This project is archived and no longer maintained. These docs stay available as a reference. The code stays available under its original license.

What I build now:

Kubmin — per-workload cost, efficiency, and SCI/SEI intelligence for Kubernetes
infai — a terminal control plane for local LLM inference: one profile shape for llama.cpp and vLLM, many models at once, live tokens/sec plus CPU/RAM/VRAM telemetry, hardware fit checks before launch
infaiw — a daemon-based agent harness: sessions keep running after you close the terminal, sidecar loops must pass an acceptance check before reporting back, every event lands on a durable branchable timeline, and approvals wait for you

Want to take over maintenance of kli? If you still depend on it and would like to become the maintainer, reach out on Discord or open an issue.

Core functionalities

How does the core functionalities of ksctl work

Basic cluster operations

HA self-managed cluster (VM is provisioned and ssh into and configure them just like ansible)
Managed (cloud provider creates the clusters and we get the kubeconfig in return)

Only for ha cluster as the user has manual ability to increase the number of worknodes
Example: if workernode 1 then it will create 2 then 3…

Only for ha cluster as the user has manual ability to decrease the number of worknodes
Example: if workernodes are 1, 2 then it will delete from the last to first aka 2 then 1

It will just use the request from the user to get the kubeconfig from specific cluster and save to specific folder that is ~/.ksctl/kubeconfig

Both ha and manage cluster it will search folders in specific directory to get what all cluster have been created for specific provider

Example: Here for get request of azure it will scan the directory .ksctl/state/azure/ha and for managed as well to get all the folder names

Was this page helpful?

Sorry to hear that. Please tell us how we can improve.