Kula shrinks Linux monitoring to single binary

This is the kind of infra tool people actually keep running: small, opinionated, and easy to drop onto a box without building a whole observability stack around it.

• –The no-agent, no-database design makes it a strong fit for VPSes, homelabs, and closed networks where Prometheus/Grafana feels too heavy
• –Tiered ring-buffer storage gives it predictable disk use and local history without requiring extra cleanup jobs or retention plumbing
• –Having both a web UI and terminal UI is a practical win for ops workflows, especially when SSH is the primary interface
• –It overlaps with host monitoring stacks more than it competes with full observability platforms, so the sweet spot is single-server or small-fleet visibility
• –The AGPLv3 open-source model may matter for teams that want to inspect, modify, and self-host the whole stack