Why it exists
AI agents now hold and spend money, and they can be talked into bad decisions. The data they read can be wrong too: anyone can create a token, put a price on it and send it to a wallet, and a look-alike address can be planted in a wallet's history so the next payment goes to a thief. Svalin checks these things before the money moves.
What it does
| Product | The question it answers |
|---|---|
| Wallet ScanLive | How risky is this wallet's mix of holdings? |
| Check Before PayingLive | Is it safe to send money to this address? |
| Svalin IndexLive | Is there real money behind this token's price? |
| Registry and GuardPlanned | Grades that contracts can read on-chain, and spending rules the chain enforces. |
How we work
- Read-only. Svalin never asks for wallet keys, approvals or signatures, and has no way to move anyone's funds.
- Deterministic. The same data always gives the same answer. No language model computes a number.
- Verifiable. Every Index edition archives its inputs, so anyone can recompute every grade.
- Open. The code and the methodology are public.
The name
In Norse myth, Svalinn is the shield that stands before the sun and keeps it from burning the earth. Svalin stands between an agent and a bad payment.
Status
Svalin is early. The scan, the check and the Index are live, and each chain moves from its test network to real USDC on its own. The Registry and the Guard are designed and described in the whitepaper, and the roadmap shows what comes next.
Questions, bugs and ideas are welcome on GitHub.