This guide runs Vibgrate drift scanning inside an Argo Workflows template on Kubernetes. It targets platform teams running CI as Kubernetes-native workflows.
Prerequisites
- Argo Workflows installed on a cluster.
- A way to source your repository into the step, such as a git artifact input or an init container.
Workflow template
apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
generateName: vibgrate-drift-
spec:
entrypoint: drift
templates:
- name: drift
container:
image: node:22
command: [sh, -c]
args: ['npx @vibgrate/cli scan --fail-on error']
Ensure your repository is mounted as the working directory so npx @vibgrate/cli scan scans the project root.
Gating the workflow
A non-zero exit code marks the node, and therefore the workflow, as failed. Use --fail-on error for severity gating, or --drift-budget to gate on the overall DriftScore:
npx @vibgrate/cli scan --drift-budget 60
Capturing the report
Write the report to a path declared as an output artifact so Argo archives it:
npx @vibgrate/cli scan --format markdown --out /tmp/drift-report.md
Caching
Mount a persistent volume claim for the Vibgrate cache directory so repeat scans reuse prior work. See Cache Vibgrate scans in CI.
Related
- Monorepo matrix scans map naturally to Argo workflow fan-out.
- Push scan results to Vibgrate Cloud from CI for visibility.