Recently, people have asked me again how to use cross-chain bridges so you can feel “more at ease”… To put it plainly, what I fear most isn’t the transaction fees—it’s messy bookkeeping. On the bridge side, how many people are required for the multi-signature, and whether the keys are concentrated in the same company; for the oracle, who is feeding the price/data and who is feeding the status—what happens if it gets broken. If you don’t get these straight, don’t complain that I’m being long-winded. And that “waiting for confirmation”—it’s not me just being picky. Once your side has already treated the cross-chain as successful, if the other side rolls back or gets stuck, then if the records don’t match, it’s going to look really bad.

These past few days, AI Agents and automated trading have been getting hot again. Everyone’s talking up “fully automated on-chain interactions,” but nobody wants to go into the security details, and I can’t help feeling pretty uneasy… There’s just too much information, and it makes me anxious. My filtering method is pretty crude: first, go check the project treasury/bridge contract permissions—who can move the funds, and how many keys are needed to move them once. Then check whether there are a pause switch and clear incident logs in abnormal situations. If these don’t pass, no matter how good the narrative sounds, I’ll treat it as if I didn’t see it.