General use
Buying, retiring or claiming for any voluntary purpose.No integrity events, a fresh registry proof, and the profile's rules are met.
See what the contracts say to a claimant, a seller and an integrator right now, next to what happened when the same calls were sent on-chain.
No integrity events, a fresh registry proof, and the profile's rules are met.
This profile needs a host-country letter of authorization, and none is on the proven registry record.
No integrity events, a fresh registry proof, and the profile's rules are met.
No integrity events, a fresh registry proof, and the profile's rules are met.
Every event ever opened for this project, oldest first. AI-proposed events are evidence for review, not findings: they act only after a reviewer countersigns or the dispute window closes.
Satellite imagery shows forest loss inside the project boundary (a possible reversal). Confidence 50%.
From issuance to claims, read from Verra's registry on 21 Oct 2026 and from the contracts now.
| Beneficiary, as published by the registry | Tonnes | Retirements |
|---|---|---|
| Not disclosed | 753,898 | 286 |
| PetroChina International Company Limited (“PCI”) | 206,166 | 4 |
| Saipem Group | 100,000 | 1 |
| Qantas Airways Limited | 54,449 | 6 |
| Shell | 31,650 | 2 |
| Xero Limited | 21,341 | 5 |
Names appear exactly as the registry publishes them; personal ID numbers are redacted. A name here says nothing about that buyer's conduct.
The project's status comes from Verra's own API response, proven with a zkTLS proof and verified inside RegistryOracle. The request it accepts is pinned byte for byte: URL, method, headers, body and the extraction pattern.
Beneficiaries shown as published by the registry.
Sentinel-2 dry-season composites inside the boundary filed with the Verra registry, 2023 against 2025. The detector flags canopy loss pixel by pixel; Hansen Global Forest Change is the independent yardstick.
2023 dry season
2025 dry season
Detected lossPre-registered rule met (validFraction >= 0.5 AND (lossPct >= 1.0 OR lossHa >= 1000)): Forest loss detected, 1.21% haircut proposed at 50% confidence.
Detected 1,508 ha against 202 ha in Hansen; 90.1% of forest pixels had clear observations.
Everything sent on-chain about this project, including the calls the contracts refused on purpose.
| What | By | Result | Transaction |
|---|---|---|---|
| Registry status proven 1477 REGISTERED RegistryOracle.submitProjectStatus | watcher | Accepted | 0x7a7459d9…e5bc2f93 |
| Attestation REVERSAL_SUSPECTED 1477 121bps EAS.attest | oracle sat | Accepted | 0x2b6f607d…8d3e993f |
| Dispute opened independent Hansen GFC loss does not corroborate the flagged pixels IntegrityRegistry.dispute | reviewer | Accepted | 0x79005a45…d426c7a7 |
| Arbiter decision rejected: haze speckle in a cloudy composite, no clearings in the chips IntegrityRegistry.resolve | arbiter | Accepted | 0xf7f7ff79…11076a19 |
| Claim accepted after the satellite alarm was rejected ClaimsLedger.registerClaim | relayer | Accepted | 0x21b98669…2b9181a4 |