The research-plan-implement workflow I use to build software with Claude Code, and why I never let it write code until I've approved a written plan.

💫 Toolkit to help you get started with Spec-Driven Development - github/spec-kit

It’s Friday at 4pm. I’ve just closed my 12th bug of the week. My brain is completely fried. And I’m staring at the bug leaderboard, genuinely sad that Monday means going back to regular work. Which is weird because I love regular work. But fixit weeks have a special place in my heart. What’s a fixit, you ask? Once a quarter, my org with ~45 software engineers stops all regular work for a week. That means no roadmap work, no design work, no meetings or standups. Instead, we fix the small things that have been annoying us and our users:

an error message that’s been unclear for two years a weird glitch when the user scrolls and zooms at the same time a test which runs slower than it should, slowing down CI for everyone

The rules are simple: 1) no bug should take over 2 days and 2) all work should focus on either small end-user bugs/features or developer productivity.

I recently shipped a non-trivial Ghostty feature (unobtrusive macOS automatic updates) that was largely developed with AI.

A REST client for the Web

Computers are starting to have opinions on how our APIs should look like

Software teams seeking to provide better products and services must focus on faster release cycles.

Announcing Entire with $60 million seed round and shipping our first product, called Checkpoints.

At <a href="https://www.datawire.io" target="_blank">Datawire</a>, all of our cloud services are developed and deployed on Kubernetes. When we started developing services, we noticed that getting code changes into Kubernetes was a fairly tedious process. Typically, we had to:

VS Code remote support for WSL, Containers, and SSH

Kubernetes for Prod, Tilt for Dev

Create systems that allow you to be consistently productive