LangChain Tools

The @smartagentkit/langchain package provides LangChain-compatible tools for AI agents to interact with SmartAgentKit wallets.

createSmartAgentKitTools

createSmartAgentKitTools(
  client: ISmartAgentKitClient,
  walletAddress: string,
  sessionKey?: string
): DynamicStructuredTool[]

Create an array of LangChain DynamicStructuredTool instances for wallet interaction.

Parameter	Type	Required	Description
client	ISmartAgentKitClient	Yes	SDK client or MockSmartAgentKitClient
walletAddress	string	Yes	Wallet address
sessionKey	string	No	Session key private key for scoped access

Returns: Array of 5 DynamicStructuredTool instances.

import { createSmartAgentKitTools } from "@smartagentkit/langchain";
import { SmartAgentKitClient } from "@smartagentkit/sdk";

const client = new SmartAgentKitClient({ ... });
const wallet = await client.createWallet({ ... });

const tools = createSmartAgentKitTools(client, wallet.address);

Tools

All tools return JSON strings for LLM consumption.

check_wallet_balance

Check the ETH balance of the wallet.

Input schema: {} (no inputs)

Returns:

{
  "wallet": "0x...",
  "eth": "1.5",
  "ethWei": "1500000000000000000"
}

check_spending_allowance

Check the remaining spending allowance for a token.

Input schema:

Field	Type	Required	Description
token	string	Yes	Token address (use zero address for ETH)

Returns:

{
  "wallet": "0x...",
  "token": "0x0000000000000000000000000000000000000000",
  "remainingWei": "500000000000000000",
  "remaining": "0.5"
}

send_transaction

Send a single transaction through the wallet.

Input schema:

Field	Type	Required	Description
target	string	Yes	Target contract address
value	string	No	ETH value in wei
data	string	No	Calldata hex string

Returns (success):

{
  "success": true,
  "transactionHash": "0x..."
}

Returns (failure):

{
  "success": false,
  "error": "Spending limit exceeded for 0x0000...0000: attempted 2000000000000000000, remaining 500000000000000000"
}

send_batch_transaction

Send multiple transactions atomically through the wallet.

Input schema:

Field	Type	Required	Description
calls	Array<{ target, value?, data? }>	Yes	Array of transaction calls

Returns (success):

{
  "success": true,
  "transactionHash": "0x...",
  "callCount": 3
}

Returns (failure):

{
  "success": false,
  "error": "Wallet 0x... is currently paused"
}

check_wallet_status

Check whether the wallet is paused.

Input schema: {} (no inputs)

Returns:

{
  "wallet": "0x...",
  "paused": false,
  "status": "active"
}

Usage with LangChain Agents

import { ChatOpenAI } from "@langchain/openai";
import { AgentExecutor, createReactAgent } from "langchain/agents";
import { createSmartAgentKitTools } from "@smartagentkit/langchain";

const tools = createSmartAgentKitTools(client, wallet.address);

const llm = new ChatOpenAI({ model: "gpt-4" });
const agent = await createReactAgent({ llm, tools, prompt });
const executor = new AgentExecutor({ agent, tools });

const result = await executor.invoke({
  input: "Check my wallet balance and send 0.01 ETH to 0xAlice...",
});

Usage with MockSmartAgentKitClient

The tools work identically with the mock client for testing:

import { MockSmartAgentKitClient } from "@smartagentkit/testing";
import { createSmartAgentKitTools } from "@smartagentkit/langchain";

const mockClient = new MockSmartAgentKitClient();
const wallet = await mockClient.createWallet({ ... });
const tools = createSmartAgentKitTools(mockClient, wallet.address);