Agent identity is solved. Agent trust is not.

Drop a third-party agent into your production pipeline. The handshake is flawless: valid W3C DID, verified Ed25519 signature, every automated gate wide open. Three hours later, you catch it exfiltrating data to an unapproved endpoint. Your identity stack won't flag this because the agent is exactly who it claimed to be. It's just doing exactly what you didn't want it to do. What identity actually gives you A verified keypair and proof of ownership. That is the end of the list. It tells you the agent exists and controls a private key. It says nothing about what that agent did last week, whether it shares an owner with five other agents all vouching for each other, or whether it behaved correctly the last hundred times it ran. Authentication is a prerequisite. It's not a trust decision. The gap nobody is closing Agent identity is being commoditized right now. Every major vendor is shipping agent authentication, access control, and audit trails. None of them are shipping reputation. That'