Claude Code Tracker for Garmin (github.com)

A new tool called Claude Code Tracker has been introduced, enabling users to log their coding sessions as activities on Garmin watches. This innovative functionality captures metrics such as wrist heart rate alongside custom graphs displaying tokens per second, tools used per minute, estimated server-side watt usage, and cumulative tokens. Users can seamlessly sync these activities to Garmin Connect, making coding sessions quantifiable like athletic workouts. This development is significant for the AI/ML community as it offers a novel approach to tracking productivity in coding, linking physical health metrics with cognitive performance. The implementation runs locally on the user's machine without third-party API calls, ensuring privacy and security, as it utilizes bearer key authentication to connect with the user's Garmin watch via a Cloudflare Tunnel. The sideloaded app does not engage with Anthropic credentials, and it categorizes coding activities as SPORT_GENERIC to avoid influencing traditional training metrics. This project encourages a more holistic view of productivity in AI development, promoting the integration of physical and intellectual activities.