Claudelytics

Claudelytics
⚠️ BETA SOFTWARE: This is pre-release software under active development. Features may change, and bugs may exist. Please report issues on GitHub.

A fast, parallel Rust CLI for auditing Claude Code and Codex usage, token consumption, and costs from local logs. It answers the practical question “where did my agent budget go?” with unified daily/session totals and source-aware model breakdowns.

✨ Features

Core Analytics
- 📊 Daily Reports: Analyze usage patterns by day with token counts and costs
- 📈 Session Reports: Break down usage by Claude Code and Codex sessions
- 🤖 Cross-Agent Audit: Normal execution shows separate Claude Code/Codex totals and per-model usage after the unified daily report
- ♻️ Cache Accounting: Keep cache writes and reads separate; Codex cache writes are shown as unavailable when local logs do not expose them
- 💰 Cost Tracking: Quick cost summaries for today, specific dates, or totals
- ⏰ 5-Hour Billing Blocks: Track usage in Claude's actual billing periods (00:00-05:00, 05:00-10:00, etc. UTC)
- 🔍 Flexible Filtering: Filter data by date ranges with --since and --until
- 🎯 Model Filtering: Filter Fable, Opus, Sonnet, Haiku, OpenAI families, aliases, or exact models
- 📚 Current Model Pricing: Audit Claude Fable 5 and GPT-5.6 Sol/Terra/Luna, including GPT-5.6 long-context pricing
- 📋 Model Registry: Future-proof model management with aliases and automatic recognition
- 💾 Offline Pricing Cache: 7-day cache for pricing data to work without internet connection

Interactive & Real-time
- 🎯 Interactive Mode: peco-style fuzzy searchable session selector
- 🎨 Enhanced TUI: Eight responsive views including audit overview, agent/model usage, sessions, cache, and billing blocks
- 👀 Watch Mode: Real-time monitoring with automatic…