[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-osmo-user":3,"mdc-kwbdi2-key":37,"related-repo-nvidia-osmo-user":1035,"related-org-nvidia-osmo-user":1078},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":26,"repoUrl":27,"updatedAt":28,"license":29,"forks":30,"topics":31,"repo":32,"sourceUrl":35,"mdContent":36},"osmo-user","manage cloud robotics workflows with OSMO","Drive the OSMO CLI for cloud-robotics compute on behalf of an end user: check resources, submit\u002Fmonitor\u002Fdebug\u002Fexplain workflows, fetch logs and Grafana\u002FKubernetes links, inspect direct data storage, manage workflow apps, and set workflow credentials. Use whenever the user asks about OSMO pools, quota, GPUs, or nodes, or about submitting, listing, querying, monitoring, or troubleshooting workflows — including failed, PENDING, queued, stuck, or image-pull-blocked workflows, or when they ask to inspect or transfer direct storage URIs such as `s3:\u002F\u002F...`, even when they describe a workflow, cluster resource, or storage URI without saying \"OSMO\". Do not use for Kubernetes admin, server-side `osmo config` changes, OSMO install\u002Fdeploy, non-OSMO compute, or general NVIDIA hardware questions.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,16,19,20,23],{"name":13,"slug":14,"type":15},"Monitoring","monitoring","tag",{"name":17,"slug":18,"type":15},"CLI","cli",{"name":9,"slug":8,"type":15},{"name":21,"slug":22,"type":15},"Grafana","grafana",{"name":24,"slug":25,"type":15},"Kubernetes","kubernetes",192,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FOSMO","2026-07-14T05:33:25.461692",null,40,[],{"repoUrl":27,"stars":26,"forks":30,"topics":33,"description":34},[],"The developer-first platform for scaling complex Physical AI workloads across heterogeneous compute—unifying training GPUs, simulation clusters, and edge devices in a simple YAML","https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FOSMO\u002Ftree\u002FHEAD\u002Fskills\u002Fosmo-user","---\nname: osmo-user\ndescription: >\n  Drive the OSMO CLI for cloud-robotics compute on behalf of an end user:\n  check resources, submit\u002Fmonitor\u002Fdebug\u002Fexplain workflows, fetch logs and\n  Grafana\u002FKubernetes links, inspect direct data storage, manage workflow apps,\n  and set workflow credentials.\n  Use whenever the user asks about OSMO pools, quota, GPUs, or nodes, or about\n  submitting, listing, querying, monitoring, or troubleshooting workflows —\n  including failed, PENDING, queued, stuck, or image-pull-blocked workflows, or\n  when they ask to inspect or transfer direct storage URIs such as `s3:\u002F\u002F...`,\n  even when they describe a workflow, cluster resource, or storage URI without\n  saying \"OSMO\".\n  Do not use for Kubernetes admin, server-side `osmo config` changes, OSMO\n  install\u002Fdeploy, non-OSMO compute, or general NVIDIA hardware questions.\n---\n\n# osmo-user\n\nRun OSMO CLI workflows from natural-language requests. Keep this file as a\nrouter: load only the reference files needed for the current task.\n\n## Prerequisites\n\nBefore the first OSMO command in a conversation:\n\n0. For cancel\u002Fdelete\u002Fforce\u002Fdestructive requests, ask for explicit confirmation\n   before running any `osmo` command, including `osmo --version` or query\n   commands. After confirmation, continue with the checks below.\n1. Confirm the CLI is available: `osmo --version`. If it fails, tell the user\n   the OSMO CLI is unavailable and stop.\n2. If any command returns an authentication error, ask the user to run\n   `osmo login` and stop until they confirm.\n3. Resource and workflow operations rely on the user's profile and pool access\n   (`osmo profile list`, `osmo pool list`).\n\n## Operating Rules\n\n- Classify the request using the Reference Routing section below, then load only\n  the reference file(s) it names before running commands.\n- Do not guess command names or flags from memory. Use the linked reference for\n  the user's use case, then run the commands yourself.\n- Obtain workflow and resource state (status, logs, events, capacity, spec) by\n  running the `osmo` CLI yourself. Do not infer OSMO state by reading, cat-ing,\n  or grepping files in the workspace — run the command and use its output.\n- Never fabricate command output; report only what the `osmo` commands actually returned.\n- Cache workflow query JSON during the conversation; do not re-query just to\n  extract another field from the same response.\n- Surface `grafana_url` and `dashboard_url` according to\n  `references\u002Fworkflow-status.md`; if either value is null, say it is\n  unavailable instead of omitting it.\n- Do not edit cluster config, node taints, quota policies, or non-OSMO\n  Kubernetes resources. Those are admin-side operations.\n- Do not edit server-side OSMO configuration (`osmo config`): pod or group\n  templates, resource validations, pool\u002Fbackend config, roles, or storage\n  buckets. That is the OSMO admin surface — say it is out of scope and do not\n  attempt it here.\n\n## Default Workflow\n\n1. Complete the Prerequisites above (CLI check, auth, profile\u002Fpool access)\n   before the first OSMO command.\n2. Classify the request and read only the reference file(s) named in the\n   Reference Routing section below.\n3. Run the `osmo` commands yourself — cache the query JSON and never infer state\n   from workspace files — unless the selected reference says to spawn a subagent.\n4. Verify the outputs before reporting: surface `grafana_url`\u002F`dashboard_url`\n   (or say a null one is unavailable), confirm returned workflow IDs, and check\n   status and exit codes.\n5. Summarize in the user's terms: available capacity, workflow state, progress,\n   error cause and concrete fix, dashboard link, output, or next action.\n6. For submit -> monitor -> fix loops, keep monitoring and final reporting in\n   the main conversation; delegate only setup, submission, or log summarization.\n\n## Reference Routing\n\nClassify the request and read only the reference file(s) for the matched intent.\nEach heading is a reference file; the bullets under it are the user intents and\nexample wordings that route there. Error and failure cases are listed under the\nreference that handles them. These are routing cues, not complete command recipes.\n\n### `references\u002Fresource-check-format.md`\nResources, pools, GPUs, nodes, or quota.\n- \"What resources are available to me?\", \"Any H100s free?\", \"Do I have quota?\"\n- Discover profile\u002Fpool access and report effective capacity (`min(Quota Free, Total Free)`).\n- If pool\u002Fresource output is empty or ambiguous, re-check access here and state uncertainty instead of guessing.\n\n### `references\u002Fcli-commands.md`\nSafe end-user OSMO command lookup when no dedicated reference applies.\n- \"Check whether I can read from s3:\u002F\u002F...\", \"Download this s3:\u002F\u002F path\", \"What is the command for data download?\", \"How do I list direct storage?\", \"Set my default pool\" — use this for auth\u002Fversion\u002Fprofile\u002Fdata\u002Ftask syntax.\n- For workflow, app, credential, resource, or troubleshooting details, follow this file's routing to the dedicated reference.\n- If the requested command is `osmo config`, `osmo user`, role\u002Fbucket\u002Fadmin mutation, or Kubernetes administration, treat it as out of scope.\n\n### `references\u002Fworkflow-commands.md`\nWorkflow submit\u002Flist\u002Fquery\u002Flog\u002Fevent\u002Fspec subcommand and flag lookup.\n- \"Validate this workflow\", \"Dry run it\", \"What flags does workflow list support?\", \"How do I fetch logs?\", \"Show the rendered spec\"\n- Use for command syntax only; use `references\u002Fworkflow-submit.md`, `references\u002Fworkflow-status.md`, or `references\u002Ftroubleshooting.md` for procedures.\n\n### `references\u002Fworkflow-runtime-commands.md`\nLive workflow runtime operations.\n- \"Cancel workflow X\", \"Exec into task Y\", \"Port-forward task Z\", \"Rsync files into this workflow\", \"Add a tag to this workflow\"\n- Ask for confirmation before cancellation and other destructive operations. A\n  bare request like \"Cancel workflow X\" is not confirmation.\n\n### `references\u002Fworkflow-submit.md`\nSubmit a supplied or generated workflow.\n- \"Submit workflow.yaml\", \"Pick a free H100 pool\", \"No need to ask\" — read the supplied YAML as-is, choose a pool (use `references\u002Fresource-check-format.md` for pool selection), and submit only if authorized.\n- \"Submit this Jinja workflow with 4 GPUs\" — preserve Jinja placeholders and pass values at submit time.\n\n### `references\u002Fcookbook-fetching.md`\nGenerate a workload from an OSMO cookbook example.\n- \"Generate 1000 Isaac Sim images and submit\" — fetch\u002Fadapt the cookbook workflow, compute the run count, then submit via `references\u002Fworkflow-submit.md`.\n\n### `references\u002Fworkflow-status.md`\nStatus, logs, links, live metrics, recent workflows, and workflow explanation.\n- \"Show my recent workflows\", \"What's still running?\", \"What finished?\"\n- \"Is workflow gr00t-train-1 done?\", \"How is my run going?\", \"Show progress\" — query, fetch logs, summarize state.\n- \"How much memory\u002FGPU is it using?\", \"Open metrics\" — surface `grafana_url`; do not invent live utilization.\n- \"Give me the Kubernetes dashboard link\", \"I want to inspect the pod\" — surface `dashboard_url` or say it is unavailable.\n- \"What does this workflow do?\", \"Explain this run before I rerun it\" — fetch the templated spec and summarize purpose\u002Fimage\u002Fcommand\u002Foutput.\n- \"Why is it PENDING\u002Fqueued\u002Fstuck?\", \"Why won't it schedule?\" — establish state here, then `references\u002Ftroubleshooting.md`.\n- Logs or events are sparse or time out — establish state here; for failures route to `references\u002Ftroubleshooting.md`.\n\n### `references\u002Ftroubleshooting.md`\nFailed, stuck, sparse-log, or misbehaving workflows.\n- \"The logs are empty\", \"Why did it fail?\", \"Exit code 137\u002F139\u002F143\u002F127\" — match the failure signature and propose a concrete fix.\n- For private-image pull failures, establish the failure here, then read `references\u002Fworkflow-registry-credentials.md`.\n\n### `references\u002Fworkflow-credentials.md`\nWorkflow credential types and generic\u002Fdata credential setup: `REGISTRY` overview, `GENERIC` for task secrets, and `DATA` for object-storage access.\n- \"What credential type do I need?\", \"Set an API token\", \"Set a data credential\" — check existing credentials and create only the needed workflow credential.\n- Never print secret values; prefer environment variables and ask for secrets only after automatic sources are exhausted.\n\n### `references\u002Fworkflow-registry-credentials.md`\nPrivate image-pull and registry credential setup.\n- \"Use this private nvcr.io image\", \"Set up NGC credentials\", \"Create an OSMO registry credential\", \"ImagePullBackOff unauthorized\", \"pull access denied\" — check existing registry credentials and create or fix only the needed OSMO registry credential.\n- Registry image-pull credentials are managed by OSMO outside workflow YAML.\n\n### `references\u002Fvalidation-error-recovery.md`\nSubmission capacity validation errors.\n- `osmo workflow submit` returns a capacity validation error — edit only the allowed hard-coded `resources` values and resubmit.\n\n### `references\u002Fworkflow-apps.md`\nWorkflow app lifecycle.\n- \"Create an app from this workflow\", \"Publish this completed run\", \"List apps\", \"Show app parameters\", \"Update app\", \"Submit app\", \"Rename\u002Fdelete app\" — use the app lifecycle reference.\n- Ask for confirmation before create\u002Fupdate\u002Fsubmit\u002Frename\u002Fdelete unless the user already clearly authorized the action.\n\n### `references\u002Fworkflow-spec.md`\nWorkflow YAML schema and field shapes.\n- \"Is this workflow YAML valid?\", \"Add files\u002Fenvironment\u002Fresources\", \"What shape do tasks\u002Fresources\u002Fgroups use?\" — use this as a compact field map.\n- For URL inputs\u002Foutputs, also read `references\u002Fworkflow-io-spec.md`; for topology placement, also read `references\u002Fworkflow-advanced-patterns.md`.\n- For design examples, continue to `references\u002Fworkflow-patterns.md`; for checkpointing\u002Ftopology\u002Fnode exclusion, continue to `references\u002Fworkflow-advanced-patterns.md`.\n\n### `references\u002Fworkflow-io-spec.md`\nWorkflow data flow, runtime tokens, and Jinja parameter fields.\n- \"Where does `default-values` go?\", \"What shape do inputs\u002Foutputs use?\", \"How do `{{output}}` or `{{input:0}}` work?\", \"How do I use `--set`?\" — use this as the compact IO\u002Ftemplate map.\n- For full workflow design examples, continue to `references\u002Fworkflow-patterns.md`.\n\n### `references\u002Fworkflow-patterns.md`\nWorkflow structure and authoring patterns.\n- Workflow structure, multi-task execution, dependencies, Jinja templates, checkpointing, exit\u002Fretry behavior, node exclusion, or topology placement — start here; it routes to `references\u002Fworkflow-advanced-patterns.md` for niche patterns.\n\n### `references\u002Fworkflow-expert.md` (subagent prompt)\nEnd-to-end submit \u002F monitor \u002F fix loops.\n- \"Monitor this from submit to completion\", \"Fix and resubmit if it fails\" — spawn or resume this subagent for setup\u002Fsubmission; keep final monitoring and reporting in the main conversation. After a failed loop, fetch logs as directed, resume the subagent, and stop after three failures.\n\n### `references\u002Flogs-reader.md` (subagent prompt)\nMulti-task log summarization.\n- A workflow has two or more tasks and logs are needed — spawn `logs-reader` subagents as directed; do not inline all logs.\n\n### When NOT to use this skill\n- \"What GPUs does NVIDIA sell?\", \"How do I deploy OSMO?\", \"Set up AKS \u002F Kubernetes \u002F NIM Operator infrastructure\", \"Configure Kubernetes taints\", \"Edit a pod template \u002F pool quota \u002F `osmo config`\" — answer with general help or another skill; do not run `osmo`. Server-side config, cluster admin, and infrastructure scaffolding are outside this skill.\n",{"data":38,"body":39},{"name":4,"description":6},{"type":40,"children":41},"root",[42,49,55,62,67,141,147,237,243,297,303,308,319,324,349,359,364,396,406,411,446,456,461,474,483,488,508,518,523,538,547,552,616,625,630,650,660,689,702,711,716,729,739,744,766,776,781,794,804,809,856,865,870,921,930,935,950,962,967,975,986,991,1007,1013],{"type":43,"tag":44,"props":45,"children":46},"element","h1",{"id":4},[47],{"type":48,"value":4},"text",{"type":43,"tag":50,"props":51,"children":52},"p",{},[53],{"type":48,"value":54},"Run OSMO CLI workflows from natural-language requests. Keep this file as a\nrouter: load only the reference files needed for the current task.",{"type":43,"tag":56,"props":57,"children":59},"h2",{"id":58},"prerequisites",[60],{"type":48,"value":61},"Prerequisites",{"type":43,"tag":50,"props":63,"children":64},{},[65],{"type":48,"value":66},"Before the first OSMO command in a conversation:",{"type":43,"tag":68,"props":69,"children":71},"ol",{"start":70},0,[72,95,107,120],{"type":43,"tag":73,"props":74,"children":75},"li",{},[76,78,85,87,93],{"type":48,"value":77},"For cancel\u002Fdelete\u002Fforce\u002Fdestructive requests, ask for explicit confirmation\nbefore running any ",{"type":43,"tag":79,"props":80,"children":82},"code",{"className":81},[],[83],{"type":48,"value":84},"osmo",{"type":48,"value":86}," command, including ",{"type":43,"tag":79,"props":88,"children":90},{"className":89},[],[91],{"type":48,"value":92},"osmo --version",{"type":48,"value":94}," or query\ncommands. After confirmation, continue with the checks below.",{"type":43,"tag":73,"props":96,"children":97},{},[98,100,105],{"type":48,"value":99},"Confirm the CLI is available: ",{"type":43,"tag":79,"props":101,"children":103},{"className":102},[],[104],{"type":48,"value":92},{"type":48,"value":106},". If it fails, tell the user\nthe OSMO CLI is unavailable and stop.",{"type":43,"tag":73,"props":108,"children":109},{},[110,112,118],{"type":48,"value":111},"If any command returns an authentication error, ask the user to run\n",{"type":43,"tag":79,"props":113,"children":115},{"className":114},[],[116],{"type":48,"value":117},"osmo login",{"type":48,"value":119}," and stop until they confirm.",{"type":43,"tag":73,"props":121,"children":122},{},[123,125,131,133,139],{"type":48,"value":124},"Resource and workflow operations rely on the user's profile and pool access\n(",{"type":43,"tag":79,"props":126,"children":128},{"className":127},[],[129],{"type":48,"value":130},"osmo profile list",{"type":48,"value":132},", ",{"type":43,"tag":79,"props":134,"children":136},{"className":135},[],[137],{"type":48,"value":138},"osmo pool list",{"type":48,"value":140},").",{"type":43,"tag":56,"props":142,"children":144},{"id":143},"operating-rules",[145],{"type":48,"value":146},"Operating Rules",{"type":43,"tag":148,"props":149,"children":150},"ul",{},[151,156,161,173,185,190,219,224],{"type":43,"tag":73,"props":152,"children":153},{},[154],{"type":48,"value":155},"Classify the request using the Reference Routing section below, then load only\nthe reference file(s) it names before running commands.",{"type":43,"tag":73,"props":157,"children":158},{},[159],{"type":48,"value":160},"Do not guess command names or flags from memory. Use the linked reference for\nthe user's use case, then run the commands yourself.",{"type":43,"tag":73,"props":162,"children":163},{},[164,166,171],{"type":48,"value":165},"Obtain workflow and resource state (status, logs, events, capacity, spec) by\nrunning the ",{"type":43,"tag":79,"props":167,"children":169},{"className":168},[],[170],{"type":48,"value":84},{"type":48,"value":172}," CLI yourself. Do not infer OSMO state by reading, cat-ing,\nor grepping files in the workspace — run the command and use its output.",{"type":43,"tag":73,"props":174,"children":175},{},[176,178,183],{"type":48,"value":177},"Never fabricate command output; report only what the ",{"type":43,"tag":79,"props":179,"children":181},{"className":180},[],[182],{"type":48,"value":84},{"type":48,"value":184}," commands actually returned.",{"type":43,"tag":73,"props":186,"children":187},{},[188],{"type":48,"value":189},"Cache workflow query JSON during the conversation; do not re-query just to\nextract another field from the same response.",{"type":43,"tag":73,"props":191,"children":192},{},[193,195,201,203,209,211,217],{"type":48,"value":194},"Surface ",{"type":43,"tag":79,"props":196,"children":198},{"className":197},[],[199],{"type":48,"value":200},"grafana_url",{"type":48,"value":202}," and ",{"type":43,"tag":79,"props":204,"children":206},{"className":205},[],[207],{"type":48,"value":208},"dashboard_url",{"type":48,"value":210}," according to\n",{"type":43,"tag":79,"props":212,"children":214},{"className":213},[],[215],{"type":48,"value":216},"references\u002Fworkflow-status.md",{"type":48,"value":218},"; if either value is null, say it is\nunavailable instead of omitting it.",{"type":43,"tag":73,"props":220,"children":221},{},[222],{"type":48,"value":223},"Do not edit cluster config, node taints, quota policies, or non-OSMO\nKubernetes resources. Those are admin-side operations.",{"type":43,"tag":73,"props":225,"children":226},{},[227,229,235],{"type":48,"value":228},"Do not edit server-side OSMO configuration (",{"type":43,"tag":79,"props":230,"children":232},{"className":231},[],[233],{"type":48,"value":234},"osmo config",{"type":48,"value":236},"): pod or group\ntemplates, resource validations, pool\u002Fbackend config, roles, or storage\nbuckets. That is the OSMO admin surface — say it is out of scope and do not\nattempt it here.",{"type":43,"tag":56,"props":238,"children":240},{"id":239},"default-workflow",[241],{"type":48,"value":242},"Default Workflow",{"type":43,"tag":68,"props":244,"children":245},{},[246,251,256,268,287,292],{"type":43,"tag":73,"props":247,"children":248},{},[249],{"type":48,"value":250},"Complete the Prerequisites above (CLI check, auth, profile\u002Fpool access)\nbefore the first OSMO command.",{"type":43,"tag":73,"props":252,"children":253},{},[254],{"type":48,"value":255},"Classify the request and read only the reference file(s) named in the\nReference Routing section below.",{"type":43,"tag":73,"props":257,"children":258},{},[259,261,266],{"type":48,"value":260},"Run the ",{"type":43,"tag":79,"props":262,"children":264},{"className":263},[],[265],{"type":48,"value":84},{"type":48,"value":267}," commands yourself — cache the query JSON and never infer state\nfrom workspace files — unless the selected reference says to spawn a subagent.",{"type":43,"tag":73,"props":269,"children":270},{},[271,273,278,280,285],{"type":48,"value":272},"Verify the outputs before reporting: surface ",{"type":43,"tag":79,"props":274,"children":276},{"className":275},[],[277],{"type":48,"value":200},{"type":48,"value":279},"\u002F",{"type":43,"tag":79,"props":281,"children":283},{"className":282},[],[284],{"type":48,"value":208},{"type":48,"value":286},"\n(or say a null one is unavailable), confirm returned workflow IDs, and check\nstatus and exit codes.",{"type":43,"tag":73,"props":288,"children":289},{},[290],{"type":48,"value":291},"Summarize in the user's terms: available capacity, workflow state, progress,\nerror cause and concrete fix, dashboard link, output, or next action.",{"type":43,"tag":73,"props":293,"children":294},{},[295],{"type":48,"value":296},"For submit -> monitor -> fix loops, keep monitoring and final reporting in\nthe main conversation; delegate only setup, submission, or log summarization.",{"type":43,"tag":56,"props":298,"children":300},{"id":299},"reference-routing",[301],{"type":48,"value":302},"Reference Routing",{"type":43,"tag":50,"props":304,"children":305},{},[306],{"type":48,"value":307},"Classify the request and read only the reference file(s) for the matched intent.\nEach heading is a reference file; the bullets under it are the user intents and\nexample wordings that route there. Error and failure cases are listed under the\nreference that handles them. These are routing cues, not complete command recipes.",{"type":43,"tag":309,"props":310,"children":312},"h3",{"id":311},"referencesresource-check-formatmd",[313],{"type":43,"tag":79,"props":314,"children":316},{"className":315},[],[317],{"type":48,"value":318},"references\u002Fresource-check-format.md",{"type":43,"tag":50,"props":320,"children":321},{},[322],{"type":48,"value":323},"Resources, pools, GPUs, nodes, or quota.",{"type":43,"tag":148,"props":325,"children":326},{},[327,332,344],{"type":43,"tag":73,"props":328,"children":329},{},[330],{"type":48,"value":331},"\"What resources are available to me?\", \"Any H100s free?\", \"Do I have quota?\"",{"type":43,"tag":73,"props":333,"children":334},{},[335,337,343],{"type":48,"value":336},"Discover profile\u002Fpool access and report effective capacity (",{"type":43,"tag":79,"props":338,"children":340},{"className":339},[],[341],{"type":48,"value":342},"min(Quota Free, Total Free)",{"type":48,"value":140},{"type":43,"tag":73,"props":345,"children":346},{},[347],{"type":48,"value":348},"If pool\u002Fresource output is empty or ambiguous, re-check access here and state uncertainty instead of guessing.",{"type":43,"tag":309,"props":350,"children":352},{"id":351},"referencescli-commandsmd",[353],{"type":43,"tag":79,"props":354,"children":356},{"className":355},[],[357],{"type":48,"value":358},"references\u002Fcli-commands.md",{"type":43,"tag":50,"props":360,"children":361},{},[362],{"type":48,"value":363},"Safe end-user OSMO command lookup when no dedicated reference applies.",{"type":43,"tag":148,"props":365,"children":366},{},[367,372,377],{"type":43,"tag":73,"props":368,"children":369},{},[370],{"type":48,"value":371},"\"Check whether I can read from s3:\u002F\u002F...\", \"Download this s3:\u002F\u002F path\", \"What is the command for data download?\", \"How do I list direct storage?\", \"Set my default pool\" — use this for auth\u002Fversion\u002Fprofile\u002Fdata\u002Ftask syntax.",{"type":43,"tag":73,"props":373,"children":374},{},[375],{"type":48,"value":376},"For workflow, app, credential, resource, or troubleshooting details, follow this file's routing to the dedicated reference.",{"type":43,"tag":73,"props":378,"children":379},{},[380,382,387,388,394],{"type":48,"value":381},"If the requested command is ",{"type":43,"tag":79,"props":383,"children":385},{"className":384},[],[386],{"type":48,"value":234},{"type":48,"value":132},{"type":43,"tag":79,"props":389,"children":391},{"className":390},[],[392],{"type":48,"value":393},"osmo user",{"type":48,"value":395},", role\u002Fbucket\u002Fadmin mutation, or Kubernetes administration, treat it as out of scope.",{"type":43,"tag":309,"props":397,"children":399},{"id":398},"referencesworkflow-commandsmd",[400],{"type":43,"tag":79,"props":401,"children":403},{"className":402},[],[404],{"type":48,"value":405},"references\u002Fworkflow-commands.md",{"type":43,"tag":50,"props":407,"children":408},{},[409],{"type":48,"value":410},"Workflow submit\u002Flist\u002Fquery\u002Flog\u002Fevent\u002Fspec subcommand and flag lookup.",{"type":43,"tag":148,"props":412,"children":413},{},[414,419],{"type":43,"tag":73,"props":415,"children":416},{},[417],{"type":48,"value":418},"\"Validate this workflow\", \"Dry run it\", \"What flags does workflow list support?\", \"How do I fetch logs?\", \"Show the rendered spec\"",{"type":43,"tag":73,"props":420,"children":421},{},[422,424,430,431,436,438,444],{"type":48,"value":423},"Use for command syntax only; use ",{"type":43,"tag":79,"props":425,"children":427},{"className":426},[],[428],{"type":48,"value":429},"references\u002Fworkflow-submit.md",{"type":48,"value":132},{"type":43,"tag":79,"props":432,"children":434},{"className":433},[],[435],{"type":48,"value":216},{"type":48,"value":437},", or ",{"type":43,"tag":79,"props":439,"children":441},{"className":440},[],[442],{"type":48,"value":443},"references\u002Ftroubleshooting.md",{"type":48,"value":445}," for procedures.",{"type":43,"tag":309,"props":447,"children":449},{"id":448},"referencesworkflow-runtime-commandsmd",[450],{"type":43,"tag":79,"props":451,"children":453},{"className":452},[],[454],{"type":48,"value":455},"references\u002Fworkflow-runtime-commands.md",{"type":43,"tag":50,"props":457,"children":458},{},[459],{"type":48,"value":460},"Live workflow runtime operations.",{"type":43,"tag":148,"props":462,"children":463},{},[464,469],{"type":43,"tag":73,"props":465,"children":466},{},[467],{"type":48,"value":468},"\"Cancel workflow X\", \"Exec into task Y\", \"Port-forward task Z\", \"Rsync files into this workflow\", \"Add a tag to this workflow\"",{"type":43,"tag":73,"props":470,"children":471},{},[472],{"type":48,"value":473},"Ask for confirmation before cancellation and other destructive operations. A\nbare request like \"Cancel workflow X\" is not confirmation.",{"type":43,"tag":309,"props":475,"children":477},{"id":476},"referencesworkflow-submitmd",[478],{"type":43,"tag":79,"props":479,"children":481},{"className":480},[],[482],{"type":48,"value":429},{"type":43,"tag":50,"props":484,"children":485},{},[486],{"type":48,"value":487},"Submit a supplied or generated workflow.",{"type":43,"tag":148,"props":489,"children":490},{},[491,503],{"type":43,"tag":73,"props":492,"children":493},{},[494,496,501],{"type":48,"value":495},"\"Submit workflow.yaml\", \"Pick a free H100 pool\", \"No need to ask\" — read the supplied YAML as-is, choose a pool (use ",{"type":43,"tag":79,"props":497,"children":499},{"className":498},[],[500],{"type":48,"value":318},{"type":48,"value":502}," for pool selection), and submit only if authorized.",{"type":43,"tag":73,"props":504,"children":505},{},[506],{"type":48,"value":507},"\"Submit this Jinja workflow with 4 GPUs\" — preserve Jinja placeholders and pass values at submit time.",{"type":43,"tag":309,"props":509,"children":511},{"id":510},"referencescookbook-fetchingmd",[512],{"type":43,"tag":79,"props":513,"children":515},{"className":514},[],[516],{"type":48,"value":517},"references\u002Fcookbook-fetching.md",{"type":43,"tag":50,"props":519,"children":520},{},[521],{"type":48,"value":522},"Generate a workload from an OSMO cookbook example.",{"type":43,"tag":148,"props":524,"children":525},{},[526],{"type":43,"tag":73,"props":527,"children":528},{},[529,531,536],{"type":48,"value":530},"\"Generate 1000 Isaac Sim images and submit\" — fetch\u002Fadapt the cookbook workflow, compute the run count, then submit via ",{"type":43,"tag":79,"props":532,"children":534},{"className":533},[],[535],{"type":48,"value":429},{"type":48,"value":537},".",{"type":43,"tag":309,"props":539,"children":541},{"id":540},"referencesworkflow-statusmd",[542],{"type":43,"tag":79,"props":543,"children":545},{"className":544},[],[546],{"type":48,"value":216},{"type":43,"tag":50,"props":548,"children":549},{},[550],{"type":48,"value":551},"Status, logs, links, live metrics, recent workflows, and workflow explanation.",{"type":43,"tag":148,"props":553,"children":554},{},[555,560,565,577,589,594,605],{"type":43,"tag":73,"props":556,"children":557},{},[558],{"type":48,"value":559},"\"Show my recent workflows\", \"What's still running?\", \"What finished?\"",{"type":43,"tag":73,"props":561,"children":562},{},[563],{"type":48,"value":564},"\"Is workflow gr00t-train-1 done?\", \"How is my run going?\", \"Show progress\" — query, fetch logs, summarize state.",{"type":43,"tag":73,"props":566,"children":567},{},[568,570,575],{"type":48,"value":569},"\"How much memory\u002FGPU is it using?\", \"Open metrics\" — surface ",{"type":43,"tag":79,"props":571,"children":573},{"className":572},[],[574],{"type":48,"value":200},{"type":48,"value":576},"; do not invent live utilization.",{"type":43,"tag":73,"props":578,"children":579},{},[580,582,587],{"type":48,"value":581},"\"Give me the Kubernetes dashboard link\", \"I want to inspect the pod\" — surface ",{"type":43,"tag":79,"props":583,"children":585},{"className":584},[],[586],{"type":48,"value":208},{"type":48,"value":588}," or say it is unavailable.",{"type":43,"tag":73,"props":590,"children":591},{},[592],{"type":48,"value":593},"\"What does this workflow do?\", \"Explain this run before I rerun it\" — fetch the templated spec and summarize purpose\u002Fimage\u002Fcommand\u002Foutput.",{"type":43,"tag":73,"props":595,"children":596},{},[597,599,604],{"type":48,"value":598},"\"Why is it PENDING\u002Fqueued\u002Fstuck?\", \"Why won't it schedule?\" — establish state here, then ",{"type":43,"tag":79,"props":600,"children":602},{"className":601},[],[603],{"type":48,"value":443},{"type":48,"value":537},{"type":43,"tag":73,"props":606,"children":607},{},[608,610,615],{"type":48,"value":609},"Logs or events are sparse or time out — establish state here; for failures route to ",{"type":43,"tag":79,"props":611,"children":613},{"className":612},[],[614],{"type":48,"value":443},{"type":48,"value":537},{"type":43,"tag":309,"props":617,"children":619},{"id":618},"referencestroubleshootingmd",[620],{"type":43,"tag":79,"props":621,"children":623},{"className":622},[],[624],{"type":48,"value":443},{"type":43,"tag":50,"props":626,"children":627},{},[628],{"type":48,"value":629},"Failed, stuck, sparse-log, or misbehaving workflows.",{"type":43,"tag":148,"props":631,"children":632},{},[633,638],{"type":43,"tag":73,"props":634,"children":635},{},[636],{"type":48,"value":637},"\"The logs are empty\", \"Why did it fail?\", \"Exit code 137\u002F139\u002F143\u002F127\" — match the failure signature and propose a concrete fix.",{"type":43,"tag":73,"props":639,"children":640},{},[641,643,649],{"type":48,"value":642},"For private-image pull failures, establish the failure here, then read ",{"type":43,"tag":79,"props":644,"children":646},{"className":645},[],[647],{"type":48,"value":648},"references\u002Fworkflow-registry-credentials.md",{"type":48,"value":537},{"type":43,"tag":309,"props":651,"children":653},{"id":652},"referencesworkflow-credentialsmd",[654],{"type":43,"tag":79,"props":655,"children":657},{"className":656},[],[658],{"type":48,"value":659},"references\u002Fworkflow-credentials.md",{"type":43,"tag":50,"props":661,"children":662},{},[663,665,671,673,679,681,687],{"type":48,"value":664},"Workflow credential types and generic\u002Fdata credential setup: ",{"type":43,"tag":79,"props":666,"children":668},{"className":667},[],[669],{"type":48,"value":670},"REGISTRY",{"type":48,"value":672}," overview, ",{"type":43,"tag":79,"props":674,"children":676},{"className":675},[],[677],{"type":48,"value":678},"GENERIC",{"type":48,"value":680}," for task secrets, and ",{"type":43,"tag":79,"props":682,"children":684},{"className":683},[],[685],{"type":48,"value":686},"DATA",{"type":48,"value":688}," for object-storage access.",{"type":43,"tag":148,"props":690,"children":691},{},[692,697],{"type":43,"tag":73,"props":693,"children":694},{},[695],{"type":48,"value":696},"\"What credential type do I need?\", \"Set an API token\", \"Set a data credential\" — check existing credentials and create only the needed workflow credential.",{"type":43,"tag":73,"props":698,"children":699},{},[700],{"type":48,"value":701},"Never print secret values; prefer environment variables and ask for secrets only after automatic sources are exhausted.",{"type":43,"tag":309,"props":703,"children":705},{"id":704},"referencesworkflow-registry-credentialsmd",[706],{"type":43,"tag":79,"props":707,"children":709},{"className":708},[],[710],{"type":48,"value":648},{"type":43,"tag":50,"props":712,"children":713},{},[714],{"type":48,"value":715},"Private image-pull and registry credential setup.",{"type":43,"tag":148,"props":717,"children":718},{},[719,724],{"type":43,"tag":73,"props":720,"children":721},{},[722],{"type":48,"value":723},"\"Use this private nvcr.io image\", \"Set up NGC credentials\", \"Create an OSMO registry credential\", \"ImagePullBackOff unauthorized\", \"pull access denied\" — check existing registry credentials and create or fix only the needed OSMO registry credential.",{"type":43,"tag":73,"props":725,"children":726},{},[727],{"type":48,"value":728},"Registry image-pull credentials are managed by OSMO outside workflow YAML.",{"type":43,"tag":309,"props":730,"children":732},{"id":731},"referencesvalidation-error-recoverymd",[733],{"type":43,"tag":79,"props":734,"children":736},{"className":735},[],[737],{"type":48,"value":738},"references\u002Fvalidation-error-recovery.md",{"type":43,"tag":50,"props":740,"children":741},{},[742],{"type":48,"value":743},"Submission capacity validation errors.",{"type":43,"tag":148,"props":745,"children":746},{},[747],{"type":43,"tag":73,"props":748,"children":749},{},[750,756,758,764],{"type":43,"tag":79,"props":751,"children":753},{"className":752},[],[754],{"type":48,"value":755},"osmo workflow submit",{"type":48,"value":757}," returns a capacity validation error — edit only the allowed hard-coded ",{"type":43,"tag":79,"props":759,"children":761},{"className":760},[],[762],{"type":48,"value":763},"resources",{"type":48,"value":765}," values and resubmit.",{"type":43,"tag":309,"props":767,"children":769},{"id":768},"referencesworkflow-appsmd",[770],{"type":43,"tag":79,"props":771,"children":773},{"className":772},[],[774],{"type":48,"value":775},"references\u002Fworkflow-apps.md",{"type":43,"tag":50,"props":777,"children":778},{},[779],{"type":48,"value":780},"Workflow app lifecycle.",{"type":43,"tag":148,"props":782,"children":783},{},[784,789],{"type":43,"tag":73,"props":785,"children":786},{},[787],{"type":48,"value":788},"\"Create an app from this workflow\", \"Publish this completed run\", \"List apps\", \"Show app parameters\", \"Update app\", \"Submit app\", \"Rename\u002Fdelete app\" — use the app lifecycle reference.",{"type":43,"tag":73,"props":790,"children":791},{},[792],{"type":48,"value":793},"Ask for confirmation before create\u002Fupdate\u002Fsubmit\u002Frename\u002Fdelete unless the user already clearly authorized the action.",{"type":43,"tag":309,"props":795,"children":797},{"id":796},"referencesworkflow-specmd",[798],{"type":43,"tag":79,"props":799,"children":801},{"className":800},[],[802],{"type":48,"value":803},"references\u002Fworkflow-spec.md",{"type":43,"tag":50,"props":805,"children":806},{},[807],{"type":48,"value":808},"Workflow YAML schema and field shapes.",{"type":43,"tag":148,"props":810,"children":811},{},[812,817,837],{"type":43,"tag":73,"props":813,"children":814},{},[815],{"type":48,"value":816},"\"Is this workflow YAML valid?\", \"Add files\u002Fenvironment\u002Fresources\", \"What shape do tasks\u002Fresources\u002Fgroups use?\" — use this as a compact field map.",{"type":43,"tag":73,"props":818,"children":819},{},[820,822,828,830,836],{"type":48,"value":821},"For URL inputs\u002Foutputs, also read ",{"type":43,"tag":79,"props":823,"children":825},{"className":824},[],[826],{"type":48,"value":827},"references\u002Fworkflow-io-spec.md",{"type":48,"value":829},"; for topology placement, also read ",{"type":43,"tag":79,"props":831,"children":833},{"className":832},[],[834],{"type":48,"value":835},"references\u002Fworkflow-advanced-patterns.md",{"type":48,"value":537},{"type":43,"tag":73,"props":838,"children":839},{},[840,842,848,850,855],{"type":48,"value":841},"For design examples, continue to ",{"type":43,"tag":79,"props":843,"children":845},{"className":844},[],[846],{"type":48,"value":847},"references\u002Fworkflow-patterns.md",{"type":48,"value":849},"; for checkpointing\u002Ftopology\u002Fnode exclusion, continue to ",{"type":43,"tag":79,"props":851,"children":853},{"className":852},[],[854],{"type":48,"value":835},{"type":48,"value":537},{"type":43,"tag":309,"props":857,"children":859},{"id":858},"referencesworkflow-io-specmd",[860],{"type":43,"tag":79,"props":861,"children":863},{"className":862},[],[864],{"type":48,"value":827},{"type":43,"tag":50,"props":866,"children":867},{},[868],{"type":48,"value":869},"Workflow data flow, runtime tokens, and Jinja parameter fields.",{"type":43,"tag":148,"props":871,"children":872},{},[873,910],{"type":43,"tag":73,"props":874,"children":875},{},[876,878,884,886,892,894,900,902,908],{"type":48,"value":877},"\"Where does ",{"type":43,"tag":79,"props":879,"children":881},{"className":880},[],[882],{"type":48,"value":883},"default-values",{"type":48,"value":885}," go?\", \"What shape do inputs\u002Foutputs use?\", \"How do ",{"type":43,"tag":79,"props":887,"children":889},{"className":888},[],[890],{"type":48,"value":891},"{{output}}",{"type":48,"value":893}," or ",{"type":43,"tag":79,"props":895,"children":897},{"className":896},[],[898],{"type":48,"value":899},"{{input:0}}",{"type":48,"value":901}," work?\", \"How do I use ",{"type":43,"tag":79,"props":903,"children":905},{"className":904},[],[906],{"type":48,"value":907},"--set",{"type":48,"value":909},"?\" — use this as the compact IO\u002Ftemplate map.",{"type":43,"tag":73,"props":911,"children":912},{},[913,915,920],{"type":48,"value":914},"For full workflow design examples, continue to ",{"type":43,"tag":79,"props":916,"children":918},{"className":917},[],[919],{"type":48,"value":847},{"type":48,"value":537},{"type":43,"tag":309,"props":922,"children":924},{"id":923},"referencesworkflow-patternsmd",[925],{"type":43,"tag":79,"props":926,"children":928},{"className":927},[],[929],{"type":48,"value":847},{"type":43,"tag":50,"props":931,"children":932},{},[933],{"type":48,"value":934},"Workflow structure and authoring patterns.",{"type":43,"tag":148,"props":936,"children":937},{},[938],{"type":43,"tag":73,"props":939,"children":940},{},[941,943,948],{"type":48,"value":942},"Workflow structure, multi-task execution, dependencies, Jinja templates, checkpointing, exit\u002Fretry behavior, node exclusion, or topology placement — start here; it routes to ",{"type":43,"tag":79,"props":944,"children":946},{"className":945},[],[947],{"type":48,"value":835},{"type":48,"value":949}," for niche patterns.",{"type":43,"tag":309,"props":951,"children":953},{"id":952},"referencesworkflow-expertmd-subagent-prompt",[954,960],{"type":43,"tag":79,"props":955,"children":957},{"className":956},[],[958],{"type":48,"value":959},"references\u002Fworkflow-expert.md",{"type":48,"value":961}," (subagent prompt)",{"type":43,"tag":50,"props":963,"children":964},{},[965],{"type":48,"value":966},"End-to-end submit \u002F monitor \u002F fix loops.",{"type":43,"tag":148,"props":968,"children":969},{},[970],{"type":43,"tag":73,"props":971,"children":972},{},[973],{"type":48,"value":974},"\"Monitor this from submit to completion\", \"Fix and resubmit if it fails\" — spawn or resume this subagent for setup\u002Fsubmission; keep final monitoring and reporting in the main conversation. After a failed loop, fetch logs as directed, resume the subagent, and stop after three failures.",{"type":43,"tag":309,"props":976,"children":978},{"id":977},"referenceslogs-readermd-subagent-prompt",[979,985],{"type":43,"tag":79,"props":980,"children":982},{"className":981},[],[983],{"type":48,"value":984},"references\u002Flogs-reader.md",{"type":48,"value":961},{"type":43,"tag":50,"props":987,"children":988},{},[989],{"type":48,"value":990},"Multi-task log summarization.",{"type":43,"tag":148,"props":992,"children":993},{},[994],{"type":43,"tag":73,"props":995,"children":996},{},[997,999,1005],{"type":48,"value":998},"A workflow has two or more tasks and logs are needed — spawn ",{"type":43,"tag":79,"props":1000,"children":1002},{"className":1001},[],[1003],{"type":48,"value":1004},"logs-reader",{"type":48,"value":1006}," subagents as directed; do not inline all logs.",{"type":43,"tag":309,"props":1008,"children":1010},{"id":1009},"when-not-to-use-this-skill",[1011],{"type":48,"value":1012},"When NOT to use this skill",{"type":43,"tag":148,"props":1014,"children":1015},{},[1016],{"type":43,"tag":73,"props":1017,"children":1018},{},[1019,1021,1026,1028,1033],{"type":48,"value":1020},"\"What GPUs does NVIDIA sell?\", \"How do I deploy OSMO?\", \"Set up AKS \u002F Kubernetes \u002F NIM Operator infrastructure\", \"Configure Kubernetes taints\", \"Edit a pod template \u002F pool quota \u002F ",{"type":43,"tag":79,"props":1022,"children":1024},{"className":1023},[],[1025],{"type":48,"value":234},{"type":48,"value":1027},"\" — answer with general help or another skill; do not run ",{"type":43,"tag":79,"props":1029,"children":1031},{"className":1030},[],[1032],{"type":48,"value":84},{"type":48,"value":1034},". Server-side config, cluster admin, and infrastructure scaffolding are outside this skill.",{"items":1036,"total":1077},[1037,1051,1069],{"slug":1038,"name":1038,"fn":1039,"description":1040,"org":1041,"tags":1042,"stars":26,"repoUrl":27,"updatedAt":1050},"osmo-admin","manage OSMO service configuration","Use only for offline\u002Flocal OSMO service-config admin requests involving explicit config roots or values files, or to ask for one when a file-specific config request omits it. Do not inspect the workspace to infer a root. Do not use for live workflow support, resource capacity, pod\u002Fnode diagnostics, or cluster operations, except live service-config paths that must be refused.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1043,1046,1047],{"name":1044,"slug":1045,"type":15},"Configuration","configuration",{"name":9,"slug":8,"type":15},{"name":1048,"slug":1049,"type":15},"Operations","operations","2026-07-14T05:33:26.717488",{"slug":1052,"name":1052,"fn":1053,"description":1054,"org":1055,"tags":1056,"stars":26,"repoUrl":27,"updatedAt":1068},"osmo-deploy","deploy OSMO to Kubernetes clusters","How to deploy OSMO to a Kubernetes cluster on Azure (AKS), AWS (EKS), MicroK8s (single-node), or any kubectl-reachable cluster (BYO). Use this skill whenever the user asks to install, deploy, set up, or stand up OSMO; whenever they ask to provision an OSMO cluster; whenever they mention deploy-osmo-minimal.sh, deploy-k8s.sh, or \"OSMO helm install\"; whenever they ask to wire up workflow storage (MinIO \u002F Azure Blob \u002F S3); or whenever they ask to add a GPU pool to an OSMO cluster, install KAI scheduler, install the NVIDIA GPU Operator, or run the post-install smoke tests. Targets OSMO 6.3 (ConfigMap mode).\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1057,1060,1063,1066,1067],{"name":1058,"slug":1059,"type":15},"AWS","aws",{"name":1061,"slug":1062,"type":15},"Azure","azure",{"name":1064,"slug":1065,"type":15},"Deployment","deployment",{"name":24,"slug":25,"type":15},{"name":9,"slug":8,"type":15},"2026-07-14T05:33:24.181591",{"slug":4,"name":4,"fn":5,"description":6,"org":1070,"tags":1071,"stars":26,"repoUrl":27,"updatedAt":28},{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1072,1073,1074,1075,1076],{"name":17,"slug":18,"type":15},{"name":21,"slug":22,"type":15},{"name":24,"slug":25,"type":15},{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},3,{"items":1079,"total":1236},[1080,1098,1114,1125,1137,1151,1164,1178,1191,1202,1216,1225],{"slug":1081,"name":1081,"fn":1082,"description":1083,"org":1084,"tags":1085,"stars":1095,"repoUrl":1096,"updatedAt":1097},"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},[1086,1089,1092],{"name":1087,"slug":1088,"type":15},"Documentation","documentation",{"name":1090,"slug":1091,"type":15},"MCP","mcp",{"name":1093,"slug":1094,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":1099,"name":1099,"fn":1100,"description":1101,"org":1102,"tags":1103,"stars":1111,"repoUrl":1112,"updatedAt":1113},"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},[1104,1107,1108],{"name":1105,"slug":1106,"type":15},"Containers","containers",{"name":1064,"slug":1065,"type":15},{"name":1109,"slug":1110,"type":15},"Python","python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":1115,"name":1115,"fn":1116,"description":1117,"org":1118,"tags":1119,"stars":1111,"repoUrl":1112,"updatedAt":1124},"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},[1120,1123],{"name":1121,"slug":1122,"type":15},"CI\u002FCD","ci-cd",{"name":1064,"slug":1065,"type":15},"2026-07-14T05:25:59.97109",{"slug":1126,"name":1126,"fn":1127,"description":1128,"org":1129,"tags":1130,"stars":1111,"repoUrl":1112,"updatedAt":1136},"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},[1131,1132,1133],{"name":1121,"slug":1122,"type":15},{"name":1064,"slug":1065,"type":15},{"name":1134,"slug":1135,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":1138,"name":1138,"fn":1139,"description":1140,"org":1141,"tags":1142,"stars":1111,"repoUrl":1112,"updatedAt":1150},"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},[1143,1146,1147],{"name":1144,"slug":1145,"type":15},"Debugging","debugging",{"name":1134,"slug":1135,"type":15},{"name":1148,"slug":1149,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":1152,"name":1152,"fn":1153,"description":1154,"org":1155,"tags":1156,"stars":1111,"repoUrl":1112,"updatedAt":1163},"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},[1157,1160],{"name":1158,"slug":1159,"type":15},"Best Practices","best-practices",{"name":1161,"slug":1162,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":1165,"name":1165,"fn":1166,"description":1167,"org":1168,"tags":1169,"stars":1111,"repoUrl":1112,"updatedAt":1177},"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},[1170,1173,1176],{"name":1171,"slug":1172,"type":15},"Machine Learning","machine-learning",{"name":1174,"slug":1175,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":1179,"name":1179,"fn":1180,"description":1181,"org":1182,"tags":1183,"stars":1111,"repoUrl":1112,"updatedAt":1190},"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},[1184,1187],{"name":1185,"slug":1186,"type":15},"QA","qa",{"name":1188,"slug":1189,"type":15},"Testing","testing","2026-07-14T05:25:53.673039",{"slug":1192,"name":1192,"fn":1193,"description":1194,"org":1195,"tags":1196,"stars":1111,"repoUrl":1112,"updatedAt":1201},"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},[1197,1198],{"name":1064,"slug":1065,"type":15},{"name":1199,"slug":1200,"type":15},"Infrastructure","infrastructure","2026-07-14T05:25:49.362534",{"slug":1203,"name":1203,"fn":1204,"description":1205,"org":1206,"tags":1207,"stars":1111,"repoUrl":1112,"updatedAt":1215},"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},[1208,1211,1212],{"name":1209,"slug":1210,"type":15},"Code Review","code-review",{"name":1134,"slug":1135,"type":15},{"name":1213,"slug":1214,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":1217,"name":1217,"fn":1218,"description":1219,"org":1220,"tags":1221,"stars":1111,"repoUrl":1112,"updatedAt":1224},"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},[1222,1223],{"name":1185,"slug":1186,"type":15},{"name":1188,"slug":1189,"type":15},"2026-07-14T05:25:54.928983",{"slug":1226,"name":1226,"fn":1227,"description":1228,"org":1229,"tags":1230,"stars":1111,"repoUrl":1112,"updatedAt":1235},"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},[1231,1234],{"name":1232,"slug":1233,"type":15},"Automation","automation",{"name":1121,"slug":1122,"type":15},"2026-07-30T05:29:03.275638",496]