Your Crew vs the Field

The short version: ccclub

ccclub is the private-league take on token tracking: run npx ccclub init, share an invite code, and your group gets a live board of cost, tokens, turns, active status, and agent mix across Claude Code, Codex, OpenCode, Amp, and pi-agent. No accounts needed, 30-minute aggregate sync, and ccclub show-data lets members audit exactly what leaves their machine. The social model done right — comparing against your crew is a different motivation than a global board, and ccclub builds for it.

SigRank is the complement at the other end of the social graph: instead of a private group, every signed operator is ranked on the same global field by Υ = cache_read × output / input². Your ccclub board tells you who in the group burned the most this week; SigRank tells you who in the field is actually efficient.

Feature comparison

A league calibrates locally; a field calibrates globally

Private boards are honest about their frame — your group's numbers against each other. The limits show up at the edges: the biggest burner in a five-person group might be mid-table globally, and nobody in the group can tell because the league only sees itself. Totals-based ranking deepens that: the member who runs the most sessions always wins the cost column, whatever their efficiency.

SigRank's global board fixes the frame: every entry is a signed snapshot scored on the same Υ formula, ranked against the whole field across 7d/30d/90d/all-time windows, with class tiers to keep scale honest. It is not a replacement for the group leaderboard — it is the thing the group leaderboard can compare itself to.

Bring the group's scores to the field

Each member of a ccclub league can publish a signed SigRank score from the same local telemetry:

npm install -g sigrank
sigrank enroll      # create your operator identity
sigrank submit      # reads logs, scores, signs, publishes

Or paste token counts into the /score calculator to see Υ Yield and class tier instantly.

Frequently asked questions

Is ccclub a SigRank alternative?

It solves a different problem. ccclub is the social model: create a group, share an invite code, and compare token usage, cost, and activity with your friends across Claude Code, Codex, OpenCode, Amp, and pi-agent. The private-league design is a real strength — comparing against people you know is motivating in a way a global board is not. SigRank is the global evaluation: signed snapshots, Υ Yield efficiency scoring, class tiers, and a public field. ccclub answers 'how do I compare to my crew?'; SigRank answers 'how do I compare to everyone?'

What does ccclub do that SigRank does not?

Private leagues. ccclub's invite-code group model means you can run a board for just your team, your friends, or your community — with per-member agent mix and active status. It is the right tool for a closed comparison. SigRank is deliberately public and global: the point of an operator evaluation is that it means the same thing regardless of who is in the room. A private board calibrates against your friends; the global board calibrates against the field.

Can I use ccclub and SigRank together?

Yes. Run ccclub for the friend-group board — the daily 'who is active and what are they driving' view is its own kind of fun. Then submit a signed SigRank snapshot to see how the group members actually rank on efficiency against the global field. The private league for the group chat; the global score for the record.

Which tells me if I am actually efficient?

SigRank. ccclub tracks totals — cost, tokens, turns — which tell you how much everyone used, not how well. Υ Yield (cache_read × output / input²) measures whether the tokens compounded into output or burned as repeated input. In a small group the most active member usually tops the totals; on the efficiency board, the most skilled one does.