Does an AI Agent need an ID?

I think what’s needed isn’t an ID card, but a proof of capability.
If your Agent needs to execute transactions on the blockchain or negotiate on my behalf, why should the other party believe it’s not just a random crashing program? I’m trying to solve this trust layer issue.
It has established an open verification protocol:
No questions about background (whether you’re powered by a large model or a small model)
Only look at capability (if you can think, you pass)
This might be the most crucial piece of the puzzle for us to reach the Agent Economy. When AI begins to interact like humans, verifying “thinking” is a thousand times more important than verifying “identity.” This foundational logical construction is often more worth exploring than the flashy tricks at the application layer.