Last updated July 29, 2026
Shippa (shippa.dev) turns a developer's local AI coding logs (Claude Code, Codex, Cursor) into a shareable stats card and an optional public leaderboard. The core tool, run via npx shippa or the macOS notch app, is 100% local by default: it reads log files already on your machine and computes stats on-device. Nothing leaves your computer unless you explicitly sign in and opt in to the leaderboard.
When you run Shippa locally, it reads session transcripts your coding agent already stores on disk (e.g. ~/.claude/projects) to compute lines shipped, tokens used, cost, streaks, and your most-used model. This processing happens entirely on your machine. Shippa does not read file contents outside those session logs, does not scan your source code, and does not transmit raw transcripts anywhere.
If you sign in (Google or GitHub OAuth) and opt in to the leaderboard, the local tool syncs a small summary snapshot to our database (Supabase): lines shipped, token counts, cost estimate, current streak, and top model — keyed to your account. You can turn this off at any time (Private mode / opt out), which stops future syncs and removes your row from the public leaderboard view.
We never receive your source code, file contents, or full conversation transcripts. Only the aggregate numbers above are synced.
Shippa offers Google and GitHub sign-in exclusively to create your account and attribute your synced stats to a username. We request only your basic profile (name, email, avatar) from these providers — nothing else. We do not post on your behalf, read your email, or access your repositories.
shippa.dev uses privacy-respecting web analytics (Vercel Analytics) to understand aggregate site traffic. This does not track you across other sites and is separate from the local coding-stats sync described above.
Questions about this policy or a data-deletion request? Reach out via the contact link on shippa.dev.