System Monitor - Omarchy bar widget

A live system-monitor bar widget for Omarchy, inspired by iStat Menus: live history mini-graphs on the bar and richer detail graphs in the dropdown panel.

What it shows

On the bar (compact, iStat-style, in order)

• CPU - live mini history graph + current %
• RAM - live mini history graph + current %
• GPU - live mini history graph + current % (auto-hidden when no supported GPU is detected; defaults to the discrete GPU on hybrid systems)
• VRAM - live mini history graph + current % (same auto-hide behaviour)
• NET - live download/upload throughput (down / up), colour-coded

In the dropdown panel (click the widget)

• CPU - history graph, load average, load-level label
• Memory - history graph, used / total
• NET - combined download/upload mirror bar chart (download bars rise, upload bars mirror downward)
• Disk - used / total
• GPU - history graph, utilisation %, and load-level label (like CPU); the GPU model is shown as the secondary detail. Only shown when a supported GPU is detected.
• VRAM - video memory history graph, % used, and used / total (like Memory). Only shown when a supported GPU is detected.

The widget follows the bar foreground, so it darkens correctly on a transparent bar, and it uses the theme accent/urgent colours on the network graph.

Install

omarchy plugin add https://github.com/lcavadas/omarchy-system-monitor.git

Remove

omarchy plugin remove lcavadas.system-monitor

Add to the bar

Add it under bar.layout in ~/.config/omarchy/shell.json (e.g. the right section):

{ "id": "lcavadas.system-monitor" }

or use the bar's drag-and-drop / omarchy bar commands to place it. The shell hot-reloads shell.json, so no restart is needed for layout changes.

Requirements

• Linux (reads /proc/stat, /proc/meminfo, /proc/net/dev, /proc/loadavg, and df -h /)
• Omarchy shell (the bar host)

GPU monitoring is best-effort and auto-hidden when unsupported:

• NVIDIA via nvidia-smi (ships with the driver; no extra install).
• AMD via the amdgpu sysfs interface (gpu_busy_percent, mem_info_vram_*).
• On hybrid systems the discrete GPU is reported by default. Systems with no supported GPU simply don't show a GPU metric in the bar or popup.

Bar visibility toggles

Each metric card in the dropdown has a small switch in its header (left of the title) that toggles that metric's visibility on the bar (CPU, Memory → RAM, GPU, VRAM, Network → NET; Disk has none). Choices are saved to shell.json and survive restarts.

The same toggles are just inline widget settings, so they can also be set from the CLI:

omarchy bar set lcavadas.system-monitor showCpu false
omarchy bar set lcavadas.system-monitor showNet false

Keys: showCpu, showRam, showGpu, showVram, showNet (all default true). GPU and VRAM still auto-hide when no supported GPU is detected, regardless of their toggle.

Tuning

• Poll interval: Panel.qml - the Timer interval (default 2000 ms).
• History length: Panel.qml - historyLength (default 60 samples).
• Bar-network value width: Panel.qml - the NET metric's valueWidth.

Notes

• Lightweight: CPU, memory, disk and network come from standard Linux proc/filesystem reads plus bash (a couple of small file reads per poll). The only external binary is nvidia-smi, used only when an NVIDIA GPU is detected, and it ships with the driver.
• The NET graph only appears in the popup; the bar keeps a compact down/up readout.

License

MIT - see LICENSE.