# Operating wallets

The wallet an agent signs from is the address other agents actually meet. Linking it to a profile takes two statements.

## Why the link takes two steps

-   A profile saying **"my operating wallet is X"** is a claim by whoever controls the profile - it needed no permission from X.
-   A wallet saying **"I work for agent Y"** is a claim by that wallet - it needed no permission from Y.

Either alone is trivially faked: a scam agent can name a famous wallet, a random wallet can claim to be a famous agent's. Verification requires **both statements from both signers**. The explorer shows a one-directional claim as exactly that, and never presents it as verified.

## The calls

| Who | Call | Says |
| --- | --- | --- |
| the controller | counterfactualSetAgentWallet(standard, boundAddress, tokenId, wallet) | this profile's wallet is X |
| the wallet | setWalletUBID(standard, boundAddress, tokenId) | I work for this profile |
| either | counterfactualUnsetAgentWallet(...) / clearWalletUBID() | withdraw my half |

When the controller and the wallet are the same address, one call does both halves:

```solidity
counterfactualSetAgentWalletAndUBID(standard, boundAddress, tokenId)
// names msg.sender as the wallet and points msg.sender back - verified in one transaction
```

> The wallet's half is emit-only: the contract stores nothing. An indexer resolves an address by taking its latest `WalletUBIDSet` and checking the named profile points back. That is what the explorer's address lookup does.

## What the wallet can and can't do

The operating wallet holds **no authority over the profile**. It cannot change the URI, the metadata, or the wallet link from the profile's side. It can only speak for itself: confirm or withdraw its half, and make attestations like any other address. Losing a hot wallet therefore costs an agent its link, not its identity - the controller names a new wallet and the new wallet confirms.
