz):\npython skills\u002Fnv-segment-ct\u002Ffixtures\u002Ffetch_spleen_fixture.py\n",[687],{"type":44,"tag":79,"props":688,"children":689},{"__ignoreMap":534},[690,699,708],{"type":44,"tag":540,"props":691,"children":692},{"class":542,"line":543},[693],{"type":44,"tag":540,"props":694,"children":696},{"style":695},"--shiki-light:#90A4AE;--shiki-light-font-style:italic;--shiki-default:#546E7A;--shiki-default-font-style:italic;--shiki-dark:#676E95;--shiki-dark-font-style:italic",[697],{"type":49,"value":698},"# Spleen example fixture from Decathlon MSD09 (~1.5 GB tar, ~11 MB\n",{"type":44,"tag":540,"props":700,"children":702},{"class":542,"line":701},2,[703],{"type":44,"tag":540,"props":704,"children":705},{"style":695},[706],{"type":49,"value":707},"# fixture extracted into skills\u002Fnv-segment-ct\u002Ffixtures\u002Fspleen_03.nii.gz):\n",{"type":44,"tag":540,"props":709,"children":711},{"class":542,"line":710},3,[712,717],{"type":44,"tag":540,"props":713,"children":714},{"style":547},[715],{"type":49,"value":716},"python",{"type":44,"tag":540,"props":718,"children":719},{"style":564},[720],{"type":49,"value":721}," skills\u002Fnv-segment-ct\u002Ffixtures\u002Ffetch_spleen_fixture.py\n",{"type":44,"tag":504,"props":723,"children":724},{},[725,727,733],{"type":49,"value":726},"Both downloads (the bundle below, and the fixture) are gitignored\n(Medical AI Skills policy: no medical data or model weights in git). The fetch\nscript is idempotent and caches the tar under\n",{"type":44,"tag":79,"props":728,"children":730},{"className":729},[],[731],{"type":49,"value":732},".workbench_data\u002Fdatasets\u002F",{"type":49,"value":734}," so re-runs are no-ops.",{"type":44,"tag":504,"props":736,"children":737},{},[738],{"type":49,"value":739},"Runtime needs an NVIDIA GPU with CUDA. CPU fallback is supported but slow.",{"type":44,"tag":52,"props":741,"children":743},{"id":742},"usage",[744],{"type":49,"value":745},"Usage",{"type":44,"tag":504,"props":747,"children":748},{},[749],{"type":49,"value":750},"From the skills repository root, run the complete bootstrap. Invoke the virtual\nenvironment's binaries directly so the caller's active environment is not\nmodified:",{"type":44,"tag":529,"props":752,"children":754},{"className":531,"code":753,"language":533,"meta":534,"style":534},"export NV_SEGMENT_CT_VENV=\"${NV_SEGMENT_CT_VENV:-$HOME\u002F.cache\u002Fnvidia-skills\u002Fvenvs\u002Fnv-segment-ct-f9f5f51}\"\nexport NV_SEGMENT_CT_REQUIREMENTS=\"${NV_SEGMENT_CT_REQUIREMENTS:-https:\u002F\u002Fraw.githubusercontent.com\u002FNVIDIA-Medtech\u002FNV-Segment-CTMR\u002Ff9f5f51b589e5dc9c23c453cf5138398e4084056\u002FNV-Segment-CT\u002Frequirements.txt}\"\n\nif [ ! -x \"$NV_SEGMENT_CT_VENV\u002Fbin\u002Fpython\" ]; then\n  python3.10 -m venv \"$NV_SEGMENT_CT_VENV\"\nfi\n\n\"$NV_SEGMENT_CT_VENV\u002Fbin\u002Fpython\" -m pip install \\\n  -r \"$NV_SEGMENT_CT_REQUIREMENTS\" \\\n  \"transformers==4.46.3\" \\\n  \"typer>=0.9\"\n\n\"$NV_SEGMENT_CT_VENV\u002Fbin\u002Fhf\" download nvidia\u002FNV-Segment-CT \\\n  --revision afb51518689f71e6abb367ee6301b2cd0225c66a \\\n  --local-dir skills\u002Fnv-segment-ct\u002Fbundle\u002F\n\n\"$NV_SEGMENT_CT_VENV\u002Fbin\u002Fpython\" skills\u002Fnv-segment-ct\u002Fscripts\u002Frun_vista3d.py PATH_TO_CT.nii.gz \\\n  --label-prompts \"1,3,5,14\" \\\n  --output-dir vista3d_outputs\n",[755],{"type":44,"tag":79,"props":756,"children":757},{"__ignoreMap":534},[758,843,961,970,1022,1054,1063,1071,1106,1132,1153,1170,1178,1209,1227,1241,1249,1277,1302],{"type":44,"tag":540,"props":759,"children":760},{"class":542,"line":543},[761,767,772,777,782,787,792,797,802,806,811,815,820,824,829,833,838],{"type":44,"tag":540,"props":762,"children":764},{"style":763},"--shiki-light:#9C3EDA;--shiki-default:#C792EA;--shiki-dark:#C792EA",[765],{"type":49,"value":766},"export",{"type":44,"tag":540,"props":768,"children":769},{"style":553},[770],{"type":49,"value":771}," NV_SEGMENT_CT_VENV",{"type":44,"tag":540,"props":773,"children":774},{"style":580},[775],{"type":49,"value":776},"=",{"type":44,"tag":540,"props":778,"children":779},{"style":580},[780],{"type":49,"value":781},"\"${",{"type":44,"tag":540,"props":783,"children":784},{"style":553},[785],{"type":49,"value":786},"NV_SEGMENT_CT_VENV",{"type":44,"tag":540,"props":788,"children":789},{"style":580},[790],{"type":49,"value":791},":-",{"type":44,"tag":540,"props":793,"children":794},{"style":553},[795],{"type":49,"value":796},"$HOME",{"type":44,"tag":540,"props":798,"children":799},{"style":580},[800],{"type":49,"value":801},"\u002F",{"type":44,"tag":540,"props":803,"children":804},{"style":564},[805],{"type":49,"value":102},{"type":44,"tag":540,"props":807,"children":808},{"style":553},[809],{"type":49,"value":810},"cache",{"type":44,"tag":540,"props":812,"children":813},{"style":580},[814],{"type":49,"value":801},{"type":44,"tag":540,"props":816,"children":817},{"style":553},[818],{"type":49,"value":819},"nvidia-skills",{"type":44,"tag":540,"props":821,"children":822},{"style":580},[823],{"type":49,"value":801},{"type":44,"tag":540,"props":825,"children":826},{"style":553},[827],{"type":49,"value":828},"venvs",{"type":44,"tag":540,"props":830,"children":831},{"style":580},[832],{"type":49,"value":801},{"type":44,"tag":540,"props":834,"children":835},{"style":553},[836],{"type":49,"value":837},"nv-segment-ct-f9f5f51",{"type":44,"tag":540,"props":839,"children":840},{"style":580},[841],{"type":49,"value":842},"}\"\n",{"type":44,"tag":540,"props":844,"children":845},{"class":542,"line":701},[846,850,855,859,863,868,872,877,882,887,891,896,900,905,909,914,918,923,927,931,935,939,943,948,952,957],{"type":44,"tag":540,"props":847,"children":848},{"style":763},[849],{"type":49,"value":766},{"type":44,"tag":540,"props":851,"children":852},{"style":553},[853],{"type":49,"value":854}," NV_SEGMENT_CT_REQUIREMENTS",{"type":44,"tag":540,"props":856,"children":857},{"style":580},[858],{"type":49,"value":776},{"type":44,"tag":540,"props":860,"children":861},{"style":580},[862],{"type":49,"value":781},{"type":44,"tag":540,"props":864,"children":865},{"style":553},[866],{"type":49,"value":867},"NV_SEGMENT_CT_REQUIREMENTS",{"type":44,"tag":540,"props":869,"children":870},{"style":580},[871],{"type":49,"value":791},{"type":44,"tag":540,"props":873,"children":874},{"style":553},[875],{"type":49,"value":876},"https",{"type":44,"tag":540,"props":878,"children":879},{"style":580},[880],{"type":49,"value":881},":\u002F\u002F",{"type":44,"tag":540,"props":883,"children":884},{"style":553},[885],{"type":49,"value":886},"raw",{"type":44,"tag":540,"props":888,"children":889},{"style":564},[890],{"type":49,"value":102},{"type":44,"tag":540,"props":892,"children":893},{"style":553},[894],{"type":49,"value":895},"githubusercontent",{"type":44,"tag":540,"props":897,"children":898},{"style":564},[899],{"type":49,"value":102},{"type":44,"tag":540,"props":901,"children":902},{"style":553},[903],{"type":49,"value":904},"com",{"type":44,"tag":540,"props":906,"children":907},{"style":580},[908],{"type":49,"value":801},{"type":44,"tag":540,"props":910,"children":911},{"style":553},[912],{"type":49,"value":913},"NVIDIA-Medtech",{"type":44,"tag":540,"props":915,"children":916},{"style":580},[917],{"type":49,"value":801},{"type":44,"tag":540,"props":919,"children":920},{"style":553},[921],{"type":49,"value":922},"NV-Segment-CTMR",{"type":44,"tag":540,"props":924,"children":925},{"style":580},[926],{"type":49,"value":801},{"type":44,"tag":540,"props":928,"children":929},{"style":553},[930],{"type":49,"value":668},{"type":44,"tag":540,"props":932,"children":933},{"style":580},[934],{"type":49,"value":801},{"type":44,"tag":540,"props":936,"children":937},{"style":553},[938],{"type":49,"value":50},{"type":44,"tag":540,"props":940,"children":941},{"style":580},[942],{"type":49,"value":801},{"type":44,"tag":540,"props":944,"children":945},{"style":553},[946],{"type":49,"value":947},"requirements",{"type":44,"tag":540,"props":949,"children":950},{"style":564},[951],{"type":49,"value":102},{"type":44,"tag":540,"props":953,"children":954},{"style":553},[955],{"type":49,"value":956},"txt",{"type":44,"tag":540,"props":958,"children":959},{"style":580},[960],{"type":49,"value":842},{"type":44,"tag":540,"props":962,"children":963},{"class":542,"line":710},[964],{"type":44,"tag":540,"props":965,"children":967},{"emptyLinePlaceholder":966},true,[968],{"type":49,"value":969},"\n",{"type":44,"tag":540,"props":971,"children":973},{"class":542,"line":972},4,[974,980,985,990,995,999,1003,1008,1012,1017],{"type":44,"tag":540,"props":975,"children":977},{"style":976},"--shiki-light:#39ADB5;--shiki-light-font-style:italic;--shiki-default:#89DDFF;--shiki-default-font-style:italic;--shiki-dark:#89DDFF;--shiki-dark-font-style:italic",[978],{"type":49,"value":979},"if",{"type":44,"tag":540,"props":981,"children":982},{"style":580},[983],{"type":49,"value":984}," [",{"type":44,"tag":540,"props":986,"children":987},{"style":580},[988],{"type":49,"value":989}," !",{"type":44,"tag":540,"props":991,"children":992},{"style":580},[993],{"type":49,"value":994}," -x",{"type":44,"tag":540,"props":996,"children":997},{"style":580},[998],{"type":49,"value":583},{"type":44,"tag":540,"props":1000,"children":1001},{"style":553},[1002],{"type":49,"value":556},{"type":44,"tag":540,"props":1004,"children":1005},{"style":564},[1006],{"type":49,"value":1007},"\u002Fbin\u002Fpython",{"type":44,"tag":540,"props":1009,"children":1010},{"style":580},[1011],{"type":49,"value":550},{"type":44,"tag":540,"props":1013,"children":1014},{"style":580},[1015],{"type":49,"value":1016}," ];",{"type":44,"tag":540,"props":1018,"children":1019},{"style":976},[1020],{"type":49,"value":1021}," then\n",{"type":44,"tag":540,"props":1023,"children":1025},{"class":542,"line":1024},5,[1026,1031,1036,1041,1045,1049],{"type":44,"tag":540,"props":1027,"children":1028},{"style":547},[1029],{"type":49,"value":1030},"  python3.10",{"type":44,"tag":540,"props":1032,"children":1033},{"style":564},[1034],{"type":49,"value":1035}," -m",{"type":44,"tag":540,"props":1037,"children":1038},{"style":564},[1039],{"type":49,"value":1040}," venv",{"type":44,"tag":540,"props":1042,"children":1043},{"style":580},[1044],{"type":49,"value":583},{"type":44,"tag":540,"props":1046,"children":1047},{"style":553},[1048],{"type":49,"value":556},{"type":44,"tag":540,"props":1050,"children":1051},{"style":580},[1052],{"type":49,"value":1053},"\"\n",{"type":44,"tag":540,"props":1055,"children":1057},{"class":542,"line":1056},6,[1058],{"type":44,"tag":540,"props":1059,"children":1060},{"style":976},[1061],{"type":49,"value":1062},"fi\n",{"type":44,"tag":540,"props":1064,"children":1066},{"class":542,"line":1065},7,[1067],{"type":44,"tag":540,"props":1068,"children":1069},{"emptyLinePlaceholder":966},[1070],{"type":49,"value":969},{"type":44,"tag":540,"props":1072,"children":1074},{"class":542,"line":1073},8,[1075,1079,1083,1087,1091,1096,1101],{"type":44,"tag":540,"props":1076,"children":1077},{"style":547},[1078],{"type":49,"value":550},{"type":44,"tag":540,"props":1080,"children":1081},{"style":553},[1082],{"type":49,"value":556},{"type":44,"tag":540,"props":1084,"children":1085},{"style":547},[1086],{"type":49,"value":561},{"type":44,"tag":540,"props":1088,"children":1089},{"style":564},[1090],{"type":49,"value":1035},{"type":44,"tag":540,"props":1092,"children":1093},{"style":564},[1094],{"type":49,"value":1095}," pip",{"type":44,"tag":540,"props":1097,"children":1098},{"style":564},[1099],{"type":49,"value":1100}," install",{"type":44,"tag":540,"props":1102,"children":1103},{"style":553},[1104],{"type":49,"value":1105}," \\\n",{"type":44,"tag":540,"props":1107,"children":1109},{"class":542,"line":1108},9,[1110,1115,1119,1124,1128],{"type":44,"tag":540,"props":1111,"children":1112},{"style":564},[1113],{"type":49,"value":1114},"  -r",{"type":44,"tag":540,"props":1116,"children":1117},{"style":580},[1118],{"type":49,"value":583},{"type":44,"tag":540,"props":1120,"children":1121},{"style":553},[1122],{"type":49,"value":1123},"$NV_SEGMENT_CT_REQUIREMENTS",{"type":44,"tag":540,"props":1125,"children":1126},{"style":580},[1127],{"type":49,"value":550},{"type":44,"tag":540,"props":1129,"children":1130},{"style":553},[1131],{"type":49,"value":1105},{"type":44,"tag":540,"props":1133,"children":1135},{"class":542,"line":1134},10,[1136,1141,1145,1149],{"type":44,"tag":540,"props":1137,"children":1138},{"style":580},[1139],{"type":49,"value":1140},"  \"",{"type":44,"tag":540,"props":1142,"children":1143},{"style":564},[1144],{"type":49,"value":341},{"type":44,"tag":540,"props":1146,"children":1147},{"style":580},[1148],{"type":49,"value":550},{"type":44,"tag":540,"props":1150,"children":1151},{"style":553},[1152],{"type":49,"value":1105},{"type":44,"tag":540,"props":1154,"children":1156},{"class":542,"line":1155},11,[1157,1161,1166],{"type":44,"tag":540,"props":1158,"children":1159},{"style":580},[1160],{"type":49,"value":1140},{"type":44,"tag":540,"props":1162,"children":1163},{"style":564},[1164],{"type":49,"value":1165},"typer>=0.9",{"type":44,"tag":540,"props":1167,"children":1168},{"style":580},[1169],{"type":49,"value":1053},{"type":44,"tag":540,"props":1171,"children":1173},{"class":542,"line":1172},12,[1174],{"type":44,"tag":540,"props":1175,"children":1176},{"emptyLinePlaceholder":966},[1177],{"type":49,"value":969},{"type":44,"tag":540,"props":1179,"children":1181},{"class":542,"line":1180},13,[1182,1186,1190,1195,1200,1205],{"type":44,"tag":540,"props":1183,"children":1184},{"style":547},[1185],{"type":49,"value":550},{"type":44,"tag":540,"props":1187,"children":1188},{"style":553},[1189],{"type":49,"value":556},{"type":44,"tag":540,"props":1191,"children":1192},{"style":547},[1193],{"type":49,"value":1194},"\u002Fbin\u002Fhf\"",{"type":44,"tag":540,"props":1196,"children":1197},{"style":564},[1198],{"type":49,"value":1199}," download",{"type":44,"tag":540,"props":1201,"children":1202},{"style":564},[1203],{"type":49,"value":1204}," nvidia\u002FNV-Segment-CT",{"type":44,"tag":540,"props":1206,"children":1207},{"style":553},[1208],{"type":49,"value":1105},{"type":44,"tag":540,"props":1210,"children":1212},{"class":542,"line":1211},14,[1213,1218,1223],{"type":44,"tag":540,"props":1214,"children":1215},{"style":564},[1216],{"type":49,"value":1217},"  --revision",{"type":44,"tag":540,"props":1219,"children":1220},{"style":564},[1221],{"type":49,"value":1222}," afb51518689f71e6abb367ee6301b2cd0225c66a",{"type":44,"tag":540,"props":1224,"children":1225},{"style":553},[1226],{"type":49,"value":1105},{"type":44,"tag":540,"props":1228,"children":1230},{"class":542,"line":1229},15,[1231,1236],{"type":44,"tag":540,"props":1232,"children":1233},{"style":564},[1234],{"type":49,"value":1235},"  --local-dir",{"type":44,"tag":540,"props":1237,"children":1238},{"style":564},[1239],{"type":49,"value":1240}," skills\u002Fnv-segment-ct\u002Fbundle\u002F\n",{"type":44,"tag":540,"props":1242,"children":1244},{"class":542,"line":1243},16,[1245],{"type":44,"tag":540,"props":1246,"children":1247},{"emptyLinePlaceholder":966},[1248],{"type":49,"value":969},{"type":44,"tag":540,"props":1250,"children":1252},{"class":542,"line":1251},17,[1253,1257,1261,1265,1269,1273],{"type":44,"tag":540,"props":1254,"children":1255},{"style":547},[1256],{"type":49,"value":550},{"type":44,"tag":540,"props":1258,"children":1259},{"style":553},[1260],{"type":49,"value":556},{"type":44,"tag":540,"props":1262,"children":1263},{"style":547},[1264],{"type":49,"value":561},{"type":44,"tag":540,"props":1266,"children":1267},{"style":564},[1268],{"type":49,"value":567},{"type":44,"tag":540,"props":1270,"children":1271},{"style":564},[1272],{"type":49,"value":572},{"type":44,"tag":540,"props":1274,"children":1275},{"style":553},[1276],{"type":49,"value":1105},{"type":44,"tag":540,"props":1278,"children":1280},{"class":542,"line":1279},18,[1281,1286,1290,1294,1298],{"type":44,"tag":540,"props":1282,"children":1283},{"style":564},[1284],{"type":49,"value":1285},"  --label-prompts",{"type":44,"tag":540,"props":1287,"children":1288},{"style":580},[1289],{"type":49,"value":583},{"type":44,"tag":540,"props":1291,"children":1292},{"style":564},[1293],{"type":49,"value":588},{"type":44,"tag":540,"props":1295,"children":1296},{"style":580},[1297],{"type":49,"value":550},{"type":44,"tag":540,"props":1299,"children":1300},{"style":553},[1301],{"type":49,"value":1105},{"type":44,"tag":540,"props":1303,"children":1305},{"class":542,"line":1304},19,[1306,1311],{"type":44,"tag":540,"props":1307,"children":1308},{"style":564},[1309],{"type":49,"value":1310},"  --output-dir",{"type":44,"tag":540,"props":1312,"children":1313},{"style":564},[1314],{"type":49,"value":1315}," vista3d_outputs\n",{"type":44,"tag":504,"props":1317,"children":1318},{},[1319],{"type":49,"value":1320},"When the user names anatomies, translate them to VISTA3D class IDs before\nrunning. For the common abdominal CT request:",{"type":44,"tag":176,"props":1322,"children":1323},{},[1324,1341],{"type":44,"tag":180,"props":1325,"children":1326},{},[1327],{"type":44,"tag":184,"props":1328,"children":1329},{},[1330,1335],{"type":44,"tag":188,"props":1331,"children":1332},{},[1333],{"type":49,"value":1334},"Anatomy",{"type":44,"tag":188,"props":1336,"children":1338},{"align":1337},"right",[1339],{"type":49,"value":1340},"VISTA3D class ID",{"type":44,"tag":203,"props":1342,"children":1343},{},[1344,1357,1370,1383],{"type":44,"tag":184,"props":1345,"children":1346},{},[1347,1352],{"type":44,"tag":210,"props":1348,"children":1349},{},[1350],{"type":49,"value":1351},"liver",{"type":44,"tag":210,"props":1353,"children":1354},{"align":1337},[1355],{"type":49,"value":1356},"1",{"type":44,"tag":184,"props":1358,"children":1359},{},[1360,1365],{"type":44,"tag":210,"props":1361,"children":1362},{},[1363],{"type":49,"value":1364},"spleen",{"type":44,"tag":210,"props":1366,"children":1367},{"align":1337},[1368],{"type":49,"value":1369},"3",{"type":44,"tag":184,"props":1371,"children":1372},{},[1373,1378],{"type":44,"tag":210,"props":1374,"children":1375},{},[1376],{"type":49,"value":1377},"right kidney",{"type":44,"tag":210,"props":1379,"children":1380},{"align":1337},[1381],{"type":49,"value":1382},"5",{"type":44,"tag":184,"props":1384,"children":1385},{},[1386,1391],{"type":44,"tag":210,"props":1387,"children":1388},{},[1389],{"type":49,"value":1390},"left kidney",{"type":44,"tag":210,"props":1392,"children":1393},{"align":1337},[1394],{"type":49,"value":1395},"14",{"type":44,"tag":504,"props":1397,"children":1398},{},[1399,1401,1407,1409,1415],{"type":49,"value":1400},"For \"segment the spleen, liver, right kidney, and left kidney\", the correct\n",{"type":44,"tag":79,"props":1402,"children":1404},{"className":1403},[],[1405],{"type":49,"value":1406},"--label-prompts",{"type":49,"value":1408}," value is exactly ",{"type":44,"tag":79,"props":1410,"children":1412},{"className":1411},[],[1413],{"type":49,"value":1414},"\"1,3,5,14\"",{"type":49,"value":1416},". Do not substitute kidney\nIDs from another label dictionary; the wrapper validates the requested label\nset and will mark the run invalid if the emitted mask contains labels outside\nthe requested set.",{"type":44,"tag":504,"props":1418,"children":1419},{},[1420,1422,1428,1430,1435],{"type":49,"value":1421},"The install and download steps are load-bearing. The pinned upstream file owns\nthe model environment, while Transformers and Typer support this thin wrapper.\n",{"type":44,"tag":79,"props":1423,"children":1425},{"className":1424},[],[1426],{"type":49,"value":1427},"hf download",{"type":49,"value":1429}," pulls the ~832 MB model bundle into\n",{"type":44,"tag":79,"props":1431,"children":1433},{"className":1432},[],[1434],{"type":49,"value":272},{"type":49,"value":1436},"; subsequent calls reuse the caches.",{"type":44,"tag":504,"props":1438,"children":1439},{},[1440,1446],{"type":44,"tag":79,"props":1441,"children":1443},{"className":1442},[],[1444],{"type":49,"value":1445},"label-prompts",{"type":49,"value":1447}," are VISTA3D class IDs. The evidence output records input\ngeometry, output mask path, observed label IDs, unexpected labels,\nper-class voxel counts, per-class physical volumes computed from the output\nmask header spacing, runtime, model identity, and fixed code-derived artifact\nchecks such as mask shape, affine match, label set, foreground count, and\nclass-volume bounds.",{"type":44,"tag":504,"props":1449,"children":1450},{},[1451,1453,1459,1461,1467],{"type":49,"value":1452},"Pass ",{"type":44,"tag":79,"props":1454,"children":1456},{"className":1455},[],[1457],{"type":49,"value":1458},"--ground-truth PATH",{"type":49,"value":1460}," to record a reference label-map path under\n",{"type":44,"tag":79,"props":1462,"children":1464},{"className":1463},[],[1465],{"type":49,"value":1466},"input.ground_truth_path",{"type":49,"value":1468},". The skill does not compute Dice; that is the\npaired verifier's job.",{"type":44,"tag":504,"props":1470,"children":1471},{},[1472,1474,1480],{"type":49,"value":1473},"Anatomy plausibility (per-class volume bounds, fragmentation, bilateral\nsymmetry, liver larger than spleen) and optional per-class Dice\u002FIoU against\nthe recorded ground truth are checked by\n",{"type":44,"tag":79,"props":1475,"children":1477},{"className":1476},[],[1478],{"type":49,"value":1479},"verifiers\u002Fct_segmentation_quality_v1",{"type":49,"value":102},{"type":44,"tag":504,"props":1482,"children":1483},{},[1484],{"type":49,"value":1485},"Not for clinical interpretation, production deployment, or non-CT modalities.",{"type":44,"tag":1487,"props":1488,"children":1489},"style",{},[1490],{"type":49,"value":1491},"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":1493,"total":1651},[1494,1512,1529,1540,1552,1566,1579,1593,1606,1617,1631,1640],{"slug":1495,"name":1495,"fn":1496,"description":1497,"org":1498,"tags":1499,"stars":1509,"repoUrl":1510,"updatedAt":1511},"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},[1500,1503,1506],{"name":1501,"slug":1502,"type":15},"Documentation","documentation",{"name":1504,"slug":1505,"type":15},"MCP","mcp",{"name":1507,"slug":1508,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":1513,"name":1513,"fn":1514,"description":1515,"org":1516,"tags":1517,"stars":1526,"repoUrl":1527,"updatedAt":1528},"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},[1518,1521,1524],{"name":1519,"slug":1520,"type":15},"Containers","containers",{"name":1522,"slug":1523,"type":15},"Deployment","deployment",{"name":1525,"slug":716,"type":15},"Python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":1530,"name":1530,"fn":1531,"description":1532,"org":1533,"tags":1534,"stars":1526,"repoUrl":1527,"updatedAt":1539},"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},[1535,1538],{"name":1536,"slug":1537,"type":15},"CI\u002FCD","ci-cd",{"name":1522,"slug":1523,"type":15},"2026-07-14T05:25:59.97109",{"slug":1541,"name":1541,"fn":1542,"description":1543,"org":1544,"tags":1545,"stars":1526,"repoUrl":1527,"updatedAt":1551},"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},[1546,1547,1548],{"name":1536,"slug":1537,"type":15},{"name":1522,"slug":1523,"type":15},{"name":1549,"slug":1550,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":1553,"name":1553,"fn":1554,"description":1555,"org":1556,"tags":1557,"stars":1526,"repoUrl":1527,"updatedAt":1565},"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},[1558,1561,1562],{"name":1559,"slug":1560,"type":15},"Debugging","debugging",{"name":1549,"slug":1550,"type":15},{"name":1563,"slug":1564,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":1567,"name":1567,"fn":1568,"description":1569,"org":1570,"tags":1571,"stars":1526,"repoUrl":1527,"updatedAt":1578},"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},[1572,1575],{"name":1573,"slug":1574,"type":15},"Best Practices","best-practices",{"name":1576,"slug":1577,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":1580,"name":1580,"fn":1581,"description":1582,"org":1583,"tags":1584,"stars":1526,"repoUrl":1527,"updatedAt":1592},"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},[1585,1588,1591],{"name":1586,"slug":1587,"type":15},"Machine Learning","machine-learning",{"name":1589,"slug":1590,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":1594,"name":1594,"fn":1595,"description":1596,"org":1597,"tags":1598,"stars":1526,"repoUrl":1527,"updatedAt":1605},"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},[1599,1602],{"name":1600,"slug":1601,"type":15},"QA","qa",{"name":1603,"slug":1604,"type":15},"Testing","testing","2026-07-14T05:25:53.673039",{"slug":1607,"name":1607,"fn":1608,"description":1609,"org":1610,"tags":1611,"stars":1526,"repoUrl":1527,"updatedAt":1616},"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},[1612,1613],{"name":1522,"slug":1523,"type":15},{"name":1614,"slug":1615,"type":15},"Infrastructure","infrastructure","2026-07-14T05:25:49.362534",{"slug":1618,"name":1618,"fn":1619,"description":1620,"org":1621,"tags":1622,"stars":1526,"repoUrl":1527,"updatedAt":1630},"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},[1623,1626,1627],{"name":1624,"slug":1625,"type":15},"Code Review","code-review",{"name":1549,"slug":1550,"type":15},{"name":1628,"slug":1629,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":1632,"name":1632,"fn":1633,"description":1634,"org":1635,"tags":1636,"stars":1526,"repoUrl":1527,"updatedAt":1639},"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},[1637,1638],{"name":1600,"slug":1601,"type":15},{"name":1603,"slug":1604,"type":15},"2026-07-14T05:25:54.928983",{"slug":1641,"name":1641,"fn":1642,"description":1643,"org":1644,"tags":1645,"stars":1526,"repoUrl":1527,"updatedAt":1650},"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},[1646,1649],{"name":1647,"slug":1648,"type":15},"Automation","automation",{"name":1536,"slug":1537,"type":15},"2026-07-30T05:29:03.275638",496,{"items":1653,"total":1747},[1654,1671,1681,1695,1705,1718,1733],{"slug":1655,"name":1655,"fn":1656,"description":1657,"org":1658,"tags":1659,"stars":20,"repoUrl":21,"updatedAt":1670},"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},[1660,1663,1666,1667],{"name":1661,"slug":1662,"type":15},"Data Analysis","data-analysis",{"name":1664,"slug":1665,"type":15},"Data Engineering","data-engineering",{"name":9,"slug":8,"type":15},{"name":1668,"slug":1669,"type":15},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":1672,"name":1672,"fn":1673,"description":1674,"org":1675,"tags":1676,"stars":20,"repoUrl":21,"updatedAt":1680},"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},[1677,1678,1679],{"name":1522,"slug":1523,"type":15},{"name":1614,"slug":1615,"type":15},{"name":9,"slug":8,"type":15},"2026-07-14T05:29:06.667109",{"slug":1682,"name":1682,"fn":1683,"description":1684,"org":1685,"tags":1686,"stars":20,"repoUrl":21,"updatedAt":1694},"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},[1687,1690,1691],{"name":1688,"slug":1689,"type":15},"Agents","agents",{"name":9,"slug":8,"type":15},{"name":1692,"slug":1693,"type":15},"Research","research","2026-07-14T05:28:06.816956",{"slug":1696,"name":1696,"fn":1697,"description":1698,"org":1699,"tags":1700,"stars":20,"repoUrl":21,"updatedAt":1704},"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},[1701,1702,1703],{"name":1661,"slug":1662,"type":15},{"name":9,"slug":8,"type":15},{"name":1603,"slug":1604,"type":15},"2026-07-17T05:29:03.913266",{"slug":1706,"name":1706,"fn":1707,"description":1708,"org":1709,"tags":1710,"stars":20,"repoUrl":21,"updatedAt":1717},"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},[1711,1712,1713,1714],{"name":1647,"slug":1648,"type":15},{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":1715,"slug":1716,"type":15},"Video","video","2026-07-17T05:28:53.905004",{"slug":1719,"name":1719,"fn":1720,"description":1721,"org":1722,"tags":1723,"stars":20,"repoUrl":21,"updatedAt":1732},"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},[1724,1725,1728,1729],{"name":1522,"slug":1523,"type":15},{"name":1726,"slug":1727,"type":15},"Docker","docker",{"name":9,"slug":8,"type":15},{"name":1730,"slug":1731,"type":15},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":1734,"name":1734,"fn":1735,"description":1736,"org":1737,"tags":1738,"stars":20,"repoUrl":21,"updatedAt":1746},"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},[1739,1740,1743],{"name":9,"slug":8,"type":15},{"name":1741,"slug":1742,"type":15},"Quantum Computing","quantum-computing",{"name":1744,"slug":1745,"type":15},"Simulation","simulation","2026-07-14T05:26:58.898253",305]