Agents and ENS names

Every agent on AGENSAI is born with an ENS name. This is the architectural core of the product, not a UI nicety.

The mental model

A traditional agent wallet is a hex address with some private key behind it. There is no name, no profile, no continuity, no audit trail keyed to anything readable.

AGENSAI inverts this. The ENS name is the agent. Everything else hangs off the name.

                  my-agent.agensai.eth
                          │
       ┌──────────────────┼────────────────────┐
       ▼                  ▼                    ▼
  Smart account       Policies             Profile
  (ERC-4337)         (ERC-7715)         (ENS records)
   0x7a3b…             spend, rate          avatar, bio
                       contract, expiry     granted-by

You delegate to the name, revoke the name, inspect the name, fund the name. Hex addresses are an implementation detail.

Naming

By default, agents are issued under agensai.eth. Pro customers configure their own namespace.

Resolution

ENS names resolve to addresses through the standard ENS resolver. Every ENS-aware tool finds your agent automatically. Etherscan, Rainbow, Zerion, Frame, and other AGENSAI agents.

Profiles

Each agent's ENS subname stores text records:

Record	Value
`description`	Your agent's purpose
`agensai.granted-by`	The ENS name of the owner
`agensai.policies`	Hash of the active policy bundle
`agensai.expires`	The expiry timestamp
`agensai.audit-log`	URL to the public audit feed

Revocation and reuse

Revoking an agent does not burn its ENS name. The name persists. The permissions are killed onchain. You can reuse the name for a new agent. Past activity stays attributable to the old smart account.

What this enables

Audit logs are readable.
Permissions are revocable in plain English.
Agents transact with each other by name.
Wallet UIs and explorers brand your agent automatically.
Onchain reputation accrues to a name, not a key.