AgentValley
I love that someone finally looked at the Mac notch, the thing we all either ignore or complain about, and saw a status dashboard for AI agents. It's a specific solve: if you're running multiple Claude Code agents, you've probably lost track of which one is doing what. AgentValley puts that right in your peripheral vision.
This points at a bigger shift from writing code to supervising it, and supervision needs observability. The terminal isn't the right surface anymore; your menu bar and notch are. I'd want to know how it handles more than a handful of agents, given the space limits, and whether it offers any control beyond watching.
The cynical read is that this is a novelty toy for early adopters, but the practical read is that agent observability is becoming a real category. Tools that make it feel effortless will stick. I'm curious to see if AgentValley grows into something bigger or stays a sharp little utility, but either way it's a sign that tooling around agents is getting creative.