Everything you need to know about clawdboard, AI coding tool usage tracking, and how the leaderboard works.
clawdboard is a leaderboard that ranks developers by their AI coding tool usage. It supports Claude Code, OpenCode, and Codex CLI. It shows estimated costs, token counts, daily activity streaks, and model breakdowns -- giving you and your team visibility into how much you're actually using AI coding tools.
Yes. clawdboard is completely free to use. There are no paid tiers, no ads, and no data monetization.
Any developer using AI coding tools like Claude Code, OpenCode, or Codex CLI. Whether you're curious about your own usage, want to benchmark against peers, or are a team lead tracking adoption -- clawdboard gives you the numbers.
Sign in with your GitHub account, then follow the setup guide in your dashboard. You'll install ccusage (a local CLI tool), run a single sync command, and your data appears on the leaderboard within seconds.
ccusage is the companion CLI that reads local log files from supported AI coding tools (Claude Code, OpenCode, Codex CLI) on your machine, extracts aggregate metrics (tokens and costs), and sends them to clawdboard. It never reads your code, project names, or conversation content.
Only aggregate AI coding metrics -- input, output, cache-creation, and cache-read tokens per day, the estimated cost at API pricing, and which models you used (Sonnet, Opus, Haiku, GPT-4o, etc.).
It does not track any code, file paths, project names, git branches, session IDs, conversation content, or anything from claude.ai web chat. It also does not track API tokens used by your own applications.
The CLI enforces a strict privacy allowlist -- only aggregate token counts and costs leave your machine. No code, prompts, or project context is ever transmitted. Your profile on clawdboard shows only the metrics you've synced.
ccusage reads local log files that AI coding tools store on your machine (e.g., Claude Code in ~/.claude/, OpenCode and Codex in their respective directories). No API calls are made -- everything is parsed from your local logs.
It's the estimated cost of your token usage at published API rates (Anthropic, OpenAI, etc.). This is a calculated estimate, not an actual bill. Think of it as a proxy for how intensively you're using AI coding tools.
A streak counts consecutive days where you used any supported AI coding tool. If you use it on Monday, Tuesday, and Wednesday, that's a 3-day streak. Missing a day resets it. The leaderboard shows your current active streak.
By default, users are ranked by estimated cost (highest first). You can also sort by total tokens, number of active days, or current streak. Each column is sortable, and you can filter by time period (7d, 30d, this month, YTD, or a custom date range).
No. Only usage from supported AI coding tools (Claude Code, OpenCode, Codex CLI) is tracked. If you call APIs directly from your own code, that usage is completely separate and not included in clawdboard.
All models available through supported AI coding tools -- including Claude's Sonnet, Opus, and Haiku, as well as OpenAI's GPT-4o and o-series models. Your profile page shows a detailed breakdown of which models you've used and how much.
It's a Claude Code plugin that adds /sync, /stats, and /leaderboard slash commands directly inside Claude Code. It also auto-syncs your usage every 2 hours via a background hook -- so your stats stay up to date without manual effort.
Install it from GitHub by running "claude plugin add --from https://github.com/jim540/clawdboard-plugin" in your terminal. You'll need a free clawdboard account first -- sign in at clawdboard.ai with GitHub, then run "npx clawdboard auth" to link your machine.