The LLM revolution continues to provide perplexing trolly-problem moral quandaries, such as this thing: https://github.com/lombax85/clawguard
On the one hand, do I want a vibe-coded security product, whose value proposition is "alert fatigue as a service", holding root credentials to all of my critical data? Probably not. On the other hand, this project exists because the alternative is just LETTING THE LLM JUST HAVE ROOT ON EVERYTHING WITH NO OVERSIGHT. Would I rather that folks do THAT?!?!