Track and compare AI coding-agent usage across Claude, Codex, Kimi, Grok, GLM, Pi, and OpenCode. ccrank.dev turns aggregated local usage into public rankings, history, titles, and shareable cards while keeping raw prompts and responses on your machine. It is open source and powered by Cloudflare Workers, D1, Hono, and TypeScript.
We built ccrank.dev after a simple challenge in a WhatsApp group: make AI coding usage visible and competitive. What began as a Claude Code leaderboard now supports seven agent platforms, multi-machine aggregation, historical snapshots, estimated cost, and shareable cards. The public leaderboard is immediately browsable, the uploader sends aggregate usage rather than prompt content, and the whole project is open source. We would especially value feedback on cross-agent comparisons, privacy expectations, and which local coding tools to support next.