AI Pulse · Under the Radar

A small tool called codenotch is getting attention because it solves a very practical problem. It pins usage-limit meters for Claude Code, Cursor, Codex, and Antigravity to the edge of your screen, so you can see where you stand without checking each product separately.

Built by one person with no press push behind it, the project is climbing at around 336 GitHub stars a day, making it one of the faster-moving small repositories this week. You can see the project and its supported tools in the original GitHub repository.

The interesting part is not the menubar app itself. It is the behaviour behind it. Teams are now using several AI coding agents at once, each with its own plan, limits, reset dates, and overage rules. That gets messy fast. A developer hits a limit halfway through work, switches tools, and the business has little visibility into what is being used, what it costs, or where work is being interrupted.

For an owner, this is an early operational signal. AI spend is moving beyond a handful of software subscriptions. It is becoming a set of variable usage decisions made across the day by people trying to get work done. The firms that manage this well will need a simple view of usage, cost, access, and outcomes across their AI tools. This is the kind of thing we build into an AI command centre, turning scattered tool activity into something leaders can actually monitor.