The Problem
Standard monitors show what is happening, but not if a spike is normal or a sign of a leak/runaway thread. A process jumping from 1% to 90% CPU isn't flagged by standard tools if the system isn't crashing.
Project Deep Dive
Aura Process Guardian 🛡️
Unsupervised OS-Level Anomaly Detector
PythonpsutilmatplotlibStatistics (Z-Score)
The Core Idea
Performs unsupervised anomaly detection locally by:
- Learning a baseline of 'normal' behavior
- Comparing live usage
- Flagging sustained statistical deviations using Z-scores.
No ML frameworks, fully explainable.
Key Features
- Real-time Monitoring
- Process Protection ('Guardian Mode')
- Auto-Healing
- Visual Dashboard
- Logging & Alerts
Future Improvements
- Command-line flags
- Adaptive/rolling baselines
- System tray integration
- Container-aware monitoring
- Prometheus/Grafana exporter
How to Run
pip install psutil matplotlib python run_guardian.py
Live Demo Simulation
> SYSTEM NOTICE: This interface is currently a frontend simulation of the local OS monitoring tool. It demonstrates the expected behavior during baseline learning and anomaly detection.
dev@repo:~/aura-process-guardian
$