I kept running into the same tiny problem while using Claude Code, Cursor, Codex, and other AI coding tools:
I had no good feel for how large a prompt was getting until after I sent it.
That matters because token usage is not just a billing detail. It changes how fast you burn context, how noisy a session gets, and how surprising the final cost can feel.
So I built TokenBar, a small macOS menu bar utility that keeps token count visible while you work.
Why this exists
AI coding turned a lot of dev work into prompt shaping.
You paste a stack trace, a few files, maybe a spec, maybe a transcript, then suddenly the prompt is not small anymore.
The annoying part is that the meter is usually hidden somewhere else. You only notice token burn after the fact.
I wanted the opposite:
- visible while writing
- native macOS menu bar
- simple enough to keep open all day
- useful for Claude, Cursor, Codex, OpenAI, and other LLM workflows
- one-time price, not another subscription
What TokenBar does
TokenBar gives you a quick way to check token usage from the menu bar so you can trim giant prompts before sending them.
It is not trying to be a giant analytics dashboard. It is more like a tiny odometer for AI work.
If you are using AI coding tools all day, that small bit of visibility changes behavior fast. You start noticing bloated prompts before they waste context.
Who it is for
TokenBar is mostly for developers, solo founders, and anyone using AI coding tools heavily on macOS.
If you have ever thought:
- “How big is this prompt?”
- “Why did this session get so expensive?”
- “Am I about to blow the context window?”
- “I wish I could see the token count before sending this”
That is the use case.
Try it
TokenBar is $5 lifetime for macOS.
I would genuinely like blunt feedback from devs using Claude Code, Cursor, Codex, or similar tools every day.
Top comments (0)