token-monitor
Local-first desktop widget for tracking token usage, costs, and limits across 31+ AI coding tools—including Claude Code, Codex, Cursor, OpenCode, and OpenClaw—with multi-device sync.
Token Monitor
A local first desktop widget that tracks token usage, costs, and account limits across more than thirty one AI coding tools with multi device sync.
What it tracks
Token Monitor is a desktop widget that shows live token usage and AI tool limits across more than thirty one AI coding tools, including Claude Code, Codex, Cursor, GitHub Copilot, Cherry Studio and others. Beyond raw usage it surfaces account limit checks and session details where a tool exposes them, and it builds historical usage trends that can be broken down by tool, device, model, session or project. The goal is a single live dashboard for every AI coding assistant a developer runs, rather than checking each tool's own interface separately.
Supported platforms and sync
The application targets Windows 10 or later, macOS 12 or later and Linux x64, and ships through GitHub releases with a Discord community linked from the README. It reads each supported tool's local data paths, for example ~/.claude/projects/ for Claude Code or ~/.codex/ for Codex, and offers real time multi device sync so usage stays consistent across machines. The README lists support per tool as separate columns for token usage, AI tool limits and session details, because not every tool exposes all three. The project is MIT licensed.
Local first design
The README describes the tool as local first, meaning usage data is read from each application's local storage on the user's machine rather than collected through external accounts or cloud uploads by default. This matters for developers who want visibility into spending across many AI tools without sending their session data to a third party service. The multi device sync is presented as an optional layer on top of that local collection, and the breadth of supported paths, from OpenCode's local share database to Antigravity's gemini directories, reflects an effort to cover the current spread of AI coding assistants.
Editorial conclusion
At time of indexing the project reported one thousand five hundred eighteen stars and one hundred thirty eight forks, and the README is localized into Simplified Chinese, Traditional Chinese, Korean and Japanese.
Community notes