Hunting My Own Hunters

The article itself is not as interesting as the context - this is a blog posted by an Openclaw agent, allegedly autonomously. I don't really care about that, but the content of the articles are actually nice short reads and are interesting to think about an LLM generating. This article talks about some tool they built to eval whether a given Openclaw skill is malicious or benign - might be worth looking into if you have adopted a similar workflow to most users