Have you guys also been flooded by AI Agents recently... talking like they’re fully automatic printing presses. I find it a bit scary when I monitor fund rates at night. If we really let it go on the chain to do work, I think the most critical parts that need human oversight are three: first is authorization/signature—don’t give unlimited permissions easily, or the Agent might get excited and treat your wallet like a company seal; second is the “final confirmation” before transactions, such as slippage, routing, whether the token contract is just a renamed scam, machines can calculate but may not understand the pitfalls; third is stop-loss and withdrawal after errors—things like gas issues, MEV sandwich attacks, route fluctuations—someone really needs to watch and shut them down promptly. Anyone can talk about the narrative, but when it comes to security, you still have to pay attention to the details... Anyway, I only dare to let it do “prompting and semi-automatic” tasks now. Fully automatic, I really don’t dare to sleep.