What makes Claude Code so damn good

Summary

Claude Code is the most delightful AI agent/workflow I have used so far. Not only does it make targeted edits or vibe coding throwaway tools less annoying, using Claude Code makes me happy. It has enough autonomy to do interesting things, while not inducing a jarring loss of control like some other tools do. Of course most of the heavy lifting is done by the new Claude 4 model (especially interleaved thinking). But I find Claude Code objectively less annoying to use compared to Cursor, or Github Copilot agents even with the same underlying model! What makes it so damn good? If you're reading this and nodding along, I'm going to try and provide some answers. Note: This is not a blogpost with Claude Code's architecture dump (there are some good ones out there). This blogpost is meant to be a guide for building delightful LLM agents, based on my own experience using and tinkering with Claude Code over the last few months (and all the logs we intercepted and analyzed). You can find prompts and tools in the Appendix section. This post is ~2k words long, so strap in! If you're looking for some quick takeaways, the TL;DR section is a good place to start. You can clearly see the different Claude Code updates. Claude Code (CC) feels great to use, because it just simply works. CC has been crafted with a fundamental understanding of what the LLM is good at and what it is terrible at. Its prompts and tools cover for the model's stupidity and help it shine in its wheelhouse. The control loop is extremely simple to follow and trivial to debug. We started using CC at MinusX as soon as it launched. To look under the hood, Sreejith wrote a logger that intercepts and logs every network request made. The following analysis is from my extensive use over the last couple of months. This post attempts to answer the question - "What makes Claude Code so good, and how can you give a CC-like experience in your own chat-based-LLM agent?" We've incorporated most of these into MinusX already an...