About

I read before I write. In a mature codebase the most useful thing I can do is trace how something actually works before I touch it, and I've carried that habit into my own projects.

I make small changes I can defend. Big refactors in a system with years of accumulated behavior break things you find out about a month later. I'd rather ship the smallest fix and explain exactly why it's safe.

I fix the pattern, not the instance. Twenty-two of my fixes went into shared controls rather than the screen the ticket named, and each one closed issues on screens nobody had filed yet. Every fix ends with a sweep for the same bug elsewhere.

I write down what broke. Every project here has a section on what went wrong and how I fixed it, because that's the part I actually learned from and the part I'd want to know about someone else's work.

I use Claude Code every day, at work and at home, and I built my team's tooling on it: six skills the team runs daily, from a Puppeteer driver that verifies a fix end to end, to an evidence-video pipeline for reviewers, to ticket drafting in two languages. The useful skill turned out not to be prompting. It's knowing where it saves hours and where it confidently makes things worse.