Everything TokenBeaver measures lives on a dashboard served from your own machine. Nothing on it ever leaves your machine.
With the gateway running, either of these works:
The default port is 4505. The VS Code extension opens these same pages from the status bar — CLI and extension users see one and the same dashboard.
Estimated savings, in the frame that matches your setup: a cost view (~dollars you didn't spend, if you're on API keys) and a capacity view (context runway stretched and limit-hits avoided, if you're on a plan). Estimates are modeled locally from your own traffic — close, but not something you'd put in an expense report.
/account · /settings · /help on the same host and port.
--account · --settings · --help-page
Settings is where the telemetry opt-out toggle lives, along with the optimization switches described in the command reference.
Open a GitHub issue → · Questions? Support