MyTokenTracker vs ccusage
Short version: MyTokenTracker is built on ccusage. ccusage is the open source tool that reads your coding agents' local logs and prices them, and it is very good at it. MyTokenTracker adds what a terminal report cannot do on its own: plan value, history that survives log cleanup, background sync and a dashboard.
How they fit together
When you run npx mytokentracker, the CLI runs ccusage (version 20.0.26) to read the logs and gets back a daily report grouped by agent and model. Everything ccusage can read, MyTokenTracker can read. The difference is what happens next.
ccusage is maintained independently and is MIT licensed. If you only want the terminal reports, use it directly with npx ccusage@latest.
Side by side
| What you get | ccusage | MyTokenTracker |
|---|---|---|
| Reads local logs from Claude Code, Codex, Copilot CLI, Antigravity, OpenCode, Amp and 12 more agents | Yes | Yes (through ccusage) |
| Price usage at API list rates | Yes | Yes (ccusage's prices) |
| Daily reports by agent and model | Yes | Yes |
| Weekly, monthly and per-session reports | Yes | No |
| Claude 5-hour billing blocks and status line | Yes | No |
| Offline mode, custom pricing overrides, timezone and config files | Yes | No |
| Compare usage with a flat plan price (--plan) | No | Yes |
| Background sync every 30 minutes (macOS and Linux) | No | Yes (free account) |
| History after Claude Code clears its local logs | If you keep the logs (raise cleanupPeriodDays) | Yes (synced totals) |
| Several machines in one view | By hand (custom log paths) | Yes |
| Web dashboard, monthly plan-value receipt, budgets, weekly email | No | Yes |
| Works with no account, uploads nothing | Yes | Yes (until you run init) |
| License | MIT | MIT |
ccusage features from its README as of version 20.0.26.
ccusage alone is enough if
- You only want reports in the terminal, on one machine.
- You want weekly, per-session or 5-hour block reports, or a Claude Code status line. Those are ccusage features MyTokenTracker does not repeat.
- You do not want any account, ever. ccusage never uploads anything.
- You are happy to raise Claude Code's
cleanupPeriodDayssetting so your logs stay longer than the 30-day default (Claude Code docs).
MyTokenTracker earns its place if
- You pay for Claude Pro or Max and want to know, every month, whether it pays for itself.
--plandoes it in the terminal; the dashboard keeps a receipt. - You want usage from last quarter, not just what is still on disk.
- You code on more than one machine and want one total.
- You want it to happen on its own: a background sync every 30 minutes, a weekly email, budgets.
Frequently asked questions
Is MyTokenTracker a fork of ccusage?
No. The MyTokenTracker CLI installs ccusage as a dependency and calls it to read your agents' local logs. It then adds the plan comparison, the background sync and the hosted dashboard. ccusage does the parsing, and we credit it in the README.
Which version of ccusage does MyTokenTracker use?
The CLI currently pins ccusage 20.0.26 and runs its daily report as JSON, grouped by agent and with a per-model breakdown.
Why would I need more than ccusage?
ccusage reports what is on your disk right now. Claude Code deletes local transcripts after 30 days by default, so older usage disappears from any local report. MyTokenTracker syncs daily totals to your account so the history stays, combines several machines, and tracks plan value month after month.
Can I use both?
Yes. They do not conflict. ccusage keeps working as before, and you can use its weekly, session and 5-hour block reports alongside the MyTokenTracker dashboard.
Does MyTokenTracker upload my prompts or code?
No. Without an account nothing is uploaded. After npx mytokentracker init, the CLI uploads daily token totals and cost per agent and model. Prompts, responses, code and file paths never leave your machine.