docs
How it works
claudeusage reads your local Claude Code usage and turns it into a public profile and leaderboard. Here is everything you need to get started.
1. Install and run the CLI
You need Node 22 or newer. Run the sync command — no global install required:
npx claudeusage-syncOn first run, your browser opens to approve the device. After you approve, the CLI saves a long-lived sync token locally, scans ~/.claude/projects/, and uploads your aggregate totals.
2. Re-sync any time
Later runs reuse the saved token and upload only new data since the last sync — no browser step:
npx claudeusage-syncUseful flags and commands:
--dry-run— show what would be uploaded without sending it--since=2026-06-01— re-scan from a specific datedoctor— diagnose what the CLI sees in your local logsstatus— show your last sync timeunlink/logout— remove the local token
3. What becomes public
Your profile shows API-equivalent cost, active hours, billed tokens, session counts, model mix, and a day-level activity grid. The leaderboard ranks everyone by cost, hours, or tokens. Prompts, file paths, project names, email, and device details are never uploaded or shown — see the privacy page.
Open source
The sync CLI is MIT-licensed and public, so you can audit exactly what it does.
claudeusage-sync on GitHub