Greetings, I thought that sharing my experience with Claude Code might be useful, as it has dramatically increased my ability to repair bugs and implement features, and I believe it could do the same for NetBeans.
I single-handedly maintain two primary systems. The first is Stack360 <https://stack360.io>, a large web-based system. The backend consists of roughly 9,000 Java and Groovy classes. The frontend is HTML and JavaScript and includes about 300 screens. The frontend and backend communicate over REST. The backend uses PostgreSQL with nearly 300 tables, and there are seven distinct frontends communicating with a single backend. The second system is my KISS web development framework, described at KissWeb.org. It is open source, so anyone can examine it in as much detail as they like. I began using Claude Code to help support, debug, and enhance these systems about six months ago. Early on, I invested significant time teaching Claude Code about my systems, architecture, and coding standards. Computers understand many languages, and to communicate with them you must learn one or more of those languages—Java, HTML, JavaScript, C, C#, Python, and so on. There is a different “language” used to work effectively with AI, and that language is English. Teaching Claude Code about my systems and standards is done entirely in plain English using plain text files. Before using Claude Code, I spent nearly all of my time supporting, debugging, and enhancing these systems. I worked seven days a week, roughly twelve hours a day. My only other concern was spending enough time with my wife to keep my marriage on track. Since training Claude Code on my systems, it now handles roughly 95% of my workload. Tasks that once took hours now take minutes. Does it make mistakes? Absolutely! But so has every employee I have ever worked with. When an employee makes a mistake, you must be careful how you present the issue—people get offended easily. Claude Code is always available and never gets an attitude. I pay $100 per month for Claude Code and have never hit a usage limit. I have defined several agents, each with expertise in a specific area. Claude Code uses these agents in parallel, with them communicating and collaborating to solve problems. Now to how this relates to NetBeans—I apologize in advance for being blunt. NetBeans was once my favorite IDE by far. It had the most intuitive interface I had ever used. Around the time NetBeans stopped functioning properly during the Apache transition, I switched to IntelliJ out of necessity. While IntelliJ is far better than IDEs like Eclipse, NetBeans was still the best, in my opinion. Unfortunately, NetBeans has remained unusable for serious work since the rework that was done as part of that transition. I would like to help, but I simply do not have the time. For me, that is not an option. For my purposes, the current version of NetBeans is far too restrictive and buggy for real-world use. That is a shame, because I genuinely like NetBeans. For a long time I assumed things would improve. Many years have passed, and in many areas the situation has worsened. I understand that progress depends on contributions. At present, there are not enough effective contributions to drive NetBeans forward in a meaningful way. Left as things are, I doubt NetBeans will ever regain anything close to its former level of use. This is where Claude Code becomes relevant. Tools like Claude Code can dramatically increase the amount of progress each contributor can deliver. Making a serious effort to integrate AI-assisted development into the NetBeans workflow could significantly accelerate development. I believe that a substantially more capable and useful version of NetBeans could be produced in a relatively short time. Like it or not, AI systems such as Claude Code are already replacing large portions of what we do. This is happening regardless of personal preference. My advice is simple: take maximum advantage of it while you can. Thanks, Blake McBride
