Realtime token telemetry for Codex teams

Know exactly how your team uses Codex

A menu bar app that reads your local Codex sessions and a realtime team dashboard: tokens, cache hits, model mix and what it would cost at API prices.

macOS · Windows · WSL2 · Google or GitHub sign-in

Sample session Live
0.0
output tokens / s · gpt-5.5-codex
Devices
3
Cache hit
61%
Today
1.84M · $4.12

tokens

0

Sample figures · last 30 days

Cache hit

0%

Cost

$0.00

Requests

0

What you get

From every developer's machine to one live board

Menu bar app

Live daily totals, tokens per second for the running session, model breakdown and a usage heatmap — right in your menu bar or system tray.

Team dashboard

Organizations map to your Clerk org. Compare members, see active hours and weekday patterns, and spot who is live right now.

Realtime by design

Devices sync hourly buckets to Convex; every chart updates the moment new usage lands. Stored in UTC, shown in your local time.

API-equivalent cost

Subscription usage priced at standard OpenAI API list prices, with cache-hit savings called out — so you know what the plan is worth.

Three steps

How it works

  1. 1

    Run `npx codex-token-tracker login` — nothing to install (Node.js 20+).

  2. 2

    Approve the device in this dashboard with your Google or GitHub account.

  3. 3

    Usage is read from local Codex session logs and synced every minute.

Install the menu bar app

npx codex-token-tracker login --dashboard https://your-dashboard.vercel.app   # first time: sign in + approve
npx codex-token-tracker            # then: menu bar / tray app (nothing to install)
npx codex-token-tracker agent      # headless (WSL2, servers)

Works on macOS and Windows (native and WSL2). Sign in with Google or GitHub.

Mission control

Every session, every device, one board

This is the real team dashboard rendered from sample data — switch the range, hover the charts. Your own data looks exactly like this.

codex-tracker / dashboard / team
Codex Tracker