Agentic Coding Things That Didn’t Work

In this post, the author shares their experiences with various automated coding tools, particularly Claude Code, focusing on the features that did not work as intended. They discuss the limitations of automating tasks, the challenges of using slash commands, and the difficulties faced when implementing hooks. The author emphasizes the importance of regular evaluation of automations, highlighting the potential risks of mental disengagement that come with reliance on automation. They promote the idea of maintaining a hands-on approach to coding rather than leaning entirely on automated tools.