Discover Guider: A Python‑Powered Linux Observability Suite with 150+ Commands

Guider is a Python‑written, command‑line observability suite for Linux, developed over nine years by Peace Lee, an engineer at Hyundai Motor Company.

It provides more than 150 performance‑tracking and visualization commands that can monitor, analyze, trace, and log system resources such as CPU, GPU, memory, storage, interrupts, and inter‑process communication.

Originally released in 2021 under the GPL‑2 license, Guider has earned 585 stars and 86 forks on GitHub.

Lee created Guider to meet the need for a highly accurate, low‑overhead real‑time monitoring tool for complex automotive software systems.

What Guider Can Do

Guider can be installed via pip or directly from its GitHub repository. After installation it can configure monitoring for CPU, GPU, memory, storage, drivers, IRQs, and inter‑process communication.

It records measurements with microsecond timestamps and byte‑level sizes, writes data to files for later analysis, snapshots resource usage, counts occurrences, and sends notifications when thresholds are reached.

In a demo, Lee used the memtrace command to trace the memory usage of a 100 MB application.

Many of Guider’s capabilities resemble built‑in Linux tools such as top, but with richer visualizations and automated reporting.

$ python3 guider/guider.py top -a
[Top Info] [Time: 4588832.570] [Inter: 1.0] [Ctxt: 314463] [Life: +0/-0] [IRQ: 26606] [Core: 4] [Task: 498/625] [Load: 0/0/0] [RAM: 125.7G] [Swap: 4.0G]
[Cycle: 8.3G / Inst: 5.7G / IPC: 0.69 / CacheMiss : 13.7M(23%) / BrcMiss: 25.7M(1%) / Clk: 38.7G / MinFlt: 358 / MajFlt: 0]
... (additional output omitted for brevity)

If your application starts slowly, crashes midway, experiences video stutter, or the system resets unexpectedly, Guider can help diagnose the issue.

Automatic Reporting

Beyond the 150 commands, Guider includes powerful reporting and visualization tools that generate SVG charts, flame graphs, heat maps, and other familiar observability formats.

It can be configured to watch events, trigger actions when thresholds are crossed, log anomalies, and even execute additional commands in response.

Lee also demonstrated batch data analysis, capturing performance data from multiple sources such as lidar streams in automotive object‑detection systems and analyzing long‑term trends.

Developers interested in deep performance monitoring are encouraged to try Guider.