What problem does it solve?
When your Claude Code quota runs out unexpectedly or a day's usage looks unusually expensive, raw logs are hard to interpret. This Skill turns local ccusage data into an evidence-based, plain-language explanation of token consumption, cost, model mix, and 5-hour block burn.
Core Features & Use Cases
- Usage Summaries: Runs a bundled Python analyzer over ccusage daily and block data to report total tokens, estimated cost, input/output split, and cache create/read volumes for any date range and timezone.
- Model Comparison: Compares two models (e.g., fable vs opus) on both token volume and estimated cost, showing when similar token counts still produce very different bills.
- Quota Burn Diagnosis: Breaks usage into 5-hour blocks to explain rapid quota exhaustion, distinguishing real typing from cache-read pressure.
- Use Case: A user asks why their Claude quota was exhausted after only two questions; the Skill shows the session involved many internal turns with cache reads dominating over 80% of tokens.
Quick Start
Ask the assistant to explain today's Claude Code token and cost usage with evidence from ccusage.
