bareai
GitHub

See what your AI box
is doing. Right now.

A CLI + TUI for solo engineers on bare metal. Host, GPUs, Docker, and local LLM runtimes — collected, correlated, and read-only, in a single snapshot.

Linux·macOS·Windows·MIT licensed·exits 0 when parts are missing
bareai · live inspect
$ {{ typed }}
host linux/amd64 · 64c · load {{ loadStr }} · ram 48/128G · disk 62%
ok gpu0 H100 80G · util {{ gpuUtilStr }}% · vram 52/80G · {{ gpuTempStr }}°C
ok llm ollama :11434 · llama3.1:70b · healthy · {{ probeStr }}ms
··· docker 3 running · nvidia runtime present
gpu0 · H100
util
{{ gpuUtilStr }}%
vram
65%
host
load
{{ loadStr }}
ram
37%
disk
62%
correlation
:11434→ ollama→ pid 2144→ gpu0
collects & correlates {{ chip }}
01 · install

One line, no sudo required

Pick your platform. Installs to ~/.local/bin — full flags & troubleshooting live in the repo.

{{ installFile }} 
{{ installCode }}
02 · what it does

One question, answered fast

What is this box doing right now?

{{ f.title }}

{{ f.body }}
03 · commands

A verb for every layer

Bare bareai on a TTY opens the dashboard. In pipes it prints help.

{{ c.cmd }} {{ c.desc }}
04 · supported stack

Everything on the box, one tool

Collectors are optional — missing drivers or runtimes degrade cleanly and record a skipped reason.

{{ s.label }}
{{ it }}
05 · output modes

Read it however you work

{{ o.tag }}
{{ o.title }}
{{ o.body }}

Stop SSH-ing five tools together.

Install once, run bareai inspect, and read the whole box.

Get started Star on GitHub
bareai · MIT · inspect / probe only
GitHub Releases Docs Built by Basel Husam