I built the same 3-page site six ways — React, Vue, Angular, two vanilla approaches, and my own Rust server — then measured what each one actually costs. The dependency gap was expected. What surprised me was which builds are invisible to Google and AI crawlers entirely.
I tested a real theory: does npm actually cost you more than it's worth? Built the same landing page two ways — with npm packages, and with none — and measured what each one actually shipped. The gap was bigger than I expected. Here's the honest breakdown.
Diff Sniffer started as a suggestion from Claude during a conversation about AI-generated code risk. It checks three things from git metadata: did AI author a commit, did it touch sensitive paths, did a human review it. v0.1 is free on GitHub — building it also taught the author that git hooks exist.
My hand-rolled ANN index should beat brute force as data grows. Instead the gap gets worse — 200 rows it's a tie, by 4000 it's 50% slower. Digging into why, in a toy Rust database built to learn how databases actually work.
For most of us, the temptation to “just build it myself” has always been there. Before AI, it usually wasn’t always worth it. After AI, that equation has changed.
I wanted a blog on my site, not bolted on the side. I didn’t want another dashboard to babysit or a plugin carnival. I wanted something boring, reliable, and mine.
I build lean, low-maintenance SaaS with a boring, reliable stack (.NET, Docker, MySQL). This blog is where I keep it practical: what shipped, what broke, and how to go faster without shipping spaghetti. To make this intro useful, here are 8 underused prompt patterns that consistently save me time. No “act as a…” fluff—just stuff that works.
Coder B Dev