Autonomous DeFi trading agent for BNB Chain with multi-strategy engine, network switching, and reinforced learning.
Production-ready autonomous trading agent for BNB Chain testnet & mainnet. Features 3 intelligent strategies, real-time performance analytics, on-chain event logging, and self-improving reinforced learning.
clawhub install clawtrade-bnb
cd ~/.openclaw/workspace/skills/clawtrade-bnb
npm install
# Copy example config
cp config.deployed.json config.live.json
# Edit with your settings
nano config.live.json
# Set RPC endpoint, contract addresses, wallet
# Option A: Environment variable (recommended)
export PRIVATE_KEY="your_testnet_private_key"
# Option B: .env file (git-ignored)
echo "PRIVATE_KEY=your_key" >> .env
# NOTE: Never commit private keys!
# Test connection and contracts
node agent-cli.js network status
# Check wallet balance
npm run verify
# Terminal 1: Run strategy engine (60-second cycles)
node strategy-scheduler.js
# Terminal 2: Real-time dashboard
npm run dev:dashboard
# → Open http://localhost:5173
# Terminal 3: Control panel
node agent-cli.js
# Example commands:
node agent-cli.js network testnet # Switch network
node agent-cli.js perf summary # See performance
node agent-cli.js learn now # Optimize strategies
DeFi Strategy Engine
├─ Compound Yield Strategy
│ └─ Harvest when pending > $25 → Re-deposit
├─ Rebalance Strategy
│ └─ Move 20% from low-APR to high-APR vault
└─ Dynamic Harvest Strategy
└─ Harvest only if pending > 2x gas cost
↓ (runs every 60 seconds)
Strategy Scheduler
├─ Read vault APRs & pending rewards
├─ Execute all 3 strategies
└─ Log actions + TX hashes
↓ (logs to blockchain)
On-Chain Logger
├─ execution-log.jsonl (append-only)
├─ performance-metrics.json (cumulative)
└─ learning-state.json (optimization history)
↓ (analyzes continuously)
Reinforced Learning System
├─ Tracks success rates per strategy
├─ Detects failure patterns
├─ Auto-adjusts thresholds
└─ Generates improvement reports
↓ (displays real-time)
Dashboard + Control Panel
├─ React dashboard (http://localhost:5173)
├─ Agent CLI (network, perf, learn commands)
└─ Performance API (/api/logs, /api/health)
config.deployed.json - Contract addresses & ABIs
{
"chainId": 97,
"network": "BNB Testnet",
"contracts": [
{
"vaultId": "vault_eth_staking_001",
"address": "0x588eD88A145144F1E368D624EeFC336577a4276b",
"strategy": "Ethereum 2.0 Staking",
"risk_score": 0.3
}
]
}
config.scheduler.json - Strategy thresholds
{
"scheduler": {
"execution_interval_seconds": 60,
"enabled": true
},
"agent": {
"harvest_threshold_usd": 25,
"rebalance_apr_delta": 2.0,
"max_allocation_percent": 0.35,
"min_confidence": 0.6
}
}
Each 60-second cycle:
COMPOUND YIELD
REBALANCE
DYNAMIC HARVEST
Example Output:
Cycle #42 @ 2026-02-18T18:00:00Z
✓ vault_eth_staking_001: COMPOUND ($45.50 harvested)
✓ vault_high_risk_001: REBALANCE (2.1% APR delta)
✓ vault_link_oracle_001: HARVEST ($12.30 pending)
✅ Total Rewards: $57.80 | Compounded: $45.50
node agent-cli.js network status # Current network config
node agent-cli.js network testnet # Switch to testnet
node agent-cli.js network mainnet # Switch to mainnet (⚠️ production)
node agent-cli.js perf summary # Quick stats
node agent-cli.js perf report # Detailed analysis
node agent-cli.js perf vaults # Per-vault breakdown
node agent-cli.js perf strategies # Strategy effectiveness
node agent-cli.js learn now # Analyze & optimize
node agent-cli.js learn report # View improvements
node agent-cli.js learn reset # Reset learning state
| Network | Chain ID | Use Case | Harvest Min | Gas Multiplier |
|---|---|---|---|---|
| BNB Testnet | 97 | Development | $25 | 1.2x |
| BNB Mainnet | 56 | Production | $100 | 1.5x |
Switch instantly without restarting:
# Current config
node agent-cli.js network status
# → BNB Testnet
# Switch to production
node agent-cli.js network mainnet
# → Updated RPC, contract addresses, and thresholds
# All settings updated automatically
clawtrade-bnb/
├── defi-strategy-engine.js # 3 strategies (compound, rebalance, harvest)
├── on-chain-logger.js # Event logging with TX hashes
├── strategy-scheduler.js # Main loop (60s cycles)
├── network-switcher.js # Testnet ↔ mainnet toggle
├── performance-analytics.js # Real APR & metrics
├── reinforced-learning.js # Self-improving parameters
├── agent-cli.js # Control panel
├── dashboard/ # React frontend (real-time)
├── contracts/ # Vault smart contracts
├── config.deployed.json # Contract addresses & ABIs
├── config.scheduler.json # Strategy thresholds
├── execution-log.jsonl # Action history (generated)
├── performance-metrics.json # Metrics (generated)
├── learning-state.json # Learning progress (generated)
├── README.md # User guide
├── README_STRATEGY.md # Strategy details
├── README_ADVANCED.md # Network switching & learning
├── SKILL.md # This file
└── package.json # Dependencies
This is a standalone, complete skill. It can also integrate with:
| File | Purpose |
|---|---|
README.md | Complete user guide |
README_STRATEGY.md | Strategy details & examples |
README_ADVANCED.md | Network switching & reinforced learning |
SKILL.md | This installation & architecture guide |
✅ Production-ready code (tested, documented, error-handling) ✅ 3 profitable strategies (auto-optimizing, self-learning) ✅ Real-time dashboard (React, live updates) ✅ CLI control panel (manage from terminal) ✅ On-chain logging (auditable, transparent) ✅ Network switching (testnet → mainnet in seconds) ✅ Self-improvement (learns from failures automatically) ✅ Complete documentation (guides, examples, FAQ)
For someone else to replicate:
Install
clawhub install clawtrade-bnb
npm install
Configure
# Edit config files with your contracts & RPC
nano config.deployed.json
Deploy Contracts (if using new vaults)
cd contracts && npm run deploy:testnet
Run
node strategy-scheduler.js # Main engine
npm run dev:dashboard # Dashboard
node agent-cli.js # Control panel
Monitor
Total setup time: ~15 minutes
MIT - Free to use, modify, and distribute