lazyagent tracks Claude Code sessions

This is exactly the kind of utility that emerges when AI coding tools stop being novelties and start becoming real daily infrastructure. The interesting move here is not more autonomy, but better visibility into what autonomous coding sessions are actually doing.

• –It watches Claude Code’s local transcript files directly, so setup stays simple and the tool remains purely observational
• –The dual interface matters: a terminal-first TUI for active work and a macOS tray app for passive monitoring in the background
• –Surfacing tool history, last written file, token usage, and recent messages makes it useful for debugging stuck, idle, or unexpectedly expensive sessions
• –The project is still early, with minimal Product Hunt traction so far, but the feature set is already sharper than a typical weekend wrapper around an LLM tool