Opensource review often ends with a reviewer requesting you to test the changes made in your PR locally first if not already done. This talk walks through actually doing that for a Backstage plugin that correlates Kubernetes CRDs to catalog components by git URL identity instead of fragile labels, so a single profile tracks correctly across dev, staging, and prod clusters.
The CRD itself, RuntimeConditionsProfile, isn't hand applied YAML it's delivered by a Kratix Promise, meaning the platform team defines it once and any cluster can request an instance on demand. That means Backstage doesn't care which cluster or which Promise pipeline produced the resource, only what codebase it describes.
We'll cover the work that goes through when a reviewer asks you to test it locally, no runnable app existed to test the plugin against, RBAC for the Promise's CRD was silently missing and a toolchain mismatch broke native builds across Node versions all fixed by standing up a real kind cluster, the Kratix Promise, and a Backstage instance from scratch.
Attendees leave with a pattern for identity-based CRD-to-entity correlation, how Kratix Promises fit into that picture and how to verify the work done for yourself