TokenTracker: Local-First Menu Bar AI Token Usage Tracker for macOS
Summary
Architecture & Design
Core Workflow
- Auto-detects running AI coding CLI/desktop apps on macOS
- Captures token usage metrics via local process inspection (no API calls required)
- Stores all data locally in
~/.token-tracker - Presents real-time metrics in a menu bar dropdown dashboard + 4 native WidgetKit widgets
Feature Table
| Feature | Details |
|---|---|
| Supported Tools | Claude Code, Codex, Cursor, Gemini, Kiro, OpenCode, OpenClaw |
| Configuration | Zero-config out of the box, no API keys required |
| Visualization | Menu bar dashboard, 4 WidgetKit widgets (daily usage, total spend, top models, active sessions) |
| Privacy | 100% local data storage, no external telemetry |
Typical Developer Workflow
Install the app → launch your preferred AI coding tool → open the menu bar icon to see real-time token counts and estimated costs. Pin widgets to your macOS desktop to keep usage front-and-center during coding sessions.
Key Innovations
Key Pain Points Solved
- No more manual tracking: Most AI coding tools don't include built-in token cost tracking, forcing developers to use spreadsheets or third-party APIs that expose usage data
- Full privacy compliance: Unlike cloud-based trackers, all data stays on your local machine with zero outbound API calls
- Native macOS integration: Menu bar access and official WidgetKit widgets avoid cluttering your desktop with extra browser tabs or electron app windows
DX Improvements
Unlike tools like LLM Cost Calculator that require manual log input or cloud sync, TokenTracker works out of the box with zero setup. It automatically identifies running AI coding processes without requiring you to copy/paste API keys or usage logs.
Unique support for niche coding AI tools like Kiro and OpenClaw that are overlooked by larger LLM tracking platforms.
Performance Characteristics
Resource Footprint
- Menubar app uses <2% CPU and <20MB RAM during active monitoring
- WidgetKit widgets update in 1-second intervals without noticeable performance impact
Alternative Comparison
| Tool | Local-First | macOS Native | Supported AI Coding Tools | Zero Config |
|---|---|---|---|---|
| TokenTracker | ✅ | ✅ | 7+ | ✅ |
| LLM Cost Calculator | ❌ | ❌ | 5+ | ❌ |
| OpenAI Usage Dashboard | ❌ | ❌ | 1 (OpenAI only) | ✅ |
| CodeLlama Tracker | ✅ | ❌ | 1 (self-hosted only) | ✅ |
TokenTracker is the only tool that combines native macOS integration, full local-first storage, and support for all major commercial AI coding assistants.
Ecosystem & Alternatives
Integrations & Adoption
- Native support for all top macOS AI coding tools listed in the project description
- Lightweight standalone binary with no external dependencies beyond default macOS frameworks
- Currently adopted by small teams of AI-native developers and indie builders working with Claude Code and Cursor
- No official plugin ecosystem yet, but the open source codebase supports easy addition of new LLM tools via simple process matching rules
Community Status
The project has 236 GitHub stars at launch (as of April 2026) with steady weekly growth, indicating strong demand for privacy-focused AI usage tracking tools for macOS.
Momentum Analysis
AISignal exclusive — based on live signal data
| Metric | Value |
|---|---|
| Weekly Growth | +5 stars/week |
| 7-day Velocity | 252.2% |
| 30-day Velocity | 0.0% |
The project is in the early adoption phase, having launched just 2 months prior to this analysis. The 7-day velocity spike indicates a recent burst of organic interest, likely driven by increased adoption of Claude Code and Cursor among macOS developers. The flat 30-day velocity is a temporary artifact of the project's very young launch date.
Forward-looking: The tool fills a clear unmet need, and will likely see accelerated growth as more developers adopt commercial AI coding tools and demand privacy-first usage tracking.
Get analysis like this — weekly
New deep dives + trending repos, straight to your inbox. Free.
Free weekly AI intelligence digest