LLM Spend Tracker

See what Claude Code is actually costing you — live, in your status bar.

Claude Code writes a detailed usage log for every session. This extension reads those local logs and turns them into a real-time spend meter and a usage breakdown. No API key, no account, no data leaves your machine.

Status bar: $12.40 today · 38% — spend so far today and how full the context window is right now.

Free

• Live status bar — today's spend + current context-window burn (38% of the model's window), refreshed every 30s.
• Usage panel — total spend, breakdown by model, last 3 days.

Pro (license key)

• Budgets & alerts — set a daily USD budget; get warned at your threshold.
• Full history — 30-day spend trend.
• Per-project breakdown — see which repos burn the most.
• Export — pull the numbers out for your own reporting.

How it works

Reads ~/.claude*/projects/**/*.jsonl — the session logs Claude Code already writes — and prices each message against current Anthropic rates (input, output, and cache read/write tiers per model). Nothing is sent anywhere; all computation is local.

Supports Opus, Sonnet, Haiku, and Fable model families, including prompt-cache read/write pricing (cache reads are billed at 0.1× input, writes at 1.25×).

Privacy

100% local. The extension never makes a network request to read your usage — it only parses files already on disk. License validation (Pro) is the only outbound call.

Requirements

• VS Code 1.85+
• Claude Code installed and used on this machine (so the logs exist)

Pro license

One-time/subscription via Creem. After purchase, run "LLM Spend: Enter Pro License Key" from the command palette and paste the key from your email. The key is stored in VS Code's secret storage and validated against our license server — it is never kept in plaintext settings.