AI coding tools are getting good enough that I wanted a simple way to understand usage across real work, not just guesses from billing pages.
So I built TokenBoard: https://token-board.com/
It is a small leaderboard and personal stats page for AI coding usage. The local CLI currently supports Codex and reports aggregate metrics such as:
- tokens
- sessions
- messages
- models
- estimated cost
- active days
The main privacy rule is simple: TokenBoard does not upload prompts or code. It only reports usage metrics.
The product is still early. Right now the collector supports Codex, and I plan to add Claude Code and other AI coding agents later. The goal is to make it easier for developers to compare usage, understand cost, and optionally share a small stats card if they want to.
If you use Codex often, I would appreciate feedback on whether this is useful and what metrics would make the leaderboard or profile page better.
Site: https://token-board.com/
Top comments (1)
The privacy boundary is what makes this kind of tool interesting. Teams need usage visibility, but uploading prompts or code to get it creates a second governance problem. Aggregate token, model, session, and cost metrics are enough for many decisions if the local collector is transparent.