panic_mode

Built for the gap between 'I have one server' and 'I run Datadog/Prometheus'. For a small fleet, full observability stacks are overkill and underused — but you still need to know when a disk is at 95% or a process is leaking memory before it cascades into a real outage.

panic_mode runs as a single Rust binary with minimal dependencies. It maintains continuous logs of system events, watches a configurable set of metrics and conditions, and pushes notifications to a developer via the configured channel when thresholds are crossed.

The design goal is 'noisy enough to be useful, quiet enough to trust' — a monitor that pages too often gets muted, and a muted monitor is no monitor at all.