How I've been using Claude Code

This article discusses the author's experience using Claude Code to develop Hegel, an open-source property-based testing library. The author notes that roughly 90% of the code is initially written by Claude, but emphasizes that this is not 'vibecoding'—human review and design control are critical. The team uses a standard pull-request workflow with two human reviewers to catch mistakes. The article highlights both the productivity gains and the pitfalls of relying on AI for code generation, such as Claude oversimplifying complex protocols. It is relevant to IT/technology as it covers AI-assisted software development, coding practices, and code review workflows.