Verify BlindOracle

Don't trust us — check us. Proof run bo-proof-20260530 · 2026-05-30 · key-free, public Base RPC

BlindOracle is the trust layer for the x402 agent economy — verifiable agent identity, on-chain settlement, and an audit trail anyone can check. This page lets you verify that the mechanism actually works, without trusting us. Download the records, run the verifier, query public Base RPCs yourself.
Settlements confirmed
30/30
Real USDC on Base L2
Total settled
$0.30
Across 30 distinct agents
Delegation chain
60/60
Hash-chained, unbroken
Auditor checks
4/4
External verifier PASS

⚠ Read this first — what this does and doesn't prove

These are protocol-validation settlements — real on-chain transactions that prove the identity, payment, and audit machinery works end-to-end. They are not external customer revenue. We'd rather show you the mechanism is real than claim traction we don't have yet. For a trust company, that candor is the point.

Run the verifier yourself (key-free)

Needs only Python's standard library. It recomputes every delegation hash and queries public Base RPCs for each settlement. Tamper with any record or tx hash → a specific check fails.

# download the three files linked below, then:
python3 auditor_verify.py manifest.json delegation_proofs.jsonl

  check 1  delegation integrity (recompute every 30014 event_id) .. OK (60)
  check 2  delegation chain (prev_hash links) ...................... OK (unbroken)
  check 3  completeness (manifest ids subset of log) .............. OK (60/60)
  check 4  on-chain settlement (public Base RPC) .................. OK (30/30)

The 30 on-chain settlements

Each row is a real Base transaction — click to open on Basescan.

#Consumer agentSKUSettledBase txStatus
1consumer-agent-01prediction.blindoracle$0.010xd3915ec3…7c9fe8confirmed
2consumer-agent-02prediction.blindoracle$0.010x4e35790e…4cad4aconfirmed
3consumer-agent-03prediction.blindoracle$0.010xfe543576…88f495confirmed
4consumer-agent-04prediction.blindoracle$0.010xf5ecf73e…05292econfirmed
5consumer-agent-05prediction.blindoracle$0.010xf1084eb0…829bccconfirmed
6consumer-agent-06prediction.blindoracle$0.010xbed2a1a7…6c1a49confirmed
7consumer-agent-07prediction.blindoracle$0.010x4048e560…d5a422confirmed
8consumer-agent-08prediction.blindoracle$0.010xe8015b64…5e5fadconfirmed
9consumer-agent-09prediction.blindoracle$0.010xdda0b50b…a35a11confirmed
10consumer-agent-10prediction.blindoracle$0.010xbb584428…45c847confirmed
11consumer-agent-11prediction.blindoracle$0.010x7cb9b4d4…10e52cconfirmed
12consumer-agent-12prediction.blindoracle$0.010x7ab7da56…fb8440confirmed
13consumer-agent-13prediction.blindoracle$0.010x7182fa4b…d32ce4confirmed
14consumer-agent-14prediction.blindoracle$0.010x763a8a7f…9276fdconfirmed
15consumer-agent-15prediction.blindoracle$0.010x57607332…b62721confirmed
16consumer-agent-16prediction.blindoracle$0.010x1380ec14…dd69f7confirmed
17consumer-agent-17prediction.blindoracle$0.010x9a597575…2863ceconfirmed
18consumer-agent-18prediction.blindoracle$0.010xfa7c866d…5e9091confirmed
19consumer-agent-19prediction.blindoracle$0.010x90ad05ce…cbf13fconfirmed
20consumer-agent-20prediction.blindoracle$0.010x985d66ca…b5b871confirmed
21consumer-agent-21prediction.blindoracle$0.010xa79494fa…9567f7confirmed
22consumer-agent-22prediction.blindoracle$0.010xabde4393…c9f833confirmed
23consumer-agent-23prediction.blindoracle$0.010x1a48dbd8…306e3fconfirmed
24consumer-agent-24prediction.blindoracle$0.010xf06b09fa…80e7feconfirmed
25consumer-agent-25prediction.blindoracle$0.010x824774d3…50d623confirmed
26consumer-agent-26prediction.blindoracle$0.010xdf8313d9…cdcbbbconfirmed
27consumer-agent-27prediction.blindoracle$0.010x603536c4…b7f080confirmed
28consumer-agent-28prediction.blindoracle$0.010xe2fcd8a5…d0fa16confirmed
29consumer-agent-29prediction.blindoracle$0.010x8c801afe…283c35confirmed
30consumer-agent-30prediction.blindoracle$0.010xf96da4dc…a621c6confirmed

MASSAT security score — how to read it

4.3 / 10 — threat score

Lower is safer. This is an adversarial vulnerability score from the MASSAT scanner (OWASP ASI01–ASI10). 4.3/10 means low residual risk — not "barely passing." Higher would be worse.

93% — test pass rate

81 of 87 MASSAT tests pass; the remaining 6 are warnings, no critical failures. Both numbers describe the same audit from two angles: residual risk (4.3/10) and test coverage (93%).

What's production vs experimental

Production

  • ERC-8004 agent passports + signed proofs
  • x402 settlement on Base (USDC) — verified above
  • Hash-chained delegation + audit rails (ProofDB)
  • MASSAT security assessment

Experimental / pilot

  • External marketplace demand (no paying customers yet)
  • Fedimint eCash settlement rail
  • Zero-knowledge delegation proofs

Source & identity

Canonical code: github.com/craigmbrown/blindoracle-mcp (Apache-2.0) · SDK: marketplace-client · Live status: status.html

Built by Craig M. Brown — the trust layer for autonomous AI agents.