KernelPulse
Link to open source: https://github.com/Aneeshpoetries/Sketch_n_ship
KernelPulse reimagines Linux server monitoring for people who don't want to run a monitoring stack just to monitor a server. Its purpose is to strip away the operational weight of Prometheus, exporters, Grafana, and a separate database, and replace all of it with one Python process that reads the kernel directly and ships with real intelligence built in — not bolted on.
The goal isn't just visibility, it's judgment. Most lightweight dashboards show you numbers and leave the interpretation to you; most tools that do interpret the numbers require standing up real infrastructure to get there. KernelPulse closes that gap with a transparent AI layer — a z-score anomaly classifier, a slow memory-leak detector that catches growth patterns z-score alone would miss, and a capacity-exhaustion forecaster that tells you "RAM full in ~2h 14m" instead of just "RAM: 87%." Every one of those calls is explainable: the dashboard shows the raw value, the moving average, and the exact statistical threshold behind each status, so nothing is a black box.
Ultimately, KernelPulse is built for the developer, student, or small team running a personal server or home lab who wants Prometheus-grade insight without Prometheus-grade overhead — a monitoring tool that installs in one command and actually tells you when something needs your attention, not just what the numbers are.
This build was uploaded as a hackathon project
