NEO WORLDProof series
Case 01 · Commerce workflow evidence

The Million-Receipt Proof

How one reviewed synthetic commerce workflow became a million-receipt proof for the NEO World economy.

1,000,000source receipts verified
25,000signed workflow commitments
30 mincontinuous source run
0quarantined submissions

Start with a real commerce workflow

The test used CommerceAgentBench task api-amazon-margin-floor-audit. One execution performs 40 HTTP operations: reading listings, probing marketplaces, previewing and applying qualified price updates, creating and submitting a feed, polling its result, and generating a settlement report.

The workflow was executed 25,000 times with a unique identity and hash-linked receipt chain for every run. Expected 404 probes remained visible as honest receipts; they were not hidden as failures.

Can one million commerce receipts cross the external NEO rail—and still be independently proven afterward?

One continuous evidence path

01 · GenerateCommerceAgentBench

40 receipts close into one meaningful workflow commitment.

02 · AdmitNEOfX2

25,000 Ed25519-signed envelopes enter a dedicated external boundary.

03 · PreserveNEOF1

Exact commitments become durable append-only frames.

04 · FoldNEOCL

Gap-free delivery forms verifiable roots.

05 · CloseNEOseal

The measured proof horizon is closed.

06 · WitnessNEOva2

62 folds are confirmed at positions 90,006,547–90,006,608.

Reconstructed, not merely reported

A separate fail-closed verifier replayed the million source receipts, verified every signature and batch link, re-read the append-only journal, checked the gap-free NEOCL span, verified the endpoint Merkle proofs and confirmed final NEOva2 readback. Every required check returned PASS.

What this proves

  • One reviewed commerce workflow can be repeated at sustained scale.
  • All 25,000 commitments remained intact across the full external rail.
  • Delivery was gap-free and all 62 folds were witnessed on NEOva2.
  • An independent verifier can reconstruct the result afterward.

What this does not claim

  • The test used one pinned synthetic scenario, not 25,000 different businesses.
  • It measures evidence architecture—not commercial outcome quality.
  • The large source database remains on the test host, bound by its published hash.
  • Scenario diversity is the next measurement dimension.

Don't trust the counter.
Follow the evidence.

The public case, test branch and upstream commerce harness provide the path from claim to executable source.

← All proof casesNEO World · 01 / 03Next · Two-Hour Proof →