The most complete Claude Code tracker.

Reads the statusline, the same source behind your bill — not the broken session logs other trackers use. (Billing-grade for Claude Code run in your terminal.)

wtclaude compare puts your real number next to the broken-log number, with the gap multiple. The moment that changes how you see your costs.

Each figure is tagged billing-grade, inferred, or estimate. We never show an estimate as if it were exact.

Opus 5.5 vs Sonnet 5 vs Fable 5.1, re-priced from your own billing-grade usage. Per-surface split; every projection labeled estimate.

Prices Claude Code’s default model correctly ($4/$20, cache reads $0.20 per million) off the model’s own tokens — no unknown-model gaps.

Prices Sonnet 5 correctly ($2/$10 — the $3/$15 increase once scheduled for September 1, 2026 was cancelled) off the model’s own tokens — no unknown-model gaps.

Billing-grade Fable cost in the terminal + a labeled forecast of usage-credit cost on the plans where Fable bills as credits — Pro and Team Standard from the first token; Max, Team Premium and Enterprise Premium are included up to 50% of the weekly limit. Covers Fable 5.1 and Fable 5, each priced at its own rates.

Tokens, cost, and model breakdown over any window.

Shows the always-loaded skills and subagents you’re paying for but never use, with an estimate of what re-reading them costs at each model’s list cache-read rate. CLAUDE.md files are listed as always loaded but not judged. A recommendation to review, never an automatic change.

A visual timeline of your sessions — see exactly which turns burned the most.

A plain-language summary of your day's usage.

Compare plans against your real usage before you switch.

Where your tokens go, broken down by what you were doing.

How often your edits land first try, no rework loop. The quality command ships; the edit-target field it reads isn’t in current Claude Code payloads yet.

See costs in your currency (the billing-grade figure stays in USD).

Everything runs on your machine by default. Nothing leaves it unless you turn sync on.

Sync your usage history to the cloud so you can see it on the dashboard and across machines — opt-in, with a preview of what syncs.

Laptop, desktop, work machine — one true total across all of them. The number no single-machine tracker can give you.

Track more than one Claude account or profile, side by side.

Opt-in everything, preview before sync, and a clear view of what’s shared.

Everything the CLI tracks, in a dashboard you can look at: overview, session timeline, the compare view, the model-comparison and context-waste tiles, the what-if calculator, your limit gauge, and the agent-pool forecast tile.

Anthropic’s announced Agent-SDK billing split was paused on June 15, before it took effect. WTClaude’s dual-pool view is built and ready — if a revised version returns, it would light up with a WTClaude update.

Projected agent-pool spend — a labeled estimate; if a revised split lands, it is compared against the credits that split includes. (Billing-grade math, heuristic pool split — a forecast, not your exact bill.)

Paused with the split — there are no Agent-SDK credits to check today. If a revised split lands: one line, are the credits it includes enough, yes or no, and the one thing to do about it.

How close you are to your overall plan limit, with reset countdowns. (This is your shared plan-limit bucket, read from your latest Claude Code status update.)

Recognizes Claude Fable 5.1 and Fable 5 in your terminal, reads their cost billing-grade, and forecasts what your pace would cost as usage credits on the plans that bill Fable as credits with wtclaude fable — a labeled estimate, kept separate from the paused June-15 change.

The tracker is free forever. The moat is the data and the paid layers, not the tracker code.

Not affiliated with Anthropic. On your side — we recommend what’s right for you.

Chat, Cowork, and Code together — with honest, clearly-labeled estimates where billing-grade isn't possible.

Personalized recommendations, anomaly detection, spend forecasting, and coaching. $2.99/mo · $24.99/year (≈$2.08/mo — save 30%).