What is SentinelGuard?
An autonomous exploit detection and circuit-breaker layer for Solana DeFi — built to act in under one slot.
The Problem
Every major DeFi exploit follows the same pattern. A flash loan is initiated, TVL drops across 2-3 transactions, and by the time the protocol team coordinates a manual pause, funds are already bridged out. The average response window is 4-22 minutes. SentinelGuard closes that window to under 400ms.
T+0s
🔴 "Attack begins"
T+8s
🔴 "Vault drained"
T+4min
🟡 "Team sees Twitter alert"
T+18min
🟡 "Multisig submitted"
T+22min
⚫ "Funds bridged — too late"
How SentinelGuard Fixes It
Sub-slot Detection
Watches every transaction via Geyser gRPC stream. Three detection rules score each slot in real time.
Automated On-chain Pause
When severity exceeds threshold, pause_withdrawals fires on-chain within the same slot. No human needed.
Public Threat Feed
Open WebSocket feed streams live alerts. No API key. Any wallet or aggregator can consume it.
3-line Integration
Protocols add SentinelGuard via npm SDK. No smart contract rewrite required.
Architecture at a Glance
SentinelGuard turns raw Solana transaction data into threat scoring, automated defense, and public alert distribution through a single low-latency pipeline.
SentinelGuard Architecture Pipeline
From raw slot activity to automated defense and public distribution.
Helius Geyser gRPC
- Raw Solana transaction stream
- Real-time slot monitoring
Rust Watcher Engine
- Low-latency detection engine
- Flash loan detection
- TVL velocity monitoring
- Bridge exploit detection
Threat Analysis
- Severity scoring
- Risk classification
- Confidence analysis
Automated Response
- Emergency pause transaction
- Discord & Telegram alerts
- Webhook notifications
Public Alert Feed
- WebSocket stream
- Dashboard consumers
- SDK/API integration
Helius Geyser gRPC
- Raw Solana transaction stream
- Real-time slot monitoring
Rust Watcher Engine
- Low-latency detection engine
- Flash loan detection
- TVL velocity monitoring
- Bridge exploit detection
Threat Analysis
- Severity scoring
- Risk classification
- Confidence analysis
Automated Response
- Emergency pause transaction
- Discord & Telegram alerts
- Webhook notifications
Public Alert Feed
- WebSocket stream
- Dashboard consumers
- SDK/API integration
Hypothetical Scenario
Imagine a protocol loses $232M in a drain that runs across ~12 transactions over ~8 seconds — the pattern of the Drift Protocol class of exploits. SentinelGuard would detect the anomaly after transaction 2–3 and pause withdrawals before transaction 4 fires.
70-80%
Estimated funds that could have been saved
400ms
Time from detection to on-chain pause
Tx 2-3
When detection would have triggered
Next Steps
Continue from the architecture overview into setup, rule logic, or protocol integration.