I hate reviewing code. I hate reviewing bad code even more.
Yet there I was, reciting Martin Fowler's Refactoring to Claude for the 99th time—despite the elaborate CLAUDE.md and every skill I'd built up to prevent exactly this.
And the most annoying part?
"You're absolutely right, I ignored the project rules..."
When I'm annoyed, I dig for solutions. So join me on this journey, and let's find out:

* Why LLMs can't write good code—even though they can refactor to it
* How to fix code quality automatically, so you stop reviewing AI slop
* Why this costs fewer tokens, not more
* What it all means for the future of agentic engineering