Skip to content
StackBlazeDocs

Philosophy

StackBlaze is a managed PaaS. You compose projects, environments, and services in a dashboard. The control plane runs Kubernetes underneath; you never kubectl, edit CRDs, or pick a node pool.

That is a deliberate product bet:

  • One canvas per environment. Git services, Docker images, databases, cron, functions, and WAF sit on the same board and share a private network.
  • Staged changes. Topology edits accumulate, then you deploy. Half-applied graphs are the failure mode we optimize against.
  • Config as code is optional. Most teams click. stackblaze.yaml is there when you want plan/apply and git review. Apply never deletes.
  • Your token, your RBAC. Dashboard, CLI, API, and MCP all use a PAT (kbr_pat_). A privileged platform token is only a fallback.

Compare that to operating a cluster: you would own upgrades, ingress, certificates, storage classes, and RBAC. Here those are settings on a service.

See The basics for the object model and Platform for how a deploy actually runs.