I kept SSHing into servers and running the same commands just to see what was running:
kubectl get pods
docker ps
systemctl status
top
Different tools, different boxes, no single picture. So I built InfraCanvas.
You install one agent on a Linux server. It draws everything — Kubernetes, Docker, systemd, processes — as one live map. Color-coded by health, updates in real time.
Click anything to see its details. And you can act right there: restart, scale, rollback, open a terminal, tail logs. No SSH.
A few things I cared about building it:
Try it (read-only demo, no signup): [https://demo.infracanvas.app](https://demo.infracanvas.app)
Free tier on your own box: [https://cloud.infracanvas.app](https://cloud.infracanvas.app)
I'm a solo dev. Honest question: does seeing infra as a map actually help, or is kubectl muscle memory good enough? Tell me where it breaks.
source & further reading
dev.to — original article
n8n's Guardrails node masks your PII. It doesn't give it back. Here's why that matters.
When the model is the marketing device: A Protobuf short story
The Hidden Architecture of the Agentic Enterprise: Model Stack, Tokenomics, and Harness