llmusage is a command-line tool that answers a simple question: how much is my AI habit actually costing me?
Every AI coding tool β Claude Code, Cursor, Codex, Gemini CLI β stores usage data in its own silo. llmusage pulls from all of them, deduplicates into a local SQLite database, and gives you a unified view of your token usage and spend.
Key Features
- One command sync β
llmusage syncauto-detects local tools and pulls usage data - Unified reporting β
summary,daily,weekly,monthlybreakdowns by provider/model - 8 sources supported β Claude Code, Codex, Cursor, OpenCode, Gemini CLI, Anthropic API, OpenAI API, Ollama
- Zero dependencies β single static binary, no runtime needed
- Privacy-first β everything stays local in SQLite, nothing phones home
- Export anywhere β CSV and JSON output for dashboards or spreadsheets
Install
# Quick install (no Rust toolchain needed)
curl -LsSf https://raw.githubusercontent.com/openrijal/llmusage/main/install.sh | sh
# Or via Homebrew
brew install openrijal/tap/llmusage
# Or via Cargo
cargo install llmusage
Built in Rust for single-binary distribution and fast parsing of months of session logs. MIT licensed.