Get started

  1. 1 npx wtclaude setup

    wires up the statusline so WTClaude reads billing-grade data; reads the credential Claude already wrote, so there’s no login.

    npx wtclaude setup
  2. 2 wtclaude today

    your real cost so far today.

  3. 3 wtclaude compare

    the reveal: your billing-grade number next to the broken-log number.

Requirements: macOS or Linux, Node, and the Claude Code CLI in your terminal. Free, MIT-licensed.

WTClaude reads the statusline, which only the terminal CLI emits — the Claude desktop app and Cowork don’t run it. (Desktop/Cowork tracking comes later, as clearly-labeled estimates.)

CLI reference

The launch commands. We show only what ships.

setup
configure tracking — wires the statusline and creates your local data dir.
today / week / month billing-grade
real cost, tokens, and model breakdown for the period — and after June 15 the summary splits the total into your Interactive (subscription) and Agent-SDK (credit) pools automatically. Flags: --json, --since/--until, --group-by .
compare
billing-grade vs broken-log number + the gap. compare --share exports a shareable image card to your Desktop + clipboard.
session [id] billing-grade
list every session (newest first), or per-turn detail for one session.
blocks billing-grade
the rolling 5-hour-block view — buckets every turn into the windows your subscription limit resets on, so you can pace against it.
statusline
render the short status string on demand — mirrors what the collector prints back into Claude Code’s status line.
watch
a live terminal monitor: today’s cost, current 5h-block burn, and the shared plan-limit gauge with a basic, non-predictive time-to-limit estimate.
limit billing-grade
your overall plan limit status (% used + reset countdown). Note: this reads Anthropic’s shared rate-limit bucket, which spans Claude Code, Cowork, and Chat — so it’s your overall plan limit, not a Code-only or per-surface number.
whatif
plan- and model-comparison math. Flags: --plan, --model, --pool, --speed.
forecast estimate
projected Agent-pool spend vs included credits + a June-15 countdown.
readiness estimate
your June-14 credit-coverage check.
credits estimate
Agent-SDK credit balance and burn. Before June 15: your allocation + a countdown. After: live burn rate.
devices
cross-device combined totals + a per-device breakdown. With sync off it degrades to a local-only view of this machine, plus a hint to enable sync.
quality
your one-shot success rate — edits the model got right without a retry.
sync
opt-in cloud backup + multi-device totals. --enable / --disable / --status / --configure. Local-first — nothing leaves your machine until you enable it, and you preview exactly what syncs first.
dashboard
open the web dashboard in your browser — everything the CLI tracks, in one view.
tasks coming soon
token/cost breakdown by task category — coming soon (the live payload doesn’t expose tool names yet, so categories aren’t populated).

Global flags: --json and --csv/--clipboard on the summary commands, and --currency=<ISO> for display-only conversion (USD stays the billing-grade source).

FAQ

Is it accurate? +

For Claude Code, yes — billing-grade. We read the statusline path, the same source behind your bill. Forward-looking projections (the forecast/readiness) are clearly-labeled estimates.

Does it track Cowork or Chat? +

Not yet. Today WTClaude tracks Claude Code. A unified view across Cowork and Chat is coming (as honest estimates, clearly labeled).

Does it work with the Claude desktop app? +

Tracking requires the Claude Code CLI in your terminal — the statusline WTClaude reads only fires there. The desktop app and Cowork don’t emit it, so install via the terminal. (Desktop/Cowork coverage is on the roadmap.)

Will it slow me down? +

No. The collector runs in the background, fire-and-forget. You won’t notice it.

Is my data private? +

Everything runs locally by default. Cloud sync is opt-in, with a preview before anything leaves your machine.

Is it free? +

Yes, and open-source (MIT). A paid Guardian intelligence layer is coming at $2.99/mo · $24.99/year (≈$2.08/mo — save 30%); the tracker stays free.

Who’s behind it? +

An independent project. Not affiliated with Anthropic.