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
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/NVIDIA/nvcfnpx agentmods add skills/nvidia/nvcf/nvca-values-customizationWrote 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/nvca-values-customization)<a href="https://agentmods.dev/skills/nvidia/nvcf/nvca-values-customization"><img src="https://agentmods.dev/badge/skills/nvidia/nvcf/nvca-values-customization/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/nvca-values-customization"><img src="https://agentmods.dev/badge/skills/nvidia/nvcf/nvca-values-customization.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Privilege Escalation · line 72 Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.Fix: Remove references to credential paths. Use environment variables or secrets managers. For docs, use placeholder paths (e.g., /path/to/config). Never load .env or token files in production code paths.
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 | $0.00051 | $0.01063 |
| Opus 5.5 | $0.00020 | $0.00425 |
| Sonnet 5.5 | $0.00010 | $0.00213 |
| Haiku 4.5 | $0.00005 | $0.00106 |
Grade A, and why
nvca-values-customization 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 8d 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 — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Customizing NVCA Operator Chart Values
Use this skill from deploy/helm/nvca-operator.
Values Flow
Two install paths, and only one of them renders values from a stack.
nvca-operator/values.yaml the chart's own defaults
-> make install values=<path> the values file, used directly
stack environment
-> scripts/render_values_from_stack_env.sh stack-aware generated values
-> make install-from-stack the generated values
Either accepts additional_values=<path> for further overrides.
Permanent Defaults
Edit nvca-operator/values.yaml directly. There is one chart and no vendoring
step, so that file is the source of truth.
Only defaults that suit every consumer belong there. Values tied to one deployment are supplied by whoever installs the chart:
- the compute-plane stack sets them under
deploy/stacks/nvcf-compute-plane/, includingnameOverride,fullnameOverrideandselfManaged.nvcaVersion - an ngc-managed install passes
ngcConfig.serviceKeyand thehelmManaged.*values on the command line
image.tag ships empty so templates fall back to appVersion, which the
release stamps at packaging time.
Deploy-time Overrides
Use additional_values for one-off validation or environment-specific values:
make install-from-stack \
stack_repo=../../../deploy/stacks/self-managed \
stack_env=local \
additional_values=override.yml
Use deploy-time overrides for secrets, credentials, cluster-specific IDs, and temporary validation changes.
Validation
make lint
make template
make validate
tools/ci/validate-helm-chart deploy/helm/nvca-operator/nvca-operator \
-f tools/ci/helm-validate-values/nvca-operator.yaml
Gotchas
- Install-time values are layered after generated stack-aware values.
- Use
clusterValidator.role: control-planeonly in a single-cluster topology, where the NVCF control plane runs in the same cluster. On a compute-only cluster every run is Not-Ready. - With
clusterValidator.role: control-plane, setclusterValidator.gatewayNamesto every stack Gateway asnamespace/name. Unset, the validator finds Gateways from route labels, and those can never fail Tier-1.make render-values-from-stackfills it in. - List stack quorum components that run outside the cluster (
nats,openbao,cassandra) inclusterValidator.externalComponents, or Tier-2 reports them missing after install.make render-values-from-stacklists each one the stack environment disables. make render-values-from-stackrewritesgatewayNames,storageClassandexternalComponentsfrom the stack on every render, empty included. Set a different value withadditional_values, not on the release.- The node-to-node probe pods have no image pull secrets. On a registry other
than NGC (
nvcr.ioor a subdomain),make render-values-from-stackpoints an unsetclusterValidator.nodeToNodeProbeImageatbusybox:1.36under the stack's repository, unlessnetworkChecks.enforcement.testImageis set. The default an earlier render wrote follows a registry change; any other probe image already set is kept. PassNODE_TO_NODE_PROBE_IMAGEto choose another image. - Use
yqcarefully for nested keys and quoted strings. Chart.yamlname stays in git and must match the subproject's service_name; the release refuses to publish when they differ. Only the version is set at packaging time, andappVersionis stamped from the nvca release the chart installs.- Never commit real service keys or rendered secret material.
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.
- 8d ago Changed · +21 lines 796c80177fee
- 15d ago Changed · -13 lines 679660e39d04
- 1mo ago First seen · 102 lines · 51 tokens per session scan A d8eb95fc398f
nvca-values-customization is a skill published in the GitHub repository NVIDIA/nvcf (238 stars, last pushed today), licensed Apache-2.0. It adds 51 tokens to every session and 1,063 once invoked, about $0.0002 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
helm
Package Kubernetes applications as Helm charts, configure values and dependencies, and inspect release upgrades or rollbacks.
helm-expert
Expert-level Helm 3 package management, chart development, templating, and production operations. Use when the user mentions Kubernetes, package manager, charts, templates, or deployment, or when the task involves Helm Architecture, Chart Structure, Values.yaml, or Helm Hooks.
tao-setup-nvidia-gpu-host
Host setup for TAO GPU backends. Checks and, after user approval, installs minimum-compatible NVIDIA driver, CUDA Toolkit, and NVIDIA Container Toolkit versions for Docker/local-Docker and Kubernetes GPU worker hosts. TAO-wide defaults can be overridden by the selected model's runtime profile. The --check-only path…
obol-stack
Monitor the Kubernetes cluster running the Obol Stack. Use when asked about pod status, logs, events, services, deployments, statefulsets, PVCs, or diagnosing issues — in your own namespace or across the cluster (local Ethereum nodes, DV pods, monitoring). Read-only; Secrets are never readable.
argocd
Operate Argo CD GitOps with AppProjects, safe automated sync/prune, and RespectIgnoreDifferences—debug OutOfSync/Degraded without kubectl drift that fights the controller.
helm
A guide to Helm, a package manager for Kubernetes applications. Helm uses reusable packages called Charts to install, configure, upgrade, inspect, and remove groups of Kubernetes resources.