const blade = new BladeEngine({
network: 'mainnet',
wallet: process.env.PRIVATE_KEY
});
// Execute intent
await blade.execute(
"Swap 1 ETH for USDC and stake it on Lido"
);
/* Output:
✓ Route mapped
✓ Swap executed (Tx: 0x...)
✓ Stake executed (Tx: 0x...)
*/
FEATURES
Agents think.
BLADE ENGINE executes.
Build for specific
- Smart routing
- Risk management
- Execution boundaries
- Transaction batching
Connect seamlessly
- Any LLM framework
- Natural language input
- Policy enforcement
- Full audit logs
ARCHITECTURE
Infrastructure,
not a wrapper.
User
AI Agent
↓
Routing Layer
Intent mapping
Intent mapping
↓
Policy Engine
Limits, Rules, Audits
Limits, Rules, Audits
↓
Execution Node
BLADE Agent
BLADE Agent
↓
Target Network
EVM / Solana
EVM / Solana
INTEGRATION
Natural language.
Real execution.
Input parsed by LLM from user
const request = {
intent: "Buy 0.5 ETH worth of LINK",
maxSlippage: "1%"
};
BLADE ENGINE output transaction
// Engine maps intent to DEX
{
target: "UniswapV3Router",
data: "0x...",
value: "500000000000000000"
}
SECURITY
You define
the boundaries.
const policy = {
maxSpend: "1.0 ETH",
allowTokens: ["USDC", "LINK"],
denyProtocols: ["UnknownDEX"]
};
blade.setPolicy(policy);
// If agent attempts breach:
try {
await blade.execute(intent);
} catch (e) {
console.error(e.message);
// "Policy violation: Max spend exceeded"
}