nvcrectl Overview | NVIDIA Cluster Readiness Engine

nvcrectl is the command-line interface for the NVIDIA Cluster Readiness Engine. It handles installation, certification lifecycle, workload execution, and reporting — without requiring direct kubectl access for most operations.

Install

curl -fsSL https://github.com/NVIDIA/cluster-readiness-engine/releases/latest/download/installer | bash
nvcrectl --version

To pin a version: curl -fsSL https://github.com/NVIDIA/cluster-readiness-engine/releases/download/<tag>/installer | bash -s -- -v <tag>.

The installer also creates a kubectl-nvcre symlink so the CLI is available as a kubectl plugin (kubectl nvcre ...).

Command groups

Command group	Purpose
`nvcrectl setup`	Install and uninstall the controller and its dependencies
`nvcrectl certification`	Run, render, report, and list-categories for Certification resources
`nvcrectl workloadrun`	Run, render, report, status, and cancel WorkloadRun resources
`nvcrectl cluster`	Inspect GPU nodes, platform, and network topology
`nvcrectl workflow`	Render Workflow manifests offline with overrides applied
`nvcrectl mcp`	Serve read-only certification state to MCP agents over stdio

Global flags

These flags are available on subcommands that connect to a cluster:

Flag	Description
`--kubeconfig`	Path to kubeconfig (defaults to `~/.kube/config`)
`--context`	Kubernetes context to use
`--namespace` / `-n`	Namespace override (not available on all subcommands)

Shell completion

nvcrectl completion bash >> ~/.bashrc
nvcrectl completion zsh >> ~/.zshrc