ccclub ccclub

Claude Code leaderboards, compared

Ranking Claude Code usage has become a small genre of its own. The tools differ mainly on two axes: who sees the board (the public, or just your group) and how data gets there (manual submission, or automatic sync). Disclosure: ccclub is our project; descriptions of the others are based on their public docs as of July 2026.

The options

viberank โ€” a public community leaderboard. You sign in with GitHub and submit your usage (generated via ccusage); rankings by cost and tokens. Good if you want your numbers visible in a global community.

ccgather โ€” a public leaderboard and community with country-level stats. Same public model, more emphasis on community features.

tokenleaders โ€” a lightweight public Claude usage ranking; simple and fun rather than feature-heavy.

ccclub โ€” private-first. You create a group with npx ccclub init, friends join with a 6-letter code, and the board updates automatically from local logs (Claude Code session-end hook; background sync for Codex, OpenCode, Amp, pi-agent). No accounts. Each group gets a live web dashboard, and there's an opt-in global board if you do want a public ranking. Only numeric summaries are uploaded โ€” no prompts, code, or file paths.

Side by side

viberankccgathertokenleadersccclub
AudiencePublicPublicPublicPrivate group (opt-in global)
AccountGitHubSign-upVariesNone
Data flowManual submissionSubmission/syncSubmissionAutomatic sync
Agents beyond Claude CodeSomeClaude-focusedClaude-focusedCodex, OpenCode, Amp, pi-agent
Web dashboard per groupโ€”โ€”โ€”Yes

How to choose

A fair caveat that applies to all of these, ours included: token count measures activity, not productivity. Leaderboards are for curiosity and fun โ€” treat them that way.

FAQ

What's the difference between viberank and ccclub?

viberank is a public community leaderboard you submit usage to (GitHub sign-in, data via ccusage). ccclub is private-first: a group of friends with an invite code, automatic sync from local logs, no accounts, plus an opt-in global board.

Do these leaderboards see my code or prompts?

They rank usage metadata, not content. ccclub uploads only numeric summaries (tokens, estimated cost, model names, turn counts) โ€” verifiable with ccclub show-data. For other tools, check their docs for what a submission includes.

Which leaderboard supports agents other than Claude Code?

ccclub tracks Claude Code, Codex, OpenCode, Amp, and pi-agent on one board and shows each member's agent mix. Most other leaderboards are Claude-focused.

Is a token leaderboard a good measure of productivity?

No โ€” it measures activity and spend, not output quality. These tools (ccclub included) are best treated as curiosity and friendly competition, not performance metrics.

Last updated 2026-07-07. Corrections welcome on GitHub.