ClaudeGauge
Token Usage Widget for macOS
Overview
Slide open your macOS Notification Center while using Claude Code, claude.ai, or any Anthropic-powered tool and instantly see how fast you're burning tokens, where you stand in your current 5-hour window, and your usage trends over the week.
Built when I was very early to AI-assisted coding and needed visibility into consumption that didn't exist yet. The Opus 3.5–4.1 era models would eat through token limits fast on any larger project — and there was no way to see how close you were to the wall until you hit it.
The early days of AI coding
In the Opus 3.5–4.1 era, Claude Code was already the most capable AI coding tool available — but token budgets were tight, and the models were expensive to run. A complex refactoring session or multi-file feature could burn through your 5-hour allocation in under an hour.
The problem wasn't the limit itself — it was the lack of visibility. You'd be deep in a coding session, the model would slow down or refuse requests, and you'd realize you'd used 80% of your window without knowing it. I needed a glanceable, always-visible indicator of where I stood.
Glanceable token intelligence
The dashboard and widgets are designed for peripheral awareness — you should be able to glance at a gauge and know instantly whether you're fine or need to pace yourself.
Token count gauge (left), burn rate gauge (right), with usage history and AI predictions below
Still works today
ClaudeGauge monitors token usage across Claude Code, Claude Co-work, claude.ai, and any other Anthropic-powered tool by reading local session data. It was built for the early days when this kind of observability didn't exist, but the tool still works well today.
The ecosystem has matured since then — there are now several good monitoring tools available. But ClaudeGauge was one of the first, born from the practical need to not get blindsided mid-session. It's a small example of something I do consistently: when the tool I need doesn't exist, I build it.