NVIDIA Cloud Functions is a platform for deploying, managing, and running GPU-based workloads such as inference, streaming, and batch processing, while routing requests to GPU worker clusters. Its monorepo contains the service code, deployment materials, documentation, examples, command-line tools, agent skills, and validation tools used to build and operate the platform.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add NVIDIA/nvcf --skill nvcf-self-managed-prerequisitegit clone --depth 1 https://github.com/NVIDIA/nvcfWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/nvidia/nvcf/nvcf-self-managed-prerequisite)<a href="https://agentmods.dev/skills/nvidia/nvcf/nvcf-self-managed-prerequisite"><img src="https://agentmods.dev/badge/skills/nvidia/nvcf/nvcf-self-managed-prerequisite/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/nvidia/nvcf/nvcf-self-managed-prerequisite"><img src="https://agentmods.dev/badge/skills/nvidia/nvcf/nvcf-self-managed-prerequisite.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00195 | $0.02444 |
| Opus 5.5 | $0.00078 | $0.00978 |
| Sonnet 5.5 | $0.00039 | $0.00489 |
| Haiku 4.5 | $0.00019 | $0.00244 |
Grade A, and why
nvcf-self-managed-prerequisite scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 178 lines — stays where its author put it; the contents beside it link to each section on GitHub.
NVCA prerequisites — KAI Scheduler + SMB CSI
One operator tool plus two cluster-level components the NVCA operator / compute plane depends on. Satisfy all three before running nvcf-nvca-install.
| Prereq | Why it is needed | Detail |
|---|---|---|
nvcf-cli |
The compute-plane stack's make register-cluster (and install/apply/sync, which abort without the registration values it writes) shells out to nvcf-cli. The shipped stack defaults to building it from a sibling ../cli checkout that the release does not include, so a green-field repo fails with ensure-nvcf-cli / "Registration values not found". |
See Step 0b below |
| KAI Scheduler | selfManaged.featureGateValues includes KAIScheduler; NVCA polls Queue CRs and refuses to become healthy until their quotas are -1. Skip the standalone helm step below when addons.kaiScheduler.enabled is set in nvcf-compute-plane (required when enabling Grove or Dynamo). |
references/kai-scheduler.md |
SMB CSI driver (smb.csi.k8s.io) |
NVCA's selfManaged.sharedStorage runs Samba sidecar pods that export file shares; the resulting PVCs need this CSI driver to bind |
references/smb-csi.md |
The KAI Scheduler and SMB CSI installs are cloud-neutral helm commands pinned to NVCF-validated versions. These are upstream third-party charts (not NVCF images), so they are not in manifest.yaml; the per-component reference docs carry the current pin and link the NVCF docs version table. nvcf-cli is an operator workstation tool, not an in-cluster install.
Prerequisites
- A running Kubernetes cluster (any cloud — AKS, EKS, GKE, k3d, MicroK8s) with
kubectlconfigured and admin access. helm>= 3.12 and < 4. Helm 4 is NOT supported (matchesnvcf-self-managed-stack/README.md). On Helm 4 the KAI install below hangs silently for many minutes — Helm 4 runs the chart's pre-installcrd-managerhook through abefore-hook-creationdelete and then waits--timeoutper already-absent hook resource, so the release sits inpending-installwith no pods and never errors cleanly. Use Helm 3.x.- Cluster has CPU headroom on a general-purpose node pool for KAI's 7 pods.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 178 lines · 195 tokens per session scan A b06737d73ab3
nvcf-self-managed-prerequisite is a skill published in the GitHub repository NVIDIA/nvcf (244 stars, last pushed today), licensed Apache-2.0. It adds 195 tokens to every session and 2,444 once invoked, about $0.0008 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
olares-chart
Olares app packaging and chart authoring via olares-cli chart — port a repo, docker-compose, or generic Helm chart; build/push the image; author, lint, package, and deploy an OlaresManifest; wire storage, middleware, entrances, env, and GPU; edit the chart after diagnosis. Runtime failure diagnosis is olares-doctor…
portal-relay
Set up and run a public Portal relay on any Linux host with a public IP — Docker Compose deployment, embedded authoritative DNS with one-time NS delegation, optional TCP/UDP lease ports for game hosting, and registration in the public relay pool. Use when the user asks to run their own relay, contribute a relay to the…
runpod-usage
Explain Runpod architecture and design choices: Pods versus Serverless, GPU/VRAM, image construction, storage, networking, and workload development. For creating or improving a complete template project use runpod-build-template; execute infrastructure with runpodctl/runpod-mcp and code-first work with flash.
runpod-build-template
Create or improve a complete Runpod template project from an application request, local code, or GitHub repository, with a documented Dockerfile and app-specific files for Pods or queue/load-balanced Serverless. Use runpod-templates for finding or operating existing official templates, runpod-mcp/runpodctl to register…
kubernetes-mesh-provisioner
Kubernetes Mesh Provisioner atomic skill. Stands up an RKE2 cluster (server + agents) with Cilium CNI and the NVIDIA GPU device plugin, the Kubernetes parallel of swarm-mesh-provisioner. Idempotent — re-runnable.
openshell-cli
Guide agents through using the OpenShell CLI (openshell) for sandbox management, gateway registration, provider configuration and refresh, profile management, policy iteration, settings, service exposure, BYOC workflows, and attached-provider inference. Covers basic through advanced multi-step workflows. Trigger…