Agent Overview
Read how Nexxore's autonomous AI agents work: the agent swarm architecture, roles and responsibilities, and how they coordinate to execute strategies.
What Are Nexxore Agents?
Nexxore Agents are autonomous AI programs that run 24/7 to:
Monitor markets and on-chain data in real-time
Identify opportunities and risks
Execute strategies and rebalancing
Generate research and actionable insights
Unlike traditional DeFi bots that follow simple if-then rules, Nexxore agents use AI to make intelligent decisions in complex, changing market conditions.
The Agent Swarm
Nexxore operates a coordinated swarm of specialized agents, each with distinct responsibilities:
On-Chain Analyst
The Analyst
Market intelligence, technicals, fundamentals for 20 perp markets
Prediction Markets Agent
The Oracle
AI signals for Polymarket & Kalshi, arbitrage detection
Market Data Agent
The Collector
Dexscreener integration, top movers, new pairs, liquidity monitoring
Signal Engine
The Scorer
4-signal alpha detection, composite scoring (0–100), WebSocket feed
Alpha Agent
The Strategist
Trading signal generation, entry/exit recommendations
Web3 Intelligence Agent
The Auditor
Contract analysis, security scoring, rug pull detection
Execution Agent
The Trader
Order routing, slippage minimization, venue selection
Risk Agent
The Guardian
CVaR monitoring, drawdown enforcement, position limits
Agent Architecture
Data Layer
All agents draw from a shared data layer that includes:
On-chain
Transactions, wallet balances, contract events
Price Feeds
Real-time prices from CEXs and DEXs
News/Social
Crypto news, Twitter sentiment, Discord activity
Whale Alerts
Large transaction monitoring
DeFi Analytics
TVL, yields, protocol metrics
LLM Engine
At the core of each agent is a Large Language Model (LLM) that provides:
Analysis and reasoning capabilities
Pattern recognition across data streams
Natural language synthesis for reports
Decision-making under uncertainty
Execution Layer
Agents execute actions through:
Smart contract calls (ERC-20 approvals, swaps, positions)
Multi-chain support (Arbitrum, Ethereum, future chains)
Transaction optimization (gas estimation, MEV protection)
Agent Coordination
Agents don't operate in isolation—they coordinate through a shared messaging layer:
Research → Alpha
Market analysis informs signal generation
Alpha → Execution
Signals trigger trade execution
Web3 Intel → All
Security scores gate asset exposure
Risk → All
Risk limits override all other agents
Example Workflow
Research Agent detects unusual whale accumulation in ETH
Web3 Intelligence confirms no smart contract risks
Alpha Agent generates a long signal with 72% confidence
Risk Agent approves position size within CVaR limits
Execution Agent routes order to GMX for best fill
Available Agents
On-Chain Analyst (Research Agent)
Comprehensive market intelligence for perpetual trading.
Provides:
Real-time price tracking for 20 markets (BTC, ETH, SOL, etc.)
Technical analysis: RSI, MACD, EMA crossovers, Bollinger Bands
Fundamental analysis: TVL, dev activity, tokenomics
Derivatives data: funding rates, open interest, liquidations
Support/resistance level calculation from OHLC data
Trading signals: BULLISH/BEARISH/NEUTRAL with confidence
API integration for execution agents (
window.nexxoreAgent)
Market Data Agent
Dexscreener data integration and real-time DEX pair monitoring.
Provides:
Async Dexscreener API client with rate limiting (60/300 rpm)
Top movers discovery across 5 chains (Solana, Ethereum, Base, BSC, Arbitrum)
New pair detection (age <24h, liquidity >$100K)
Liquidity change monitoring (>20% in 10 min)
Token search across all Dexscreener pairs
Time-series SQLite storage with 7-day retention
Redis + in-memory LRU caching (10s–300s TTLs)
REST API on port 3860
Signal Engine (Market Intelligence Agent)
Composite alpha signal scoring system with 4 independent detectors.
Provides:
Momentum spike detection (price_5m >5%, vol >2× avg, liq >$500K → 0-40 pts)
Volume breakout detection (vol_1h >3× 24h avg → 0-30 pts)
Liquidity inflow detection (>20% increase in 10 min → 0-20 pts)
New pair scoring (age <24h, liq >$100K → 0-10 pts)
Composite scoring (0–100) with classification: >70 strong trade, 50-70 watchlist
Rolling window data pipeline (5m/15m/1h/24h)
WebSocket live feed (/ws/signals/live)
REST API on port 3861
AI-powered intelligence for prediction market traders.
Provides:
Real-time odds analysis for Polymarket & Kalshi
Cross-platform arbitrage detection
AI trading signals with confidence scoring
Market categorization: Politics, Crypto, Sports, Economy, Tech
Odds movement tracking and alerts
Portfolio analytics and performance tracking
Auto-trade integration (coming soon)
Alpha Agent
Trading signal generation engine.
Provides:
Entry and exit signals
Risk/reward analysis
Position sizing recommendations
Multi-timeframe analysis
Confidence scoring
Web3 Intelligence Agent
On-chain data analysis and security auditing.
Provides:
Token contract analysis
Liquidity monitoring
Holder distribution analysis
Honeypot and rug pull detection
Security scoring (0-100)
Agent Performance
Agents are continuously monitored for performance:
Uptime
Agent availability (target: 99.9%)
Latency
Time from signal to execution
Accuracy
Signal win rate over time
Risk Adherence
Compliance with CVaR limits
Technical Specifications
Update Frequency
Every block (~12s on Ethereum)
Supported Chains
Arbitrum, Ethereum (more coming)
LLM Provider
OpenAI GPT-4 / Anthropic Claude
Execution Venues
GMX, Uniswap, 1inch
Next Steps
Last updated