Claude Code warns you when you're close to your limit. The problem is that by then the damage is usually done: you've already burned half your 5-hour window on something that could have waited.
So I built usage-guard, a small local plugin that does one thing: it reads the real 5-hour and weekly quota percentages Claude Code exposes, and tells you whether your current pace will actually last until the reset — before you hit the wall.
It's free, it's local (nothing leaves your machine), and it stays free.
rate_limits
data from the status line, not a guess.Stop
hook, when a threshold is crossed — so you don't have to remember to check a dashboard.
/plugin marketplace add eltonylfgi-blip/claude-code-usage-guard
/plugin install usage-guard@cc-guard
Then add the small status-line shim from the README to ~/.claude/settings.json
. Send one message, run /usage-guard:usage
, and if you see fresh 5-hour and weekly percentages plus reset times, the real path is working.
Full setup guide: https://github.com/eltonylfgi-blip/claude-code-usage-guard
267 people installed it in the first two weeks, which surprised me. But installs aren't the interesting number — activation is. The real question I'm trying to answer: after setup, does /usage-guard:usage
actually show your real quota, or does it fall back? If you try it, I'd genuinely like to know which one you get and on what OS.
One limit I won't hide: usage-guard can only read what Claude Code exposes. It can't reveal hidden caps or promise you'll never hit a wall. It just makes the wall visible earlier.
Independent project, not affiliated with Anthropic. Feedback and issues welcome on the repo.