Sentinel

Sentinel is an AI-powered Anti-Money Laundering (AML) and compliance oracle that brings bank-grade regulatory enforcement to DeFi protocols. Built on the Chainlink Runtime Environment (CRE), Sentinel automates the entire AML compliance stack — from real-time transaction screening to sanctions list synchronization — without human intervention. At its core is USDA, an AML-compliant stablecoin where every transaction is screened against global sanctions databases, scam registries, and AI-powered risk models before execution.

How It Works — AML-First Architecture

Sentinel's 5 autonomous workflows form a comprehensive AML compliance layer:

1. AML-Compliant Minting (EVM Log Trigger)

• Every USDA mint begins with multi-layered AML screening
• Depositor addresses checked against ScamSniffer database and sanctions lists
• 3-source price consensus prevents oracle manipulation for money laundering
• xAI Grok evaluates transaction patterns for suspicious activity
• Only DON-signed, AML-cleared transactions execute

2. Real-Time AML Transaction Monitoring (EVM Log on Transfer)

• Every USDA transfer is screened in real-time before settlement
• Recipients verified against:
  • GoPlus Security API (SlowMist + ScamSniffer aggregation)
  • ScamSniffer GitHub blacklist
  • Sanction Source (sanctions lists)
• xAI Grok analyzes transfer patterns, amounts, and counterparties for AML red flags
• Automatic freeze of suspicious addresses via DON-signed reports
• Non-blocking pause mechanism prevents fund movement during investigation

3. Daily Sanctions List Synchronization (Cron)

• Aggregates global AML data sources inside CRE TEE:
  • Sanction Source (DEPENDS ON TYPE OF STABLECOIN)
  • GoPlus Security API (SlowMist aggregation)
  • ScamSniffer Database
• Merges and deduplicates blacklists
• Computes Merkle root for efficient on-chain verification
• Updates PolicyEngine daily with latest AML requirements

4. Reserve-Based Volume Limits (Cron)

• Prevents money laundering through volume manipulation
• AI-adjusted limits based on real-world bank reserves
• Automatic restriction when reserve ratios indicate suspicious activity

5. Sentinel Guard — Pre-Hack AML Intervention (HTTP Triggered)

• Detects complex money laundering patterns (flash loans, layering)
• Investigates via GoPlus API (honeypot, blacklist, mintable analysis)
• xAI Grok decides PAUSE/MONITOR for AML violations
• Broadcasts to EmergencyGuardianDON for immediate protocol protection

What AML Problem It Solves

The DeFi AML Gap:

Traditional DeFi protocols lack automated AML compliance, exposing them to:

• Sanctions violations (OFAC, UN, EU lists)
• Money laundering through anonymous wallets
• Scam and fraud fund circulation
• Regulatory enforcement actions

Sentinel's AML Solution:

• Automated Sanctions Screening: Every address checked against global sanctions lists before any transaction
• Real-Time Transaction Monitoring: Continuous surveillance of all transfers for suspicious patterns
• Immutable Audit Trail: DON-signed reports provide regulatory-compliant evidence
• Pre-Execution Freezing: Suspicious funds stopped before they move, not after
• Regulatory Readiness: Built-in compliance reporting for FATF, MiCA, and banking regulations

Sentinel makes DeFi protocols regulator-ready by embedding bank-grade AML directly into the blockchain — protecting protocols from compliance violations while maintaining decentralization.

Core Infrastructure — Chainlink Runtime Environment (CRE)

Sentinel runs entirely within the Chainlink Runtime Environment (CRE), a serverless, decentralized compute layer powered by Chainlink nodes. This provides:

• Trusted Execution Environment (TEE) — All API calls, AI processing, and decision-making happen inside secure enclaves
• DON Attestation — Every workflow output is cryptographically signed by the Decentralized Oracle Network
• Confidential HTTP — External API requests (LLM requests, bank reserves) are fetched without exposing sensitive data

External API Integrations

• GoPlus Security API — Honeypot, blacklist, mintable token checks (Includes sanctions as well)
• ScamSniffer Database — Known scam addresses and GitHub blacklist
• Sanction Source (additional sanction source) — Government sanctions lists (Used Limited API for my demo - did not use the full list)
• First PlaidyPus Bank API — Real-world reserve verification (Sandbox API)
• xAI Grok — AI decision engine for risk assessment

AI & Decision Layer

• xAI Grok runs inside TEE for tamper-proof AI decisions
• Analyzes transaction patterns, counterparties, and risk scores
• Outputs PAUSE/MONITOR/FREEZE/ADJUST... decisions with DON signatures