Pick an attack on a wire transfer and watch execution refuse. The digests are computed in your own browser. Then take the record it hands you and check the second signature against a published key, again in your own browser.
What the demo does not do. It proves the instruction did not change in custody. It does not prove the answer was right or that any model was honest. The second signer is a separate key, a separate clock and a separate log operated by Hive, so it is not an outside party and the service reports that itself. It signs Ed25519, which is not post quantum. The operator side signs ML-DSA-65 under NIST FIPS 204. The second signer's log does not survive a redeploy today, and it says so too. Patent pending.
Hive
Hive Proof Architecture
Then it shipped a stack that makes agents economically capable. Hive adds a receipt another party can verify without asking the operator to restate what happened.
Circle is building the regulated money, the wallet, the marketplace, the policy layer, the chain, and the settlement path for agents. That is the hard part, and it is Circle's. Hive is an additive, removable proof sidecar that binds independently verifiable evidence beside those actions, without changing Circle's route, price, policy, or settlement logic.
Stated up front because it determines whether anything else on this page is worth reading. This is a proposed structure, not a current commercial arrangement.
The customer-facing proof product, the packaging, the pricing, and the relationship. The wallet policy stays enforced at Circle's wallet layer. Routing, orchestration, attestation and settlement stay authoritative and unchanged. Nothing moves out of Circle's control.
The signer and verifier underneath, so the artifact a counterparty checks has cryptographic separation from the operating record it references. That independent signature adds portability without replacing Circle's native evidence.
Circle's payment path runs across the top. Hive receipts bind underneath as each step passes. Switch Hive off and watch what happens to the path.
Take Hive out and Circle's behavior is identical to today. There is no shim to unwind, no policy to restore, and no settlement dependency to migrate, because Hive never held any of them. A receipt already issued still checks from the artifact itself, offline, against published key material, with no call to Hive.
Hive makes no claim about Circle's internal systems beyond what Circle publishes. Every item above is linked at source in the footer.
Circle Gateway supports ERC-1271, so a smart contract, a smart wallet, or an application can reach a unified USDC balance while its own existing authorization logic keeps making the decision, with no delegate accounts and with expanded crosschain access. That is a clean fit for evidence: the authorization already happens in one place, and the flow already produces the exact references a receipt needs. Hive adds an independently checkable receipt beside the existing flow and changes nothing about it.
The contract’s ERC-1271 validation result and Circle’s Gateway attestation remain the native evidence for what was authorized and what settled. Hive preserves those references as supplied and never restates them as its own claim.
Beside the flow, Hive can operate a separate countersigner and log whose evidence is independently checkable; true institutional independence requires non-Hive custody. Hive does not evaluate, gate, replace, or change any authorization policy, does not sign for any wallet, does not hold or move funds, and is never in the authorization path. Emission is out of band, so the transfer never waits on a receipt.
Private keys, seeds, session material, raw call data, and raw request bodies are never sent, held, or stored. Addresses, chain identifiers, public references, and caller computed digests are the only inputs.
Source analysis of the existing contracts found no unfilled proof boundary here, so the mapping reuses primitives that already exist and already verify. A new type would duplicate boundaries rather than fill a gap.
Maturity labels are taken from the Hive Proof Architecture pages themselves. Nothing is labeled live unless a Hive production endpoint answers today. Instruments filed as art and not yet exposed as an endpoint say so.
Pick a Circle boundary. The architecture keeps each system in its proper role.
A suitable scope is one Agent Stack or x402 route for 30 days, with evidence bound beside the action. Circle's route, price, policy, and settlement logic stay exactly as they are.
The integration scope is limited to the proof fields and the signing path Circle chooses to expose, and Hive removes itself on request at any point.
These receipts add a direct comparison of two settlement records and a check that one transaction fit its delegated mandate beside the settlement, authority, and delegation work already described here. They answer in production right now, and the runs below prove it end to end. Three more from The Authority Line sit directly on that work: whether the granter held what it granted, whether every hop narrowed, and whether one mandate was spent twice across acceptors with no shared ledger.
This receipt lets settlement and operations teams compare two named records at specific cursors and times. It fixes the field list, uses separate registered attestor keys, and commits a fingerprint for each record. The service recomputes whether the records match within the declared timing window. A divergence remains explicit when a source and destination view need investigation. Legal, risk, and counterparties can inspect the result without seeing balances, holders, or account identifiers.
What it does not do. It does not reveal positions, balances, holder identities, or account identifiers, and it cannot confirm that either fingerprint accurately represents the record named because it grants no access to that record. It does not choose the correct record, assign fault, determine whether the settlement, transfer, or register update was proper, or carry out or reverse settlement, and a late window says only that the observations were too far apart to decide.
Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.
Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.
Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.
This receipt compares one named transaction with a delegated authority receipt signed before that transaction was authorized. It checks the declared amount, currency, timing, and scope against that mandate's constraints. The service recomputes the outcome from the comparison rather than relying on a caller supplied result. That gives an Arc or stablecoin settlement reviewer a narrow, checkable statement about this action and the supplied delegation. It can accompany the authority record without changing the system that decides or settles the transaction.
What it does not do. It does not prove that a cardholder granted the delegation, that the agent identity is genuine, that a network authorized or settled the transaction, that goods or services were delivered, or that a person read the displayed terms. It is not payment authorization, holds no cardholder credential, has no recognition as authentication data, compelling evidence, or liability shift from a network, issuer, or regulator, cannot resolve or affect a dispute or consumer rights, cannot assess aggregate spend or velocity, and cannot show that the supplied mandate remained unrevoked when the transaction was authorized.
Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.
Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.
Every delegation scheme we could find checks the chain downward. Each link is signed, each link narrows, the math works. None of them check the very first step: whether the party at the root held the authority it delegated. UCAN says so in its own spec, that a cryptographically valid chain can still be semantically invalid and the executor has to go look. This receipt is that look, made into evidence. It reads an entitlement source and a grant instrument, verifies the grant instrument's own signature, and answers whether the grant sits inside the entitlement on scope, on amount, and on time. A third party who is neither the granter nor the grantee has to sign the check.
What it does not do. It does not judge whether the entitlement source is telling the truth. It proves the grant did not exceed what that source said the granter had.
Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.
Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.
Authority is supposed to shrink as it passes down a chain. In practice a hop can quietly add a scope, raise a cap, or push an expiry out, and the chain still verifies because every signature is good. This receipt compares each hop against the one above it on scope, amount, and time, and reports whether the chain only ever narrowed. It reads the whole chain, not just the last link.
What it does not do. It does not decide whether the top of the chain should have had that authority. That is the qualification receipt. This one only answers whether anything widened on the way down.
Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.
A single delegated mandate can be presented to two acceptors who have no shared ledger and no reason to trust each other's books. Each one sees a valid mandate and each one is right about its own slice. This receipt fixes each acceptor's consumption as a commitment, then reports the cumulative relation of all of them against the mandate constraint, so the total is checkable without either acceptor handing the other its transaction history.
What it does not do. It does not reconcile the two ledgers, reverse anything, or say which acceptor should have declined. It says whether the sum of what was consumed stayed inside the constraint or went past it.
Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.
Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.
Every run above posts a verified request body from this domain to the open verify route and prints what came back. The example receipts are signed with published example keys, so verify reports key_trust example_registry. That is on purpose. Nothing on this page is a production issuance, a customer record, or an endorsement. Patent Pending.
Each link opens that entry in the canon implementation explorer, where its schema, mint route, open verify route, auth requirement and implementation state are stated. The state shown here is read from the same registry file the explorer renders from, so the two cannot drift apart. Nothing here implies a customer, a deployment or an endorsement.