IDENTITY → INTENT → ACTION → PROOF

Nexus Shield Proof Center

Verify every agent outcome with cryptographic proof — reproducible harness benchmarks (trajectories & evaluations) and production UAR receipts. Evidence bundles cover the full evaluation lifecycle, not blocked actions only.

Open Live Trust Hub

Proof Center Playground

No registration

Simulate an agent tool call and inspect a live SHA-256 Universal Action Receipt — before/after state hashes and tamper-evident SHA-256 receipt core (hash integrity only).

Receipt JSON will appear here after simulation.

Nexus Shield v2.0

Action Verification Center

Inspect AAR receipts, simulate blast-radius what-if removal, and audit multi-agent delegation chains.

1 — Action Control

Identity, authority, intent, and policy gates before consequential tools execute.

Blast Radius & Exposure Matrix

MCP tools mapped to data boundaries — interactive what-if removal recalculates operational risk.

Risk score

62.4

HIGH

3 tools · 4 resources

3 write exposures

Tool / Resourceinvoice_1024ledger_mainvendor_api
mcp_payments_bridge—WRITEWRITE
read_invoiceWRITE——
stripe_create_transferWRITEWRITE—

3 tool(s) touch 4 resource(s); 3 write exposure(s); operational risk tier HIGH (score 62.4).

2 — Outcome Verification

Cross-check HTTP success against DB, ledger, and ERP state — detect false success (UNVERIFIED).

AAR 2.0 Receipt Inspector

Universal Action Receipts — structural JSON, SHA-256 evidence hash, Ed25519 signature.

VERIFIED

evidence_hash

sha256:d90704a8be96a9c1e12fa63b8529077f528d91298eec4b15f244f1231ef84649

Ed25519 signature

sig_nexus_ed25519_7714853662812f6f844eca644e9777d324b20c582d050904

Structural JSON
{
  "$schema": "https://nexusshield.ai/schemas/aar-v2.json",
  "receipt_id": "aar_9f8b2c4e7d1a3f6e",
  "timestamp": "2026-10-03T21:45:00.124Z",
  "agent": {
    "identity": "finance-agent-04",
    "passport_id": "pas_live_88192a",
    "owner": "Finance Department"
  },
  "intent": {
    "raw_prompt": "Pay invoice #1024 to vendor Acme Corp for $4,500",
    "parsed_intent": "EXECUTE_PAYMENT",
    "target_resource": "invoice_1024"
  },
  "authority": {
    "allowed_scopes": [
      "invoices:read",
      "payments:write"
    ],
    "financial_limit": 5000,
    "delegation_depth": 1,
    "verified_by_graph": true
  },
  "policy": {
    "policy_id": "FIN-PAY-07",
    "evaluation": "ALLOW",
    "risk_score": 0.12
  },
  "execution": {
    "tool_called": "stripe_create_transfer",
    "request_payload": {
      "amount": 450000,
      "currency": "usd",
      "destination": "ac_123456789"
    },
    "api_response": {
      "status_code": 200,
      "raw_body": "{\"id\":\"tr_1OxyZ2\",\"status\":\"succeeded\"}"
    }
  },
  "outcome_verification": {
    "status": "VERIFIED",
    "verification_method": "DB_STATE_AND_LEDGER_CROSS_CHECK",
    "state_before": {
      "invoice_status": "PENDING",
      "ledger_balance": 150000
    },
    "state_after": {
      "invoice_status": "PAID",
      "ledger_balance": 145500
    },
    "discrepancy_detected": false
  },
  "cryptographic_proof": {
    "evidence_hash": "sha256:d90704a8be96a9c1e12fa63b8529077f528d91298eec4b15f244f1231ef84649",
    "signature": "sig_nexus_ed25519_7714853662812f6f844eca644e9777d324b20c582d050904"
  }
}

3 — Cryptographic UAR

SHA-256 evidence hashes, Ed25519 seals, and delegation audit trails for every governed attempt.

Delegation Hierarchy

Human → agents → API tools. Violations and privilege escalation blocks appear in the audit trail.

human-roothuman

Scopes: invoices:*, payments:write · limit $10,000

finance-agent-04delegatee

Scopes: invoices:read, payments:write · limit $5,000

finance-agent-subdelegateeBlocked escalation

Scopes: invoices:read · limit $1,000

DELEGATION_VIOLATION: scope payments:write not in effective delegation

Audit trail

  • {"event":"GATEWAY_DELEGATION_DENIED","agent_id":"finance-agent-sub","reason":"scope payments:write not in effective delegation for finance-agent-sub"}
Public Proof Center

Nexus Shield Proof Center

Open-source harness benchmark metrics (Harness test suite (open-source baseline fixture — not live fleet telemetry)) — separate from your production UAR ledger on the data plane.

*Tested across 500+ MCP attack scenarios & multi-agent execution graphs. Open-source benchmark harness available on GitHub — OWASP GenAI / Agentic tags on every scenario.

Harness · Agent Safety

Harness agents evaluated (baseline fixture)127
Harness trajectory steps evaluated48,291
Dangerous Actions Blocked3,817 / 3,842 (99.3%)

Harness · Scenario accuracy

Intent/action misalignment (harness scenarios)98.7%
Tool misuse & hijack (harness)99.1%
Privilege escalation blocks (harness)97.8%
False positive / negative (harness)Benchmark test suite evaluated (low false-positive rate under controlled test vectors — not a production SLA)

Harness · Runtime latency

p502.4 ms
p954.8 ms
p996.1 ms

measured in Nexus benchmark harness test environment

Attack Scenarios & Evidence

MCP Attack Scenarios Tested500+
Harness evidence bundles (evaluated trajectories)48,291
Harness runs emit tamper-evident SHA-256 bundles per evaluated trajectory step (ALLOW, BLOCK, READ_ONLY, REQUIRE_APPROVAL). Counts match trajectory steps evaluated, not blocked actions alone — see harness/fixtures/sample-output.json.

Prove It — Live Attack → Block → Evidence

Interactive Verification Demo

Agent Attack Attempt

Unauthorized shell command: rm -rf / — PII exfiltration via read_customer_records

1. Agent Attack Attempt
2. Runtime Intercept
3. Tamper-evident UAR
Inspect Full Proof Center Data