Skip to content
bircniPublic

About

Track and visualize your Claude Code, Codex & Cursor usage and cost — heatmaps, stats, and multi-machine sync over a git repo you own.

Topics

Resources

Contributing

Security policy

Stars

4 stars

Watchers

0 watching

Forks

Repository files navigation

aitrack + opentrack

Your AI coding usage in the terminal and the menu bar.

Track Claude Code, Codex and Cursor with two apps built on the same usage data: aitrack for reports and git sync, opentrack for live limits and a desktop dashboard.

npm version CI license

Install the CLI · Download the desktop app · Documentation

Choose your app

aitrack opentrack
Where it runs Terminal on macOS, Windows and Linux macOS menu bar, Windows and Linux system tray
Main use Usage reports, heatmaps, exports and sync Live limits, reset times, pacing and usage at a glance
Providers Claude Code, Codex and Cursor Claude Code, Codex and Cursor
Install npm / npx Download a desktop installer
Shared history Optional git repo you control Reads the same history and can sync from the dashboard

Use either app on its own, or both together. There is no aitrack account or telemetry.

aitrack command-line app

Preview your local usage without configuring a sync repo:

npx aitrack show --tui
npx aitrack usage today
npx aitrack show

Or install the CLI once:

npm install -g aitrack

aitrack is a native binary for macOS (Apple Silicon), Linux (x86_64, arm64) and Windows (x86_64). npm picks the right one; npx and npm need Node.js 18+, aitrack itself does not. The binaries are also attached to each GitHub release. Git is needed for cross-machine sync.

aitrack heatmap showing daily usage, token totals, estimated cost and streaks

  • View daily usage by provider and model, or choose a calendar or rolling period.
  • Compare usage with the previous period and rank your busiest days, models or projects.
  • Render PNG heatmaps and export PDF or CSV reports.
  • Inspect machine totals and sync Claude Code and Codex history across computers.
aitrack usage thismonth --compare
aitrack top projects
aitrack usage month --dir
aitrack export month --csv

CLI guide: commands, configuration and data sources

opentrack desktop app

See how much of your allowance is left without leaving your work. opentrack shows live provider limits alongside the usage history and cost estimates used by aitrack.

  • Session, weekly and per-model limits where the provider exposes them.
  • Reset times, pacing indicators and notifications when limits are running low.
  • Today, yesterday and 30-day usage with daily trends and model breakdowns.
  • A configurable tray icon or bars for your available limits, with colour or monochrome rendering.
  • A tray popup with a compact view, a keep-open mode, global shortcut and optional launch at login.

Install opentrack with Homebrew on macOS 13.5+ (Apple Silicon), or with the CLI on Apple Silicon macOS or x86_64 Windows or Linux:

brew install --cask bircni/tap/opentrack
npx aitrack install-opentrack

Or download it from GitHub Releases:

Platform Download Install
Windows .exe installer Run the installer
macOS 13.5+, Apple Silicon aarch64.dmg Open the disk image and drag opentrack to Applications
Linux x86_64 amd64.AppImage Make it executable and run it
Debian, Ubuntu x86_64 amd64.deb sudo apt install ./opentrack_*_amd64.deb

opentrack checks GitHub for new releases and offers to install them from the dashboard.

The desktop app is self-contained; you do not need Node.js or the CLI to view local usage and limits. Sign in to Claude Code, Codex or Cursor as usual: opentrack reads their saved credentials. Codex live limits require a ChatGPT login.

Windows builds are unsigned; macOS builds are ad hoc signed and not notarized. SmartScreen or Gatekeeper may block the first launch. On Linux the tray icon needs a desktop with AppIndicator support; see the Linux notes.

Desktop guide: limits, credentials, settings and sync

Share history across machines

Create an empty git repository you control, then run the CLI on each machine:

aitrack init
aitrack sync

Use the same remote and a different, stable machine name for each computer. Your normal git credentials handle authentication. The desktop app reads the same configured repo, can pull history automatically, and pushes local usage when you press Sync.

Claude Code and Codex history are synced. Cursor usage stays on the current machine. You can use both apps locally without setting up a repo.

Your data and cost estimates

Claude Code and Codex usage comes from local session logs. Cursor usage comes from its HTTPS usage export using the current machine's saved Cursor login. opentrack also requests live limits from each provider using the credentials that provider's tool saved. Credentials are never written to your usage repo. Synced files do hold your machine name, timezone, and daily usage by model and project (owner/repo, which can name a private repository), so keep the repo private or turn on encryptData. Security & privacy lists everything that leaves your machine.

Costs are API-equivalent estimates, based on model pricing and token/cache counts. They show the pay-as-you-go value of your usage, not your subscription bill.

The CLI selects all three providers by default. To skip Cursor credentials and requests:

aitrack show --tui --providers claude,codex

Documentation

  • CLI guide — commands, reports, configuration, sync and data format.
  • Desktop guide — live limits, credentials and desktop behaviour.
  • Architecture — how the data flows and how the apps fit together.
  • Security & privacy — what leaves your machine and what stays local.
  • Contributing — development setup, project layout, checks and releases.
  • Changelog — changes across both apps.

Acknowledgements

opentrack's provider research and pacing model build on OpenQuota (MIT).

License

MIT

About

Track and visualize your Claude Code, Codex & Cursor usage and cost — heatmaps, stats, and multi-machine sync over a git repo you own.

Topics

Resources

Contributing

Security policy

Stars

4 stars

Watchers

0 watching

Forks

Releases

Used by

Contributors

Languages