Hook

Their other posts in the index, biggest breakout first.
Yesterday, Claude Code dropped a bunch of updates, and they're pretty incredible. I'm gonna share with you three of them and why the last one kept me up till 2:00am testing it. I'm Pete. I run a software development agency in New York City, and in the last few months, I use Claude Code to build my dream social network. This was a big month for Claude code, but yesterday was the biggest day by far. Here's three reasons why. Number one, instant compact. Previously, a compacting session took around five minutes. Now it's instant. I typically recommend using the clear command when starting new tasks, but sometimes you need the context and not getting interrupted for five minutes while compacting is a major win. No.2 usage stats. So finally some visibility in how you're actually using claude code. Type the stats command and you get your usage graphs, streaks, favorite model, and more. And now No. 3 async subagents with the ability to wake up the main thread. So let's talk about how we got here. July gave us synchronous subagents. One runs, everything waits. December 6 gave us background agents. They run while you work, but still synchronous. Yesterday full async with wait messaging. So what is this and why is it a huge deal? So async subagents run in parallel and can wake up your main agent when they need attention. Now that the main thread is unblocked, multiple agents can work simultaneously. But even more importantly, you can have background agents running sophisticated routines and they wake up the main thread when necessary. I started using this for the social network I built for iOS. I have a sub agent monitoring my logs every few minutes looking for constraint issues and errors. I created a sample project last night testing all of this and I just posted it on my Substack. It's a guide on how to create subagents, launch them, test their sync capabilities, even coordinate multiple subagents working in the background together. There's a link in my bio.