Autonomous AI security agent on Base — powered by Bankr LLM Gateway. Every scan, every trust check, every rug detection is reasoned through Claude Opus 4.6 via Bankr before your agents pay a single wei.
Each operates independently, coordinated by the IRONCLAD orchestrator.
Fork-simulates every x402 payment in <200ms. Detects unexpected transfers, unlimited approvals, and gas anomalies before settlement.
scan-transactionKYA trust scoring. Wallet history, code provenance, social links, World ID, and on-chain reputation.
check-agent-trustContinuous monitoring for rug pulls, honeypots, sandwich attacks, proxy upgrades, and flash loan exploits in real-time.
detect-rugInstant alerts via X, XMTP and webhooks. SEV-0 threats trigger auto-response: revoke approvals, rescue funds.
monitor-contractPost-exploit analysis and cross-chain fund tracing. Full incident reports with attacker profiles and IOC data.
trace-fundsIRONCLAD doesn't guess. Every scan, every rug check, every trust score is reasoned through Claude Opus 4.6 via Bankr's LLM Gateway. Raw heuristic signals are collected on-chain, then sent to Claude for deep threat analysis — returning verdicts with confidence scores and human-readable reasoning.
Bankr provides the infrastructure: LLM access, agent wallets, skill marketplace, and token launch — so IRONCLAD can focus on what it does best: protecting your agents.
// 1. Collect on-chain signals
signals = await simulate(tx)
signals += await analyzeBytecode(contract)
signals += await checkApprovals(simulation)
// 2. Send to Claude via Bankr LLM
verdict = await bankr.llm({
model: "claude-opus-4-6",
prompt: `Analyze these ${signals.length} signals
for transaction to ${tx.to}`,
gateway: "llm.bankr.bot/v1"
})
// 3. Return AI-reasoned verdict
→ { verdict: "BLOCK", confidence: 0.94,
reasoning: "Unlimited approval to
unverified proxy contract..." }
Agent submits x402 payment. IRONCLAD intercepts via middleware or XMTP command.
Fork Base chain state via Alchemy. Run transaction in sandbox. Analyze state changes.
Claude Opus 4.6 via Bankr analyzes signals. Returns SAFE / WARN / BLOCK with score.
Safe? Proceed. Threat detected? Block + alert. Critical? Auto-revoke and rescue.
$ICLAD is the access and alignment token for IRONCLAD. Hold, stake, and unlock progressively deeper security capabilities.
Skills + Token
Payment Security
Agent Messaging
Agent Identity
L2 Chain
AI via Bankr