Hermann Bjorgvin released Clawdmeter on GitHub on May 11, 2026, a custom ESP32-based device that monitors Claude Code API consumption in real-time on a compact desk display. The project quickly gained 801 stars and 47 forks, demonstrating strong community interest in physical monitoring solutions for AI tools.
Hardware Combines ESP32 Microcontroller with AMOLED Touchscreen
Clawdmeter uses a Waveshare ESP32-S3 microcontroller paired with a 2.16-inch AMOLED display running at 480×480 resolution. The device includes Bluetooth connectivity, three physical buttons, and capacitive touch input. Two side buttons integrate directly with Claude Code—one transmits Space for voice mode activation while held, the other sends Shift+Tab for mode toggling. A center power button cycles through interface screens.
The physical form factor provides persistent visibility of API usage metrics throughout the workday without requiring screen real estate on development monitors.
System Architecture Polls Anthropic API Every 60 Seconds
The monitoring system operates through three components. A daemon running on the user's computer polls the Anthropic API every 60 seconds by making minimal requests, extracting usage percentages from response headers rather than consuming significant API resources. The daemon connects via Bluetooth Low Energy, transmitting JSON payloads containing session and weekly usage percentages to the ESP32 device.
The dashboard displays three rotating screens:
- Usage screen showing session and weekly utilization percentages
- Bluetooth screen with connection status and device pairing information
- Splash screen featuring animated pixel-art "Clawd" mascot characters
Animations rotate every 20 seconds to prevent visual monotony during extended monitoring sessions. The mascot's activity level correlates dynamically with usage—higher API consumption triggers more energetic animations.
Implementation Written in C for ESP32 Platform
The project is written in C for the ESP32 platform, using Bluetooth Low Energy protocol for wireless communication between the daemon and device. The custom daemon handles API polling and data transmission, extracting usage information from API response headers to minimize consumption.
The 801 stars and 47 forks demonstrate community interest in tactile, visual feedback mechanisms for AI tool usage. The project represents an emerging trend toward physical monitoring devices that provide ambient awareness of digital AI consumption patterns without requiring active attention from developers.
Key Takeaways
- Clawdmeter is an ESP32-based desk device with a 2.16-inch AMOLED display (480×480 resolution) that monitors Claude Code API usage in real-time
- The system polls the Anthropic API every 60 seconds via a computer daemon, transmitting usage data over Bluetooth Low Energy to the ESP32 device
- Three display screens show usage metrics, Bluetooth status, and animated pixel-art mascots whose activity level correlates with API consumption
- Physical buttons provide direct Claude Code integration, transmitting Space (voice mode) and Shift+Tab (mode toggle) commands
- The project gained 801 GitHub stars and 47 forks within days, indicating strong interest in physical AI monitoring devices