infrastructure
One node. Not a cluster.
No cluster by default. No mandatory Kubernetes. No needless complexity. One machine is enough to manage your containers, end to end.
A single VM can be enough
Run the whole lifecycle on one machine — with a simple backup and recovery strategy on top.
- Development
- Validation
- Production
- Observability
your machine — one node
web
space :5280
api
space :4100
worker
space job
runc · rootless · OCI — no daemon, no kubelet
The cluster way, and the SKLP way
the cluster way
sklp
cluster + control plane
one node is enough
daemon running as root
rootless, no daemon
kubelet, etcd, CNI, CSI
runc + the Linux kernel
YAML per resource
one space, dev to prod
registry + Dockerfile to maintain
OCI builds, no Dockerfile
Pod abstraction
space — lighter, faster
Security — machines execute, SKPI observes
Spaces run locally. Data flows out to SKPI — never the other way around.
No inbound access
By default, nothing reaches in. No remote execution, no port exposed on your laptop.
One-way by design
Your machines execute. SKPI observes. The data path only goes outward.
Simple backup & recovery
A straightforward strategy you can reason about — no operational maze.
A dev environment, a CI and a production shouldn’t be three different systems. They should be the same Space.
One Space everywhere.