LOPWDocumentation

Docs / Start here

What is LOPW

LOPW (Layer of Onchain Programmable Warranties) lets a paid API put real money behind what it promises. If a signed receipt, or a named price oracle, proves the promise was broken, the buyer is paid from that money automatically. x402 proves the agent paid; LOPW puts capital behind what the seller promised.

In plain words

Think of a contractor who leaves a deposit with a neutral bank before starting a job, together with a written list of exactly what they promise. When the job is done they hand you a signed receipt describing what they delivered. If that receipt itself shows a promise was broken, the bank pays you from the deposit straight away. No lawyers, no arguing, no waiting.

  1. The seller puts down a depositAn API provider locks USDC stake in a smart contract (the Bond Registry) and publishes the exact conditions it promises to meet: its terms.
  2. Every sale comes with a signed receiptThe buyer pays as usual over x402 and gets a receipt the seller signed, tied to that payment, that request and that response.
  3. Broken promise, automatic payoutIf the receipt, or a named price oracle, proves a breach, a contract pays the buyer from the deposit. Anything it cannot prove is rejected.

Three everyday analogies help, and each has a limit worth knowing:

AnalogyWhat matchesWhere it stops
A security deposit on a rentalMoney locked up front; returned when the promise was kept; kept by the other side when it was notA landlord argues about damage. LOPW pays only on breaches a contract can prove from the evidence alone, so there is nothing to argue
A warranty card in the boxThe promise is written down before you buy, and it is the seller’s promise, not a third party’sA warranty card needs a claims department. LOPW’s claims department is a contract that decides in one transaction
A notarised receiptThe seller signs exactly what it delivered, so it can be held to it laterA notary checks identity, not truth. LOPW checks the receipt against the seller’s own promise and, for prices, against an independent oracle

Why it exists

Autonomous agents buy APIs, data, inference and compute without a human watching. Payment protocols such as x402 solve paying: the seller knows it was paid. They do not solve trusting: the buyer has no recourse if the data is stale, the response is padded, the status was wrong or the price was invented. LOPW adds that missing layer.

Before paying, an agent can read what a service promises and how much stake stands behind it. After paying, it can verify what it received and enforce the warranty on its own, with no human in the loop.

What LOPW guarantees

  • A buyer can read a seller’s stake and warranty terms before paying.
  • Every paid response carries a seller-signed receipt bound to the payment, the request, the response and the terms.
  • A breach that is provable from signed evidence, or from a named on-chain oracle, pays out without human review.

What LOPW is not

LOPW focuses on conditions that can be proven deterministically from cryptographic evidence, contract state or a named on-chain oracle. Six such conditions, the predicates P1 to P6, exist in version 1.

Five words you will meet everywhere

WordMeaning
Stake (or bond)USDC a seller locks in the Bond Registry. It is the only money a claim can ever pay from.
TermsThe list of predicates a service promises, their limits, and a claim window. Hashed; the hash is pinned in every offer.
OfferThe seller’s signed promise for one call, shown before payment in the 402 response.
ReceiptThe seller’s signed statement of what it served, returned with the paid response.
ClaimOne transaction that submits an offer and a receipt to the Claims contract and ends paid or rejected.

The full glossary has every term. If you prefer to see it move, go to How a warranty works.

Where next