Paste identifiers. You get one of three answers per project, and the difference between them is the point. The register holds records on it; it sits inside the coverage with nothing recorded against it; or it falls outside the coverage, in which case the register says nothing either way. Nothing held is a statement about the record, not about the project.
One per line, or comma separated. Registry identifiers in any common form: VCS1477, VCS 1477, Verra 1477, ACR0360, CAR1004, GS1368, CAFR0070, CALS5040, ERF109627, ART101.
A nil answer means nothing unless you know what was looked at, and how deeply. The Carbon Loss Register, edition six: records as at 3 September 2026, coverage as at 31 August 2026. 15,080 projects inside the coverage; 250 records held across 179 projects. Two registers are covered completely, from the regulator's own loss table. Five are covered partially, from the Berkeley database as it records them, and a nil in those means less.
| Verra (VCS) | partial | listed projects from the Berkeley voluntary registry database v2026-06; loss records assembled project by project from public documents, because Verra publishes no buffer pool ledger |
| Gold Standard | partial | listed projects from the Berkeley voluntary registry database v2026-06 |
| American Carbon Registry | partial | listed projects from the Berkeley voluntary registry database v2026-06, with registry confirmation of the cancellation layer |
| Climate Action Reserve | partial | listed projects from the Reserve's own project and issuance exports |
| ART TREES | partial | listed projects from the Berkeley voluntary registry database v2026-06 |
| California (CARB) | complete | the California compliance programme in full: the regulator publishes project level reversal, invalidation and buffer data in structured form |
| Australia (Clean Energy Regulator, ACCUs) | complete | the Australian ACCU register in full, including the relinquishment layer |
Only openly publishable sources are searched. 15 projects reach us solely through a source licensed for internal use; those abstain rather than return a nil, because we hold something and may not publish it either way. And a project absent from the snapshot of current listings is not a project that is not real: 67 of the projects in this register are archived, withdrawn or removed, and every one of them is missing from that snapshot for exactly that reason. For an identifier you type that the snapshot does not carry, the register cannot tell from the identifier alone whether it was removed or never issued, and says so rather than guessing.
Most of any real book comes back empty. That is the normal case, and every profession that had this problem before us solved it the same way: by naming the states and publishing the scope.
| Audit | Reports in the negative: nothing came to the practitioner's attention, within a stated scope. Explicitly not an audit opinion, and the standard makes them say so. |
| Insurance | The loss run: a carrier's record of claims against a named insured over a stated period. A clean loss run is a routine underwriting input. Nobody reads it as proof that nothing happened. |
| Consumer credit | Separates a scored file, a thin file and a credit invisible, because collapsing them penalised the invisible for the delinquent's record. |
| Catastrophe loss | Publishes inclusion thresholds, so an empty period reads as out of scope rather than as calm. |
| Drug safety | Supplies the caution we take seriously: spontaneous reporting is chronically incomplete, and absence of a signal is not evidence of safety. Our position is worse, because most voluntary reversals sit in prose inside monitoring reports and no table counts them. |