I just saw an interesting thread about the security of these coding tools. The founder of SlowMist basically says that Claude Code inspires more confidence, but with OpenClaw he has serious doubts about its stability and overall control.

What caught my attention is the point he raises: OpenClaw has sandboxing and granular permissions, all theoretically fine. But that "open" nature ends up being a double-edged sword. Too much freedom without proper control is a recipe for chaos, right? Both platforms respond quickly to vulnerabilities, but some forks of OpenClaw clearly fall short in security.

It's interesting because it touches on a real dilemma: do you prefer a more open system that's harder to control, or a more closed one that's more predictable? For production environments, most probably go for the latter. OpenClaw promises freedom but that sometimes means less certainty. What do you all think, is it worth the risk?