Real events, real dates

FOBXX published on both chains

Verified signed FOBXX reports were retained with publication transactions on X Layer mainnet and testnet. Its current status above comes from the source manifest and is admitted only because the build gate finds those reports.

The site becomes a product, and gains its Terminal

Every page is now rendered from one canonical data source — four external audits in a row had caught hand-typed facts gone stale, so the counts, addresses and sequences on this site are generated, not maintained. The Policy Terminal reads both registries live, runs gate checks, simulates and executes guarded actions through a connected wallet, and verifies a signed bundle locally in the browser.

Every approval decision is now signed

The approval ledger moved to version 2: each approval and decline carries the approver's EIP-712 signature over the exact compiler proposal it decides, verified on every load, declined entries included. Signatures are timestamped at signing — 2026-08-19 — never backdated to the decisions they attest. Reports published earlier commit to the version-1 unsigned ledger and their bundles verify unchanged, forever.

First CONFIRMED states, and Registry v2 holds its first attestations

The NAV this system refused on 2026-08-18 survived its qualifying interval unchanged and became the first CONFIRMED state, with both policy verdicts published under their own registry keys on both chains. The current confirmed-report count is shown on the reports page, not rewritten into this dated event. Registry v2 went live on testnet and mainnet the same day and holds all four policy attestations, relayer-submitted with the publisher recovered on chain from EIP-712 signatures. The consumer gate that had refused the asset flipped to (true, "allowed") with no control changed, and permitted/refused GuardedAction pairs put both outcomes on chain.

Two corrections published, one on each chain

Sequence 1 on both chains was signed carrying an operational event of RECONFIRMED. A first publication has nothing to reconfirm, so that asserted a history which did not exist. The same reports also carried a limitation reading “This local-only report” while sitting on a public chain. Both statements were wrong in the signed bytes, and signed bytes cannot be edited.

They were restated through the registry’s publishCorrection entry point: testnet sequence 3 (block 38617112) and mainnet sequence 2 (block 68307118). A correction restates; it does not re-decide. Each reproduces its original’s control_set_root, evidence_root and approval-ledger digest exactly — the builder refuses to sign one that cannot — and the reported state stays UNVERIFIABLE on both. The limitation count is unchanged, because a correction that quietly dropped a caveat would be using an honesty mechanism to say less.

epochSequence still points at each epoch’s first publication, so a correction restates an epoch without opening one. One consequence is worth stating plainly: on testnet the latest report is now sequence 3, which is an older epoch than sequence 2. That is what append-only correction looks like.

Testnet sequence 2 was left alone. Its RECONFIRMED is genuine.

Two retrieval routes re-measured

FOBXX’s regulator route was re-verified live and is bounded: data.sec.gov returns HTTP 200 in 29,354 bytes with 68 N-MFP3 filings, the latest filed 2026-08-06 for period 2026-07-31. Its status had been reported as unavailable because the daily issuer feed returns 403, which was allowed to describe the whole asset. FOBXX carries monthly controls on filed evidence.

USDY was found to have a second, bounded research surface in addition to its unbounded attestation archive. That finding did not lift the suspension: the attestation route still has no bounded retrieval, and there is no parser, fixture, approved control or adapter. The bounded daily page was measured at 1,544,898 bytes. It is not approved: the issuer’s published collateralizationRatio of 105.29 does not reconcile with its own published total and outstandingValue, which give 100.87%. That discrepancy was found in audit, after this project had already written that the figures were internally consistent. They are not, and no control can be written on them until the field definitions come from the issuer.

Control set revised

A five-control set was approved for USTB, with five candidates declined. It applies only to sequences published after sequence 1. Sequence 1 remains bound to the eight-control approval ledger it was signed under — its digest is inside the signed report, and this site never renders sequence 1 against the newer set.

First publication

USTB sequence 1, epoch ustb-2026-08-17, published to the registry on X Layer testnet at block 38526525. The reported state is UNVERIFIABLE: five of eight controls satisfied, three unevaluable. Observed at 2026-08-17T16:48:32.075003Z.

Eight controls approved for USTB

A human approved the eight-control set proposed from Superstate's own documents. Two candidates — holdings-line-items-present and ustb-fund-identifier-stable — were declined, with the reasons recorded verbatim in the approval ledger.

Coverage triage

USDY suspended: its only attestation source is a 260,431,605-byte archive with no bounded retrieval, so it cannot be scheduled. FOBXX demoted: the daily issuer feed is unreachable from a production host, so no daily-liveness or issuer-versus-regulator control may be claimed until repeatable access is verified. OUSG opened as the candidate second daily asset.

USTB controls take effect

All eight USTB controls carry effective_from 2026-08-14. From this date the controls are in force; no end date is set.