[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-tao-convert-dataset-format":3,"mdc-8bpiqe-key":34,"related-repo-nvidia-tao-convert-dataset-format":1010,"related-org-nvidia-tao-convert-dataset-format":1114},{"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},"tao-convert-dataset-format","convert NVIDIA TAO DAFT datasets","Run `tao-daft convert` to convert NVIDIA TAO DAFT datasets between supported formats. Do not use for non-DAFT data. Use when the user asks to convert a DAFT dataset, change DAFT format, change a TAO dataset format, or run `tao-daft convert`.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,16,19,22],{"name":13,"slug":14,"type":15},"Datasets","datasets","tag",{"name":17,"slug":18,"type":15},"Data Engineering","data-engineering",{"name":20,"slug":21,"type":15},"Machine Learning","machine-learning",{"name":9,"slug":8,"type":15},2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:29:15.546667","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\u002Ftao-convert-dataset-format","---\nname: tao-convert-dataset-format\ndescription: Run `tao-daft convert` to convert NVIDIA TAO DAFT datasets between supported formats. Do not use for non-DAFT data.\n  Use when the user asks to convert a DAFT dataset, change DAFT format, change a TAO dataset format, or run `tao-daft convert`.\nlicense: Apache-2.0\ncompatibility: Requires Python 3.10+ and the nvidia-tao-sdk package (pip install nvidia-tao-daft).\nmetadata:\n  author: NVIDIA Corporation\n  version: \"0.1.0\"\nallowed-tools: Read Bash\ntags:\n- tao-daft\n- dataset\n- conversion\n- vlm\n- cosmos-reason\n---\n\n# Convert a TAO DAFT Dataset\n\n## Quick start\n\n```bash\ntao-daft convert \u003Csource-format> \u003Ctarget-format> --path \u003Cinput> --output \u003Coutput>\n```\n\nSource and target are positional subcommands; `--path` and `--output` are flags.\nDiscover the supported formats and per-pair flags from the leaf `--help`\n(see \"CLI conventions\" below).\n\n## Preflight\n```bash\npython -c \"import nvidia_tao_daft\" 2>\u002Fdev\u002Fnull || {\n  echo \"MISSING: tao-daft not installed. Run:\"\n  echo \"  pip install nvidia-tao-daft\"\n  exit 1\n}\n```\n\n## Quick Start\n\nDiscover the installed CLI surface before choosing format slugs, then run the\nleaf conversion command with explicit `--path` and `--output` flags:\n\n```bash\ntao-daft --version\ntao-daft convert --help\ntao-daft convert \u003Csource-format> --help\ntao-daft convert \u003Csource-format> \u003Ctarget-format> --path \u002Fpath\u002Fto\u002Fdaft --output \u002Fpath\u002Fto\u002Fconverted\n```\n\n## Purpose\n\nDrives `tao-daft convert` to transform a DAFT dataset (or a tree of\nthem) between supported formats. The CLI does the real work; the\nskill picks the right source\u002Ftarget pair and flags, then explains the\nresult.\n\nTrigger on: converting a DAFT dataset, packaging DAFT QA \u002F\nsummarization \u002F temporal tasks for VLM training, producing a\n`meta.json`-style training set, or the command `tao-daft convert`. Do\n**not** trigger for non-DAFT → DAFT conversion (COCO, YOLO, Data\nFactory JSONL) — redirect to the upstream `nvidia-tao-daft` repo's\nconverter skills.\n\nIf the user opens ambiguously, run a few `--help` calls first.\n\n## Prerequisites\n\n- `nvidia-tao-daft` installed (wheel only, not the source repo).\n  Confirm with `tao-daft --version`.\n- A DAFT dataset, or a parent directory containing many, on local\n  disk.\n\n## Instructions\n\n### CLI conventions\n\n`tao-daft` is nested argparse subcommands. The conventions below are\nstable across versions even when format names or flags change, so\n**always discover the current surface from `--help`** rather than\nrelying on names this doc happens to mention.\n\n1. **Source and target are both positional subcommands**, not\n   `--from`\u002F`--to`: `tao-daft convert \u003Csource> \u003Ctarget> [flags]`.\n   Format slugs are versioned, lowercase, dot-separated\n   (`metropolis-v3.0`, `cosmos-reason-v1.0`, ...).\n2. **Path and output are flags** — `--path PATH` (source),\n   `--output OUTPUT` (destination). Both required at the leaf;\n   passing positionally fails.\n3. **`--path` accepts both granularities** — a single scene\u002Fdataset\n   or a parent directory; the converter walks the tree.\n4. **Per-pair flags live at the leaf** — flag sets differ between\n   targets (e.g. media-handling). Always check the leaf `--help`.\n\n**Operating procedure:**\n\n1. `tao-daft --version` — confirm install, pin version in any report.\n2. `tao-daft convert --help` — list supported source formats.\n3. `tao-daft convert \u003Csource> --help` — list valid targets for that\n   source.\n4. Infer source from layout (same directory markers as the\n   `tao-validate-dataset-format` skill's \"Format inference\"). If you cannot infer\n   or the target is unspecified, ask.\n5. `tao-daft convert \u003Csource> \u003Ctarget> --help` — pick flags for the\n   user's intent (task subset, media copy vs reference, metadata).\n6. Execute, then interpret (see below).\n\n### Reading output\n\nPer-scene progress prints to stdout; non-zero exit on failure. The\nconverted dataset is written under `--output` — spot-check it with\nthe `tao-validate-dataset-format` skill before training. For large trees, capture\nthe full output and partial-read if huge.\n\n## Limitations\n\n- DAFT-supported source formats only. For non-DAFT layouts use the\n  upstream repo's converter skills.\n- Supported pairs are whatever `--help` reports for the installed\n  version — don't pass an unconfirmed pair.\n- Source and target are positional; `--path` \u002F `--output` are flags.\n- `convert` only — `validate` and `info` have their own skills.\n- Do not reimplement conversion in Python; the CLI is the spec.\n\n## Troubleshooting\n\n- **`tao-daft: command not found`** — wheel not installed; `pip\n  install nvidia-tao-daft`, verify with `tao-daft --version`.\n- **`error: argument --path\u002F--output is required`** — passed\n  positionally; move behind the flag.\n- **`invalid choice: '\u003Cformat>'`** — slug not wired up in this\n  version. Re-run the relevant `--help`.\n- **Output rejected by `tao-daft validate`** — re-check per-pair\n  flags (media handling, task subset) via leaf `--help`; a misset\n  flag often produces a structurally valid but semantically wrong\n  target.\n",{"data":35,"body":47},{"name":4,"description":6,"license":26,"compatibility":36,"metadata":37,"allowed-tools":40,"tags":41},"Requires Python 3.10+ and the nvidia-tao-sdk package (pip install nvidia-tao-daft).",{"author":38,"version":39},"NVIDIA Corporation","0.1.0","Read Bash",[42,43,44,45,46],"tao-daft","dataset","conversion","vlm","cosmos-reason",{"type":48,"children":49},"root",[50,59,66,179,209,215,339,345,363,490,496,509,545,557,563,591,597,604,626,737,745,809,815,834,840,910,916,1004],{"type":51,"tag":52,"props":53,"children":55},"element","h1",{"id":54},"convert-a-tao-daft-dataset",[56],{"type":57,"value":58},"text","Convert a TAO DAFT Dataset",{"type":51,"tag":60,"props":61,"children":63},"h2",{"id":62},"quick-start",[64],{"type":57,"value":65},"Quick start",{"type":51,"tag":67,"props":68,"children":73},"pre",{"className":69,"code":70,"language":71,"meta":72,"style":72},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","tao-daft convert \u003Csource-format> \u003Ctarget-format> --path \u003Cinput> --output \u003Coutput>\n","bash","",[74],{"type":51,"tag":75,"props":76,"children":77},"code",{"__ignoreMap":72},[78],{"type":51,"tag":79,"props":80,"children":83},"span",{"class":81,"line":82},"line",1,[84,89,95,101,106,112,117,121,126,130,134,139,143,148,152,156,161,165,170,174],{"type":51,"tag":79,"props":85,"children":87},{"style":86},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[88],{"type":57,"value":42},{"type":51,"tag":79,"props":90,"children":92},{"style":91},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[93],{"type":57,"value":94}," convert",{"type":51,"tag":79,"props":96,"children":98},{"style":97},"--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF",[99],{"type":57,"value":100}," \u003C",{"type":51,"tag":79,"props":102,"children":103},{"style":91},[104],{"type":57,"value":105},"source-forma",{"type":51,"tag":79,"props":107,"children":109},{"style":108},"--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8",[110],{"type":57,"value":111},"t",{"type":51,"tag":79,"props":113,"children":114},{"style":97},[115],{"type":57,"value":116},">",{"type":51,"tag":79,"props":118,"children":119},{"style":97},[120],{"type":57,"value":100},{"type":51,"tag":79,"props":122,"children":123},{"style":91},[124],{"type":57,"value":125},"target-forma",{"type":51,"tag":79,"props":127,"children":128},{"style":108},[129],{"type":57,"value":111},{"type":51,"tag":79,"props":131,"children":132},{"style":97},[133],{"type":57,"value":116},{"type":51,"tag":79,"props":135,"children":136},{"style":91},[137],{"type":57,"value":138}," --path",{"type":51,"tag":79,"props":140,"children":141},{"style":97},[142],{"type":57,"value":100},{"type":51,"tag":79,"props":144,"children":145},{"style":91},[146],{"type":57,"value":147},"inpu",{"type":51,"tag":79,"props":149,"children":150},{"style":108},[151],{"type":57,"value":111},{"type":51,"tag":79,"props":153,"children":154},{"style":97},[155],{"type":57,"value":116},{"type":51,"tag":79,"props":157,"children":158},{"style":91},[159],{"type":57,"value":160}," --output",{"type":51,"tag":79,"props":162,"children":163},{"style":97},[164],{"type":57,"value":100},{"type":51,"tag":79,"props":166,"children":167},{"style":91},[168],{"type":57,"value":169},"outpu",{"type":51,"tag":79,"props":171,"children":172},{"style":108},[173],{"type":57,"value":111},{"type":51,"tag":79,"props":175,"children":176},{"style":97},[177],{"type":57,"value":178},">\n",{"type":51,"tag":180,"props":181,"children":182},"p",{},[183,185,191,193,199,201,207],{"type":57,"value":184},"Source and target are positional subcommands; ",{"type":51,"tag":75,"props":186,"children":188},{"className":187},[],[189],{"type":57,"value":190},"--path",{"type":57,"value":192}," and ",{"type":51,"tag":75,"props":194,"children":196},{"className":195},[],[197],{"type":57,"value":198},"--output",{"type":57,"value":200}," are flags.\nDiscover the supported formats and per-pair flags from the leaf ",{"type":51,"tag":75,"props":202,"children":204},{"className":203},[],[205],{"type":57,"value":206},"--help",{"type":57,"value":208},"\n(see \"CLI conventions\" below).",{"type":51,"tag":60,"props":210,"children":212},{"id":211},"preflight",[213],{"type":57,"value":214},"Preflight",{"type":51,"tag":67,"props":216,"children":218},{"className":69,"code":217,"language":71,"meta":72,"style":72},"python -c \"import nvidia_tao_daft\" 2>\u002Fdev\u002Fnull || {\n  echo \"MISSING: tao-daft not installed. Run:\"\n  echo \"  pip install nvidia-tao-daft\"\n  exit 1\n}\n",[219],{"type":51,"tag":75,"props":220,"children":221},{"__ignoreMap":72},[222,270,294,315,330],{"type":51,"tag":79,"props":223,"children":224},{"class":81,"line":82},[225,230,235,240,245,250,255,260,265],{"type":51,"tag":79,"props":226,"children":227},{"style":86},[228],{"type":57,"value":229},"python",{"type":51,"tag":79,"props":231,"children":232},{"style":91},[233],{"type":57,"value":234}," -c",{"type":51,"tag":79,"props":236,"children":237},{"style":97},[238],{"type":57,"value":239}," \"",{"type":51,"tag":79,"props":241,"children":242},{"style":91},[243],{"type":57,"value":244},"import nvidia_tao_daft",{"type":51,"tag":79,"props":246,"children":247},{"style":97},[248],{"type":57,"value":249},"\"",{"type":51,"tag":79,"props":251,"children":252},{"style":97},[253],{"type":57,"value":254}," 2>",{"type":51,"tag":79,"props":256,"children":257},{"style":91},[258],{"type":57,"value":259},"\u002Fdev\u002Fnull",{"type":51,"tag":79,"props":261,"children":262},{"style":97},[263],{"type":57,"value":264}," ||",{"type":51,"tag":79,"props":266,"children":267},{"style":97},[268],{"type":57,"value":269}," {\n",{"type":51,"tag":79,"props":271,"children":273},{"class":81,"line":272},2,[274,280,284,289],{"type":51,"tag":79,"props":275,"children":277},{"style":276},"--shiki-light:#6182B8;--shiki-default:#82AAFF;--shiki-dark:#82AAFF",[278],{"type":57,"value":279},"  echo",{"type":51,"tag":79,"props":281,"children":282},{"style":97},[283],{"type":57,"value":239},{"type":51,"tag":79,"props":285,"children":286},{"style":91},[287],{"type":57,"value":288},"MISSING: tao-daft not installed. Run:",{"type":51,"tag":79,"props":290,"children":291},{"style":97},[292],{"type":57,"value":293},"\"\n",{"type":51,"tag":79,"props":295,"children":297},{"class":81,"line":296},3,[298,302,306,311],{"type":51,"tag":79,"props":299,"children":300},{"style":276},[301],{"type":57,"value":279},{"type":51,"tag":79,"props":303,"children":304},{"style":97},[305],{"type":57,"value":239},{"type":51,"tag":79,"props":307,"children":308},{"style":91},[309],{"type":57,"value":310},"  pip install nvidia-tao-daft",{"type":51,"tag":79,"props":312,"children":313},{"style":97},[314],{"type":57,"value":293},{"type":51,"tag":79,"props":316,"children":318},{"class":81,"line":317},4,[319,324],{"type":51,"tag":79,"props":320,"children":321},{"style":276},[322],{"type":57,"value":323},"  exit",{"type":51,"tag":79,"props":325,"children":327},{"style":326},"--shiki-light:#F76D47;--shiki-default:#F78C6C;--shiki-dark:#F78C6C",[328],{"type":57,"value":329}," 1\n",{"type":51,"tag":79,"props":331,"children":333},{"class":81,"line":332},5,[334],{"type":51,"tag":79,"props":335,"children":336},{"style":97},[337],{"type":57,"value":338},"}\n",{"type":51,"tag":60,"props":340,"children":342},{"id":341},"quick-start-1",[343],{"type":57,"value":344},"Quick Start",{"type":51,"tag":180,"props":346,"children":347},{},[348,350,355,356,361],{"type":57,"value":349},"Discover the installed CLI surface before choosing format slugs, then run the\nleaf conversion command with explicit ",{"type":51,"tag":75,"props":351,"children":353},{"className":352},[],[354],{"type":57,"value":190},{"type":57,"value":192},{"type":51,"tag":75,"props":357,"children":359},{"className":358},[],[360],{"type":57,"value":198},{"type":57,"value":362}," flags:",{"type":51,"tag":67,"props":364,"children":366},{"className":69,"code":365,"language":71,"meta":72,"style":72},"tao-daft --version\ntao-daft convert --help\ntao-daft convert \u003Csource-format> --help\ntao-daft convert \u003Csource-format> \u003Ctarget-format> --path \u002Fpath\u002Fto\u002Fdaft --output \u002Fpath\u002Fto\u002Fconverted\n",[367],{"type":51,"tag":75,"props":368,"children":369},{"__ignoreMap":72},[370,382,398,429],{"type":51,"tag":79,"props":371,"children":372},{"class":81,"line":82},[373,377],{"type":51,"tag":79,"props":374,"children":375},{"style":86},[376],{"type":57,"value":42},{"type":51,"tag":79,"props":378,"children":379},{"style":91},[380],{"type":57,"value":381}," --version\n",{"type":51,"tag":79,"props":383,"children":384},{"class":81,"line":272},[385,389,393],{"type":51,"tag":79,"props":386,"children":387},{"style":86},[388],{"type":57,"value":42},{"type":51,"tag":79,"props":390,"children":391},{"style":91},[392],{"type":57,"value":94},{"type":51,"tag":79,"props":394,"children":395},{"style":91},[396],{"type":57,"value":397}," --help\n",{"type":51,"tag":79,"props":399,"children":400},{"class":81,"line":296},[401,405,409,413,417,421,425],{"type":51,"tag":79,"props":402,"children":403},{"style":86},[404],{"type":57,"value":42},{"type":51,"tag":79,"props":406,"children":407},{"style":91},[408],{"type":57,"value":94},{"type":51,"tag":79,"props":410,"children":411},{"style":97},[412],{"type":57,"value":100},{"type":51,"tag":79,"props":414,"children":415},{"style":91},[416],{"type":57,"value":105},{"type":51,"tag":79,"props":418,"children":419},{"style":108},[420],{"type":57,"value":111},{"type":51,"tag":79,"props":422,"children":423},{"style":97},[424],{"type":57,"value":116},{"type":51,"tag":79,"props":426,"children":427},{"style":91},[428],{"type":57,"value":397},{"type":51,"tag":79,"props":430,"children":431},{"class":81,"line":317},[432,436,440,444,448,452,456,460,464,468,472,476,481,485],{"type":51,"tag":79,"props":433,"children":434},{"style":86},[435],{"type":57,"value":42},{"type":51,"tag":79,"props":437,"children":438},{"style":91},[439],{"type":57,"value":94},{"type":51,"tag":79,"props":441,"children":442},{"style":97},[443],{"type":57,"value":100},{"type":51,"tag":79,"props":445,"children":446},{"style":91},[447],{"type":57,"value":105},{"type":51,"tag":79,"props":449,"children":450},{"style":108},[451],{"type":57,"value":111},{"type":51,"tag":79,"props":453,"children":454},{"style":97},[455],{"type":57,"value":116},{"type":51,"tag":79,"props":457,"children":458},{"style":97},[459],{"type":57,"value":100},{"type":51,"tag":79,"props":461,"children":462},{"style":91},[463],{"type":57,"value":125},{"type":51,"tag":79,"props":465,"children":466},{"style":108},[467],{"type":57,"value":111},{"type":51,"tag":79,"props":469,"children":470},{"style":97},[471],{"type":57,"value":116},{"type":51,"tag":79,"props":473,"children":474},{"style":91},[475],{"type":57,"value":138},{"type":51,"tag":79,"props":477,"children":478},{"style":91},[479],{"type":57,"value":480}," \u002Fpath\u002Fto\u002Fdaft",{"type":51,"tag":79,"props":482,"children":483},{"style":91},[484],{"type":57,"value":160},{"type":51,"tag":79,"props":486,"children":487},{"style":91},[488],{"type":57,"value":489}," \u002Fpath\u002Fto\u002Fconverted\n",{"type":51,"tag":60,"props":491,"children":493},{"id":492},"purpose",[494],{"type":57,"value":495},"Purpose",{"type":51,"tag":180,"props":497,"children":498},{},[499,501,507],{"type":57,"value":500},"Drives ",{"type":51,"tag":75,"props":502,"children":504},{"className":503},[],[505],{"type":57,"value":506},"tao-daft convert",{"type":57,"value":508}," to transform a DAFT dataset (or a tree of\nthem) between supported formats. The CLI does the real work; the\nskill picks the right source\u002Ftarget pair and flags, then explains the\nresult.",{"type":51,"tag":180,"props":510,"children":511},{},[512,514,520,522,527,529,535,537,543],{"type":57,"value":513},"Trigger on: converting a DAFT dataset, packaging DAFT QA \u002F\nsummarization \u002F temporal tasks for VLM training, producing a\n",{"type":51,"tag":75,"props":515,"children":517},{"className":516},[],[518],{"type":57,"value":519},"meta.json",{"type":57,"value":521},"-style training set, or the command ",{"type":51,"tag":75,"props":523,"children":525},{"className":524},[],[526],{"type":57,"value":506},{"type":57,"value":528},". Do\n",{"type":51,"tag":530,"props":531,"children":532},"strong",{},[533],{"type":57,"value":534},"not",{"type":57,"value":536}," trigger for non-DAFT → DAFT conversion (COCO, YOLO, Data\nFactory JSONL) — redirect to the upstream ",{"type":51,"tag":75,"props":538,"children":540},{"className":539},[],[541],{"type":57,"value":542},"nvidia-tao-daft",{"type":57,"value":544}," repo's\nconverter skills.",{"type":51,"tag":180,"props":546,"children":547},{},[548,550,555],{"type":57,"value":549},"If the user opens ambiguously, run a few ",{"type":51,"tag":75,"props":551,"children":553},{"className":552},[],[554],{"type":57,"value":206},{"type":57,"value":556}," calls first.",{"type":51,"tag":60,"props":558,"children":560},{"id":559},"prerequisites",[561],{"type":57,"value":562},"Prerequisites",{"type":51,"tag":564,"props":565,"children":566},"ul",{},[567,586],{"type":51,"tag":568,"props":569,"children":570},"li",{},[571,576,578,584],{"type":51,"tag":75,"props":572,"children":574},{"className":573},[],[575],{"type":57,"value":542},{"type":57,"value":577}," installed (wheel only, not the source repo).\nConfirm with ",{"type":51,"tag":75,"props":579,"children":581},{"className":580},[],[582],{"type":57,"value":583},"tao-daft --version",{"type":57,"value":585},".",{"type":51,"tag":568,"props":587,"children":588},{},[589],{"type":57,"value":590},"A DAFT dataset, or a parent directory containing many, on local\ndisk.",{"type":51,"tag":60,"props":592,"children":594},{"id":593},"instructions",[595],{"type":57,"value":596},"Instructions",{"type":51,"tag":598,"props":599,"children":601},"h3",{"id":600},"cli-conventions",[602],{"type":57,"value":603},"CLI conventions",{"type":51,"tag":180,"props":605,"children":606},{},[607,612,614,624],{"type":51,"tag":75,"props":608,"children":610},{"className":609},[],[611],{"type":57,"value":42},{"type":57,"value":613}," is nested argparse subcommands. The conventions below are\nstable across versions even when format names or flags change, so\n",{"type":51,"tag":530,"props":615,"children":616},{},[617,619],{"type":57,"value":618},"always discover the current surface from ",{"type":51,"tag":75,"props":620,"children":622},{"className":621},[],[623],{"type":57,"value":206},{"type":57,"value":625}," rather than\nrelying on names this doc happens to mention.",{"type":51,"tag":627,"props":628,"children":629},"ol",{},[630,680,706,721],{"type":51,"tag":568,"props":631,"children":632},{},[633,638,640,646,648,654,656,662,664,670,672,678],{"type":51,"tag":530,"props":634,"children":635},{},[636],{"type":57,"value":637},"Source and target are both positional subcommands",{"type":57,"value":639},", not\n",{"type":51,"tag":75,"props":641,"children":643},{"className":642},[],[644],{"type":57,"value":645},"--from",{"type":57,"value":647},"\u002F",{"type":51,"tag":75,"props":649,"children":651},{"className":650},[],[652],{"type":57,"value":653},"--to",{"type":57,"value":655},": ",{"type":51,"tag":75,"props":657,"children":659},{"className":658},[],[660],{"type":57,"value":661},"tao-daft convert \u003Csource> \u003Ctarget> [flags]",{"type":57,"value":663},".\nFormat slugs are versioned, lowercase, dot-separated\n(",{"type":51,"tag":75,"props":665,"children":667},{"className":666},[],[668],{"type":57,"value":669},"metropolis-v3.0",{"type":57,"value":671},", ",{"type":51,"tag":75,"props":673,"children":675},{"className":674},[],[676],{"type":57,"value":677},"cosmos-reason-v1.0",{"type":57,"value":679},", ...).",{"type":51,"tag":568,"props":681,"children":682},{},[683,688,690,696,698,704],{"type":51,"tag":530,"props":684,"children":685},{},[686],{"type":57,"value":687},"Path and output are flags",{"type":57,"value":689}," — ",{"type":51,"tag":75,"props":691,"children":693},{"className":692},[],[694],{"type":57,"value":695},"--path PATH",{"type":57,"value":697}," (source),\n",{"type":51,"tag":75,"props":699,"children":701},{"className":700},[],[702],{"type":57,"value":703},"--output OUTPUT",{"type":57,"value":705}," (destination). Both required at the leaf;\npassing positionally fails.",{"type":51,"tag":568,"props":707,"children":708},{},[709,719],{"type":51,"tag":530,"props":710,"children":711},{},[712,717],{"type":51,"tag":75,"props":713,"children":715},{"className":714},[],[716],{"type":57,"value":190},{"type":57,"value":718}," accepts both granularities",{"type":57,"value":720}," — a single scene\u002Fdataset\nor a parent directory; the converter walks the tree.",{"type":51,"tag":568,"props":722,"children":723},{},[724,729,731,736],{"type":51,"tag":530,"props":725,"children":726},{},[727],{"type":57,"value":728},"Per-pair flags live at the leaf",{"type":57,"value":730}," — flag sets differ between\ntargets (e.g. media-handling). Always check the leaf ",{"type":51,"tag":75,"props":732,"children":734},{"className":733},[],[735],{"type":57,"value":206},{"type":57,"value":585},{"type":51,"tag":180,"props":738,"children":739},{},[740],{"type":51,"tag":530,"props":741,"children":742},{},[743],{"type":57,"value":744},"Operating procedure:",{"type":51,"tag":627,"props":746,"children":747},{},[748,758,769,780,793,804],{"type":51,"tag":568,"props":749,"children":750},{},[751,756],{"type":51,"tag":75,"props":752,"children":754},{"className":753},[],[755],{"type":57,"value":583},{"type":57,"value":757}," — confirm install, pin version in any report.",{"type":51,"tag":568,"props":759,"children":760},{},[761,767],{"type":51,"tag":75,"props":762,"children":764},{"className":763},[],[765],{"type":57,"value":766},"tao-daft convert --help",{"type":57,"value":768}," — list supported source formats.",{"type":51,"tag":568,"props":770,"children":771},{},[772,778],{"type":51,"tag":75,"props":773,"children":775},{"className":774},[],[776],{"type":57,"value":777},"tao-daft convert \u003Csource> --help",{"type":57,"value":779}," — list valid targets for that\nsource.",{"type":51,"tag":568,"props":781,"children":782},{},[783,785,791],{"type":57,"value":784},"Infer source from layout (same directory markers as the\n",{"type":51,"tag":75,"props":786,"children":788},{"className":787},[],[789],{"type":57,"value":790},"tao-validate-dataset-format",{"type":57,"value":792}," skill's \"Format inference\"). If you cannot infer\nor the target is unspecified, ask.",{"type":51,"tag":568,"props":794,"children":795},{},[796,802],{"type":51,"tag":75,"props":797,"children":799},{"className":798},[],[800],{"type":57,"value":801},"tao-daft convert \u003Csource> \u003Ctarget> --help",{"type":57,"value":803}," — pick flags for the\nuser's intent (task subset, media copy vs reference, metadata).",{"type":51,"tag":568,"props":805,"children":806},{},[807],{"type":57,"value":808},"Execute, then interpret (see below).",{"type":51,"tag":598,"props":810,"children":812},{"id":811},"reading-output",[813],{"type":57,"value":814},"Reading output",{"type":51,"tag":180,"props":816,"children":817},{},[818,820,825,827,832],{"type":57,"value":819},"Per-scene progress prints to stdout; non-zero exit on failure. The\nconverted dataset is written under ",{"type":51,"tag":75,"props":821,"children":823},{"className":822},[],[824],{"type":57,"value":198},{"type":57,"value":826}," — spot-check it with\nthe ",{"type":51,"tag":75,"props":828,"children":830},{"className":829},[],[831],{"type":57,"value":790},{"type":57,"value":833}," skill before training. For large trees, capture\nthe full output and partial-read if huge.",{"type":51,"tag":60,"props":835,"children":837},{"id":836},"limitations",[838],{"type":57,"value":839},"Limitations",{"type":51,"tag":564,"props":841,"children":842},{},[843,848,860,879,905],{"type":51,"tag":568,"props":844,"children":845},{},[846],{"type":57,"value":847},"DAFT-supported source formats only. For non-DAFT layouts use the\nupstream repo's converter skills.",{"type":51,"tag":568,"props":849,"children":850},{},[851,853,858],{"type":57,"value":852},"Supported pairs are whatever ",{"type":51,"tag":75,"props":854,"children":856},{"className":855},[],[857],{"type":57,"value":206},{"type":57,"value":859}," reports for the installed\nversion — don't pass an unconfirmed pair.",{"type":51,"tag":568,"props":861,"children":862},{},[863,865,870,872,877],{"type":57,"value":864},"Source and target are positional; ",{"type":51,"tag":75,"props":866,"children":868},{"className":867},[],[869],{"type":57,"value":190},{"type":57,"value":871}," \u002F ",{"type":51,"tag":75,"props":873,"children":875},{"className":874},[],[876],{"type":57,"value":198},{"type":57,"value":878}," are flags.",{"type":51,"tag":568,"props":880,"children":881},{},[882,888,890,896,897,903],{"type":51,"tag":75,"props":883,"children":885},{"className":884},[],[886],{"type":57,"value":887},"convert",{"type":57,"value":889}," only — ",{"type":51,"tag":75,"props":891,"children":893},{"className":892},[],[894],{"type":57,"value":895},"validate",{"type":57,"value":192},{"type":51,"tag":75,"props":898,"children":900},{"className":899},[],[901],{"type":57,"value":902},"info",{"type":57,"value":904}," have their own skills.",{"type":51,"tag":568,"props":906,"children":907},{},[908],{"type":57,"value":909},"Do not reimplement conversion in Python; the CLI is the spec.",{"type":51,"tag":60,"props":911,"children":913},{"id":912},"troubleshooting",[914],{"type":57,"value":915},"Troubleshooting",{"type":51,"tag":564,"props":917,"children":918},{},[919,947,961,981],{"type":51,"tag":568,"props":920,"children":921},{},[922,931,933,939,941,946],{"type":51,"tag":530,"props":923,"children":924},{},[925],{"type":51,"tag":75,"props":926,"children":928},{"className":927},[],[929],{"type":57,"value":930},"tao-daft: command not found",{"type":57,"value":932}," — wheel not installed; ",{"type":51,"tag":75,"props":934,"children":936},{"className":935},[],[937],{"type":57,"value":938},"pip install nvidia-tao-daft",{"type":57,"value":940},", verify with ",{"type":51,"tag":75,"props":942,"children":944},{"className":943},[],[945],{"type":57,"value":583},{"type":57,"value":585},{"type":51,"tag":568,"props":948,"children":949},{},[950,959],{"type":51,"tag":530,"props":951,"children":952},{},[953],{"type":51,"tag":75,"props":954,"children":956},{"className":955},[],[957],{"type":57,"value":958},"error: argument --path\u002F--output is required",{"type":57,"value":960}," — passed\npositionally; move behind the flag.",{"type":51,"tag":568,"props":962,"children":963},{},[964,973,975,980],{"type":51,"tag":530,"props":965,"children":966},{},[967],{"type":51,"tag":75,"props":968,"children":970},{"className":969},[],[971],{"type":57,"value":972},"invalid choice: '\u003Cformat>'",{"type":57,"value":974}," — slug not wired up in this\nversion. Re-run the relevant ",{"type":51,"tag":75,"props":976,"children":978},{"className":977},[],[979],{"type":57,"value":206},{"type":57,"value":585},{"type":51,"tag":568,"props":982,"children":983},{},[984,995,997,1002],{"type":51,"tag":530,"props":985,"children":986},{},[987,989],{"type":57,"value":988},"Output rejected by ",{"type":51,"tag":75,"props":990,"children":992},{"className":991},[],[993],{"type":57,"value":994},"tao-daft validate",{"type":57,"value":996}," — re-check per-pair\nflags (media handling, task subset) via leaf ",{"type":51,"tag":75,"props":998,"children":1000},{"className":999},[],[1001],{"type":57,"value":206},{"type":57,"value":1003},"; a misset\nflag often produces a structurally valid but semantically wrong\ntarget.",{"type":51,"tag":1005,"props":1006,"children":1007},"style",{},[1008],{"type":57,"value":1009},"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":1011,"total":1113},[1012,1027,1041,1055,1067,1084,1099],{"slug":1013,"name":1013,"fn":1014,"description":1015,"org":1016,"tags":1017,"stars":23,"repoUrl":24,"updatedAt":1026},"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},[1018,1021,1022,1023],{"name":1019,"slug":1020,"type":15},"Data Analysis","data-analysis",{"name":17,"slug":18,"type":15},{"name":9,"slug":8,"type":15},{"name":1024,"slug":1025,"type":15},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":1028,"name":1028,"fn":1029,"description":1030,"org":1031,"tags":1032,"stars":23,"repoUrl":24,"updatedAt":1040},"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},[1033,1036,1039],{"name":1034,"slug":1035,"type":15},"Deployment","deployment",{"name":1037,"slug":1038,"type":15},"Infrastructure","infrastructure",{"name":9,"slug":8,"type":15},"2026-07-14T05:29:06.667109",{"slug":1042,"name":1042,"fn":1043,"description":1044,"org":1045,"tags":1046,"stars":23,"repoUrl":24,"updatedAt":1054},"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},[1047,1050,1051],{"name":1048,"slug":1049,"type":15},"Agents","agents",{"name":9,"slug":8,"type":15},{"name":1052,"slug":1053,"type":15},"Research","research","2026-07-14T05:28:06.816956",{"slug":1056,"name":1056,"fn":1057,"description":1058,"org":1059,"tags":1060,"stars":23,"repoUrl":24,"updatedAt":1066},"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},[1061,1062,1063],{"name":1019,"slug":1020,"type":15},{"name":9,"slug":8,"type":15},{"name":1064,"slug":1065,"type":15},"Testing","testing","2026-07-17T05:29:03.913266",{"slug":1068,"name":1068,"fn":1069,"description":1070,"org":1071,"tags":1072,"stars":23,"repoUrl":24,"updatedAt":1083},"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},[1073,1076,1079,1080],{"name":1074,"slug":1075,"type":15},"Automation","automation",{"name":1077,"slug":1078,"type":15},"Imaging","imaging",{"name":9,"slug":8,"type":15},{"name":1081,"slug":1082,"type":15},"Video","video","2026-07-17T05:28:53.905004",{"slug":1085,"name":1085,"fn":1086,"description":1087,"org":1088,"tags":1089,"stars":23,"repoUrl":24,"updatedAt":1098},"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},[1090,1091,1094,1095],{"name":1034,"slug":1035,"type":15},{"name":1092,"slug":1093,"type":15},"Docker","docker",{"name":9,"slug":8,"type":15},{"name":1096,"slug":1097,"type":15},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":1100,"name":1100,"fn":1101,"description":1102,"org":1103,"tags":1104,"stars":23,"repoUrl":24,"updatedAt":1112},"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},[1105,1106,1109],{"name":9,"slug":8,"type":15},{"name":1107,"slug":1108,"type":15},"Quantum Computing","quantum-computing",{"name":1110,"slug":1111,"type":15},"Simulation","simulation","2026-07-14T05:26:58.898253",305,{"items":1115,"total":1263},[1116,1134,1149,1160,1172,1186,1199,1211,1222,1231,1245,1254],{"slug":1117,"name":1117,"fn":1118,"description":1119,"org":1120,"tags":1121,"stars":1131,"repoUrl":1132,"updatedAt":1133},"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},[1122,1125,1128],{"name":1123,"slug":1124,"type":15},"Documentation","documentation",{"name":1126,"slug":1127,"type":15},"MCP","mcp",{"name":1129,"slug":1130,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":1135,"name":1135,"fn":1136,"description":1137,"org":1138,"tags":1139,"stars":1146,"repoUrl":1147,"updatedAt":1148},"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},[1140,1143,1144],{"name":1141,"slug":1142,"type":15},"Containers","containers",{"name":1034,"slug":1035,"type":15},{"name":1145,"slug":229,"type":15},"Python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":1150,"name":1150,"fn":1151,"description":1152,"org":1153,"tags":1154,"stars":1146,"repoUrl":1147,"updatedAt":1159},"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},[1155,1158],{"name":1156,"slug":1157,"type":15},"CI\u002FCD","ci-cd",{"name":1034,"slug":1035,"type":15},"2026-07-14T05:25:59.97109",{"slug":1161,"name":1161,"fn":1162,"description":1163,"org":1164,"tags":1165,"stars":1146,"repoUrl":1147,"updatedAt":1171},"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},[1166,1167,1168],{"name":1156,"slug":1157,"type":15},{"name":1034,"slug":1035,"type":15},{"name":1169,"slug":1170,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":1173,"name":1173,"fn":1174,"description":1175,"org":1176,"tags":1177,"stars":1146,"repoUrl":1147,"updatedAt":1185},"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},[1178,1181,1182],{"name":1179,"slug":1180,"type":15},"Debugging","debugging",{"name":1169,"slug":1170,"type":15},{"name":1183,"slug":1184,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":1187,"name":1187,"fn":1188,"description":1189,"org":1190,"tags":1191,"stars":1146,"repoUrl":1147,"updatedAt":1198},"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},[1192,1195],{"name":1193,"slug":1194,"type":15},"Best Practices","best-practices",{"name":1196,"slug":1197,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":1200,"name":1200,"fn":1201,"description":1202,"org":1203,"tags":1204,"stars":1146,"repoUrl":1147,"updatedAt":1210},"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},[1205,1206,1209],{"name":20,"slug":21,"type":15},{"name":1207,"slug":1208,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":1212,"name":1212,"fn":1213,"description":1214,"org":1215,"tags":1216,"stars":1146,"repoUrl":1147,"updatedAt":1221},"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},[1217,1220],{"name":1218,"slug":1219,"type":15},"QA","qa",{"name":1064,"slug":1065,"type":15},"2026-07-14T05:25:53.673039",{"slug":1223,"name":1223,"fn":1224,"description":1225,"org":1226,"tags":1227,"stars":1146,"repoUrl":1147,"updatedAt":1230},"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},[1228,1229],{"name":1034,"slug":1035,"type":15},{"name":1037,"slug":1038,"type":15},"2026-07-14T05:25:49.362534",{"slug":1232,"name":1232,"fn":1233,"description":1234,"org":1235,"tags":1236,"stars":1146,"repoUrl":1147,"updatedAt":1244},"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},[1237,1240,1241],{"name":1238,"slug":1239,"type":15},"Code Review","code-review",{"name":1169,"slug":1170,"type":15},{"name":1242,"slug":1243,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":1246,"name":1246,"fn":1247,"description":1248,"org":1249,"tags":1250,"stars":1146,"repoUrl":1147,"updatedAt":1253},"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},[1251,1252],{"name":1218,"slug":1219,"type":15},{"name":1064,"slug":1065,"type":15},"2026-07-14T05:25:54.928983",{"slug":1255,"name":1255,"fn":1256,"description":1257,"org":1258,"tags":1259,"stars":1146,"repoUrl":1147,"updatedAt":1262},"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},[1260,1261],{"name":1074,"slug":1075,"type":15},{"name":1156,"slug":1157,"type":15},"2026-07-30T05:29:03.275638",496]