WTClaude
Billing-grade cost tracking for Claude Code
I'm a heavy Claude Code user. For weeks I trusted a usage tracker — until the number stopped matching my bill. Turns out it (like most) reads the local session logs, which were never the billing source, so the cost drifts. Sometimes by a lot.
So I built WTClaude. It reads the Claude Code statusline — the same source behind your bill — so your today / week / month cost is billing-grade in the terminal, not a reconstruction. The command I'd start with is wtclaude compare: it shows your real number next to the log-based one, and the gap between them.
It's free, open source (MIT), local-first, and independent (not affiliated with Anthropic). It also forecasts the June 15 Claude Code billing split and tracks Claude Fable 5's June 23 cost cliff — both clearly labeled as estimates, because being honest about what's exact vs. a forecast is the whole point.
Launched today with zero pre-marketing — this is day one. Install is npx wtclaude setup. I'd love feedback from other builders, especially where your compare gap lands.
7 Comments
The "what are they actually buying" question hit close to home -- I'm also building a Claude Code statusLine tool (a status line that pays you a disclosed cut when it shows a sponsor line while you're idle), and I assumed the pitch was the money. It might not be. The honest disclosure part -- always labeled, never hidden as a normal tip -- might be what actually earns trust with this exact audience, and the money is just what gets someone to try it once.
Same statusLine hook you're using for wtclaude compare, different data source (mine polls a server, yours reads local logs) -- be interested to compare notes on refresh-interval tradeoffs sometime if you're down. Congrats on shipping.
The compare command is the right wedge. For Claude Code users, the pain is not just exact billing, it is realizing mid-session that a spike or reset window changed behavior before the invoice catches up. If you show the gap plus the moment it started drifting, that becomes much more actionable than another total cost number.
One thing I'd be careful with:
The interesting question may not be whether the tracker is more accurate.
It's whether accuracy is actually what users are buying.
Those sound similar, but they can lead to very different decisions over time.
I wouldn't make that call casually from early feedback.
You're right. I hope users find my tools valuable, like tracking Fable until the free period is over. I just want to help inform peoples decisions, not influence them.
Possibly.
I'd still be careful.
A lot of products end up measuring one thing, improving another, and creating value somewhere else entirely.
That's one of those decisions that can look obvious early and very different later.
I will be. I'm constantly looking at how people will use it, and what theyll get from it, so a pivot is easy if needed.
That's fair.
I'd just be careful assuming the ability to pivot later removes the need to make the right decision now.
Those are often two different things.