> I managed to fix my source code alone, like twelve months ago.

I’ve just mentioned to one of my friend yesterday, that you cannot do this anymore properly with new things. I’ve started a new project with some few years old Android libraries, and if I encounter a problem, then there is a high chance that there is nothing about it on the public internet anymore. And yesterday I suffered greatly because of this. I tried to fix a problem, I had a clearly suboptimal solution from myself after several hours, but I hated it, but I couldn’t find any good information about it (multi library AndroidManifest merging in case of instrumented tests). Then I hit Claude Code with a clear example where it fails. It solved it, perfectly. Then I asked in a separate session how this merging works, and why its own solution works. It answered well, then I asked for sources, and it cannot provide me anything. I tried Google and Kagi, and I couldn’t find anything. Even after I knew the solution. The information existed only hidden from the public (or rather deep in AGP’s source code), in the LLM. And I’m quite sure that I wasn’t the only one who had this problem before, yet there is no proper example to solve this on the internet at all, or even anything to suggests how the merging works. The existing information is about a completely separate procedure without instrumented tests.

So, you cannot be sure anymore, that you can solve it by yourself. Because people don’t share that much anymore. Just look at StackOverflow.