A lightweight, read-only meter for your terminal: your live token count, plan-limit fill %, and model — for whatever AI coding agent you're running, in any IDE. Just the numbers. Want it to also keep your agent on track across context limits? That's the full CLI with the Guardian.
One command prints your live usage; add --watch for a meter that refreshes as you work.
Real, live output from the Vibe Guard Meter CLI.
You don't install the Guardian inside each editor. You install it once on your Mac, and it works alongside whatever agent is open — because it watches from the outside.
Claude Code, Cursor, Antigravity, Windsurf, Codex — one install protects them all at the same time.
Read-only. It shows usage and limits — it never writes, never injects, never sees your code or prompts.
Run it once for a quick reading, or --watch for a meter that updates as you work.
It reads your local context meter only — it never sees your code or your prompts.
A few quick steps in your terminal (every IDE has one built in). After that it runs in the background for all your editors.
Get the .zip above and double-click it to unzip the vibeguard-cli folder.
In Terminal, run pipx install vibeguard-meter (or pip install . from the folder). Needs Python 3.8+.
Run vibeguard-meter for a one-shot reading of tokens + limit fill.
Run vibeguard-meter --watch for a meter that refreshes as you work.
Download the Meter CLI and run vibeguard-meter. Included with your Vibe Guard subscription. Want the Guardian too? Get the full CLI.