Just keep getting better at programming, and have a deeper understanding of a topic like how a computer works low level. That takes time, focus and effort but it will be worth it. It's knowlodge that will remain, who knows MCP how long will be there? Also, even if we reach 99% of the code generated by machines(and I don't believe it if not for trivial code) you'd still need deeper skills to understand it, not only semantically but looking at the big picture in terms of architecture and business/design implications. My suggestion is the contrary of many tech influencers, do not deep dive into prompt engineering or similar stuff, that's the trivial part, if you fail at prompting don't let them convince you you have a skill issue, you're not paid to chat, you're paid to solve problems, prompting is trivial, you must understand problems and requirements deeply. I actually refreshed my high school math and with it I've been able to do so much, from AI basics inner workings to computer graphics, there is so much in core knowledge that is underrated these days. I think I'll soon start the Computer Enhance course from Casey Muratori for low level stuff and performance. Since the advent of LLMs I actually wanted to learn more than before, it has been beneficial to me