Overview

Everything runs locally. No wrapper, no proxy, no API keys. CodeBurn reads session data directly from disk and prices every call using LiteLLM.

Key Features

• Cost tracking - prices every API call using input, output, cache read, cache write, and web search token counts
• 13 task categories - classified from tool usage patterns, fully deterministic (no LLM calls)
• Daily breakdowns - per-project, per-model, per-activity with one-shot rate
• Models - per-model token and cost table with filtering and grouping
• Optimize - scans for waste patterns and gives copy-paste fixes
• Compare - side-by-side model performance and efficiency metrics
• Yield - correlates AI sessions with git commits to track productive spend
• Menu bar - native macOS app showing today's spend at a glance
• GNOME extension - same functionality on Linux
• Plans & Currency - subscription tracking in 162 currencies

How It Works

AI coding tools store session transcripts on disk as JSONL files, SQLite databases, or JSON. CodeBurn discovers these files, parses token usage from each turn, and prices them using the LiteLLM model pricing database.

The result is a full breakdown: which model, which project, which task type, and how much each cost. The interactive dashboard lets you switch between time periods, toggle providers, and drill into individual sessions.

Quick Start

npm install -g codeburn
codeburn

That's it. CodeBurn auto-detects which AI tools you use and shows the dashboard. See Installation for more options.