Cappuccino ☕️

Cappuccino is a CLI Tool for monitor the system activity (include the CPU Active, GPU Active etc.) It can shows live CPU (E/P Cluster and pre-core), GPU, ANE(Apple Neural Engine), RAM information with UI. It reads data from macOS's built-in powermetrics and renders it with the termui package

Features

• CPU — E-core / P-core utilization and frequency, with an optional per-core breakdown.
• GPU — utilization and current frequency.
• ANE — estimated utilization and power draw.
• Memory — used/total RAM and swap.
• Power — live CPU / GPU power charts with rolling average, peak and a thermal-throttle indicator.
• Interactive TUI — a bottom hint bar, toggleable help overlay and an in-app REPL for changing settings on the fly.
• Lightweight — a single self-contained binary, no background services.

Requirements

• macOS on Apple Silicon (M1 / M2 / M3 or newer).
• Root privileges. powermetrics only runs as the superuser, so Cappuccino must be started with sudo.
• Go 1.27 or newer only if you build from source.

Installation

From source

git clone https://github.com/starxsky/Cappuccino.git
cd Cappuccino
go build -o Cappuccino ./cmd/cappuccino
sudo ./Cappuccino

With go install

go install github.com/starxsky/Cappuccino@latest
sudo "$(go env GOPATH)/bin/Cappuccino"

Tip: add an alias to your shell profile so you only ever type sudo Cappuccino. Some hints in the app refer to cappu as a shorthand for that alias.

Usage

sudo Cappuccino [flags]

Flags:
  -a, --avg int         Interval for averaged values (seconds) (default 30)
  -c, --color int       Choose display color (0-8) (default 2)
  -h, --help            help for Cappuccino
  -i, --interval int    Display interval and sampling interval for powermetrics (seconds) (default 1)
      --max-count int   Max show count to restart powermetrics (default 0 = unlimited)
      --show-cores      Show per-core usage

Examples:

sudo Cappuccino                 # default 1s sampling
sudo Cappuccino -i 2 -a 60      # sample every 2s, average over 60s
sudo Cappuccino --show-cores    # include per-core gauges
sudo Cappuccino --color 5       # magenta theme

Keyboard shortcuts

Key	Action
q / Ctrl-C	Quit
h / ?	Toggle the help overlay
:	Enter command mode (REPL)
+ / -	Increase / decrease the sampling interval
c	Cycle the bar color
Esc	Close the help overlay / cancel a command

Command mode (REPL)

Press : to open the command prompt at the bottom of the screen, type a command and press Enter. Esc cancels.

Command	Description
help	Show the help overlay
quit / exit	Quit
interval <seconds>	Set the sampling/display interval (restarts powermetrics)
avg <seconds>	Set the averaging window
color <0-8>	Set the bar color
cores <on|off>	Toggle the per-core layout
maxcount <n>	Restart powermetrics every n samples (0 = off)
clear	Clear the current status message

Supported SoCs

CPU/GPU max power and bandwidth tables are built in for:

M1, M1 Pro, M1 Max, M1 Ultra, M2, M2 Pro, M2 Max, M2 Ultra, M3, M3 Pro, M3 Max, M3 Ultra.

SoCs without a dedicated table fall back to conservative defaults, so the tool still works — only the percentage scales may be less precise.

How it implement ?

Cappuccino spawns powermetrics in the bacground and catch its plist output to a temporary file :

powermetrics --samplers cpu_power,gpu_power,thermal -f plist -i <interval_ms> -o /tmp/cappuccino_powermetrics<timecode>

Each sample is a NUL-spearated plist document. Cappuccino parses the newset one and normalizes it into internal/metrics structs, and renders the result with tool termui. Mem statistics are read separately from vm_stat and sysctl.

Contributing

Any Issues and pull requests are welcome, but please make sure you have executed below commands before you opening a PR :

gofmt -w . 
go vet ./... 
go build ./... 

Please keep changes focused and include a short description of what you changed and why.

Acknowledgements

• Inspired by asitop.
• Terminal UI powered by termui.
• CLI parsing by cobra.
• Plist decoding by howett.net/plist.