LNES-05 · Base L2 · Operational
Vanguard · SGX Sealed · Active
Avg. audit time · <8s
LNES-05 · Ghost-Witness Protocol

Your AI agent
should not be
allowed to lie.

Ghost-Witness detects logical contradictions in any AI agent conversation and generates a cryptographic Certificate of Logical Consistency — anchored permanently on Base L2.

$0.005 USDC per audit
Base L2 settlement
No subscription
ghost-witness · live detection
# Agent conversation piped to Ghost-Witness
POST /api/ghost-witness/audit

"conversation": [
  { role: "agent",
    content: "Price is $49.99. Ships in 2 days." },
  { role: "agent",
    content: "Delivery takes 5–7 business days." }
]
Audit complete · 28ms
✗ CRITICAL · Shipping contradiction detected ✗ HIGH · Temporal inconsistency (2 days ≠ 5–7 days)
consistent: false · score: 0.31
CLC sealed · Base L2 · $0.005 USDC
hash: 0x9dd05d865b826a90b61145e...
99.9%
Contradiction detection accuracy
<10s
Audit and settlement time
$0.005
USDC per audit — flat rate
CLC validity — permanent on-chain
How it works

From conversation
to certificate
in four steps.

Ghost-Witness is a pipeline, not a product. It sits downstream of your existing AI agent and verifies its outputs automatically.

01
Pipe the conversation
Send any AI agent conversation log to the Ghost-Witness API. Supports WhatsApp, Zendesk, Shopify, or any custom chatbot platform. Optional PII stripping before processing.
POST /audit
02
Claims are extracted
The system parses every agent statement and identifies verifiable claims: prices, timelines, medical assertions, legal claims, statistics, and capability promises.
8 claim types
03
Vanguard checks consistency
SEI Vanguard — running inside an Intel SGX enclave with zero data retention — analyzes the extracted claims for internal contradictions, implausibility, and hallucination risk.
SGX sealed
04
CLC sealed on Base L2
The verdict is anchored on Base L2. The Certificate of Logical Consistency is permanently verifiable at exergynet.org/clc/{hash} by anyone — your regulators, customers, or auditors.
LNES-05 · Base L2
Use cases

Any agent that
makes claims needs
a witness.

If your AI agent talks to customers, patients, or counterparties — Ghost-Witness is your permanent audit trail.

🏥
Biotech & Clinical
AI agents making clinical recommendations, quoting trial results, or describing drug interactions. Any factual inconsistency is a liability. Ghost-Witness produces a CLC for every automated patient interaction.
FDA compliance Regulatory audit CRO verification
💬
WhatsApp Business
Meta's Business Agent platform powers millions of customer conversations per day. When an agent quotes prices, delivery times, or product specs — Ghost-Witness detects contradictions before they become refunds or lawsuits.
Shopify Zendesk WhatsApp Business API
⚖️
Legal & Compliance
AI agents advising on contracts, rights, or regulatory requirements. A single inconsistent legal claim can void an agreement. Ghost-Witness provides timestamped evidence that the agent's output was verified at delivery.
Contract review Compliance advisors Regulatory bots
📈
Financial Services
AI agents quoting rates, returns, or risk levels. Contradictory financial claims trigger regulatory action. Ghost-Witness creates an immutable record that the agent's representations were internally consistent.
Rate quotation Portfolio advice Insurance underwriting
🛒
E-Commerce Agents
Product description, pricing, and availability agents. Price contradictions cost retailers in chargebacks and trust. Ghost-Witness catches shipping-time contradictions, price inconsistencies, and availability conflicts.
Product bots Customer service Returns automation
🤖
M2M Agent Networks
When AI agents contract with other AI agents, truth verification is the trust layer. Ghost-Witness provides neutral third-party attestation that a counterparty agent's representations were consistent before settlement executes.
ElizaOS Claude agents Agentic workflows
Pricing

Pay per audit.
No surprises.

USDC settlement on Base L2. No monthly fees. No hidden costs. Every audit produces a permanent, publicly verifiable CLC.

Developer
Free
to get started
  • 10 free audits / month
  • All claim types supported
  • CLC issued per audit
  • Base L2 settlement
  • Public CLC viewer
Request API key
Enterprise
Custom
volume pricing · SLA
  • Volume discount >10k/mo
  • Dedicated Vanguard node
  • Custom claim taxonomy
  • White-label CLC viewer
  • Compliance report export
  • Priority support
Contact us
API Quickstart

Three calls
to production.

REST API. USDC settlement handled automatically. No blockchain knowledge required.

1
Get your API key
Email intel@exergynet.org or request via the portal. Instant provisioning. Fund your wallet with USDC on Base L2 for pay-per-audit billing.
2
Send a conversation
POST to /api/ghost-witness/audit with your conversation array. Include conversation_id for tracking. Optional strip_pii for automatic PII removal.
3
Receive the CLC
The response contains the full Certificate of Logical Consistency with flags, severity levels, and the Base L2 receipt. Store the audit_hash for public verification.
4
Verify publicly
Anyone — your customer, your regulator, your auditor — can verify the CLC at exergynet.org/clc/{audit_hash}. No login required. Permanently on-chain.
API reference
# 1. Audit a conversation curl -X POST \ https://dt.portal.exergynet.org/intel/api/ghost-witness/audit \ -H "Content-Type: application/json" \ -H "x-api-key: YOUR_GW_API_KEY" \ -d '{ "conversation_id": "session-001", "platform": "whatsapp", "strip_pii": true, "conversation": [ {"role":"agent","content":"Price is $49.99."}, {"role":"agent","content":"Price is $99.99."} ] }' # Response { "status": "complete", "clc": { "consistent": false, "flags": [{ "issue": "Price contradiction", "severity": "critical" }] }, "receipt": { "tx_hash": "0x7a3f...e9d2", "cost_usdc": 0.005, "clc_url": "exergynet.org/clc/..." } } # 2. Verify publicly (no auth required) curl https://dt.portal.exergynet.org/intel/api/ghost-witness/clc/{audit_hash}
CLC Public Verifier

Verify any
certificate.

Enter an audit hash to verify a Certificate of Logical Consistency. No login required. Results are pulled directly from Base L2.

What a verified CLC means
A green CLC means the agent conversation was analyzed by SEI Vanguard and no logical contradictions were detected. The verdict is permanent and tamper-proof on Base L2.

A red CLC means contradictions were detected — the specific flags and severity levels are recorded on-chain for permanent reference.
The CLC does not verify factual accuracy against the real world — only internal logical consistency. Factual verification requires external data sources.