Check an address before an agent sends funds or approves it
How can an AI agent check what an address is before sending to it or approving it?
Call GET /v1/chain/address-check with the address. It tells the agent whether the address is an account or a contract, any EIP-7702 delegation, EIP-1967 or minimal proxy details, token details, native balance and transaction count, with plain signals such as a never-used address or an upgradeable proxy.
Example: the USDG token on Robinhood Chain
Example call: chain-address-check
Parameters
| Param | Values | Default |
|---|---|---|
| address | 0x followed by 40 hex characters: all lowercase or a valid EIP-55 checksum | required |
| chain | robinhood (4663) or base (8453) | robinhood |
Signals worth acting on
- A never-used address as a recipient: a typo or a look-alike address is more likely.
- An upgradeable proxy as a spender: whoever controls upgrades can change its logic.
- A token whose name or symbol imitates another: names are chosen by the deployer.
- An EIP-7702 delegation: the account runs the code of its delegate.
The check is read-only and reports facts with a level per signal. Combine it with a simulation of the exact transaction.
FAQ
Does it say whether an address is a scam?
No. It reports what the address is on chain and plain signals. Decide with those facts, a simulation and your own allow-lists.
Which chains are supported?
Robinhood Chain and Base.
Updated 2026-11-16.