AI coding usage guide

How to Monitor Claude Code and Codex Usage on Mac

Claude Code and Codex usage is easy to check once. The harder problem is seeing every work and personal account, noticing when a reading is stale, and understanding whether a new task fits before a limit interrupts it.

Updated 5 minute read

Limit Lifeboat menu-bar dashboard showing session and weekly usage meters for three Claude Code accounts and one Codex CLI account.
Limit Lifeboat keeps account identity and usage visible in the Mac menu bar.

Use the provider’s own usage view as the source of truth

In Claude Code, run /usage to view plan limits and rate-limit status. In Codex CLI, use /usage inside the interactive interface; OpenAI documents it as the command for account token activity and available rate-limit resets.

# Inside Claude Code
/usage

# Inside Codex CLI
/usage

OpenAI also directs users approaching a Codex limit to the Codex usage page or in-product limit banner. The amount consumed varies with task size, complexity, model, and where the task runs. See Using Codex with your ChatGPT plan.

A percentage needs context

A useful Claude Code or Codex usage monitor should answer more than “how much is left?” Look for:

  • The active CLI identity, especially before work involving managed company or client accounts.
  • Each available usage window, such as the current session and weekly allowance.
  • The reset time, so you can decide whether a longer task fits now or should wait.
  • The reading age, because a last-known value from an inactive or expired login is not live data.
  • Recent pace, which shows whether current consumption is moving faster than the remaining time.

Provider limits can change, and consumption is workload-dependent. Treat the provider’s own current display as authoritative whenever a local view differs.

Know where usage and credentials come from

Claude usage is fetched with the saved account’s OAuth credentials, with Claude Code’s local usage view as a fallback. Codex usage is read from recent local Codex session data. Inactive accounts keep their last successful reading and display its age.

Profiles and history stay under your user Library. App-managed credential snapshots are encrypted by macOS Keychain. Limit Lifeboat has no product analytics, advertising, or telemetry, and the website has no analytics scripts or cookies set by the project.

Review the exact storage locations, network destinations, permissions, and recovery behavior in the privacy documentation. The source and release process are also public on GitHub.

Questions and boundaries

Does Limit Lifeboat estimate token cost?

No. It focuses on subscription usage windows, remaining capacity, reset timing, reading age, and local history rather than translating subscription activity into API prices.

Why can an inactive account show an older reading?

Providers and local CLI data do not always expose every inactive account continuously. Limit Lifeboat keeps the last successful reading and shows its age rather than presenting stale data as live.

Does checking usage consume a meaningful part of the limit?

Ordinary Codex readings come from recent local session data. Claude uses provider usage services with a local Claude Code usage fallback. The app does not submit coding prompts to measure usage.

Free and open source

Keep every authorized CLI account clear.

Monitor usage and switch deliberately without replacing the rest of your setup.

Download for Mac