Skip to main content

LangChain Integration

LangChain integration for ClickSend via Model Context Protocol (MCP).

This package provides LangChain tools that wrap the ClickSend MCP Server, enabling you to send SMS, MMS, and other ClickSend communications through LangChain agents.


Installation

npm install @clicksend/langchain-clicksend-mcp

Prerequisites

You need a ClickSend account with API credentials:

  • Username: Your ClickSend account username

  • API Key: Your ClickSend API key

Get your credentials from the ClickSend Dashboard.


Available Tools

The integration automatically exposes all tools from the ClickSend MCP server, which typically include:

  • send_sms: Send SMS messages

  • send_mms: Send MMS messages

  • get_sms_history: Retrieve SMS history

  • get_account_balance: Check account balance

  • And more...

The exact tools available depend on the ClickSend MCP server version.


Configuration

import { createClickSendTools } from "@clicksend/langchain-clicksend-mcp"; const tools = await createClickSendTools({   username: "your-clicksend-username",   apiKey: "your-clicksend-api-key", });

Options

  • username (required): Your ClickSend account username

  • apiKey (required): Your ClickSend API key


Usage

Basic Example

import { createClickSendTools } from "@clicksend/langchain-clicksend-mcp"; import { ChatOpenAI } from "@langchain/openai"; import { createAgent } from "langchain"; import dontenv from "dotenv"; dontenv.config(); let agentInstance: any = null; const tools = await createClickSendTools({   username: process.env.CLICKSEND_USERNAME || "",   apiKey: process.env.CLICKSEND_API_KEY || "", }); async function getAgent() {     if(!agentInstance) {         agentInstance = await createAgent({             model: new ChatOpenAI({                  modelName: "gpt-4o-mini",                 temperature: 0              }),             tools,         }) }     return agentInstance; } export async function runAgent(userMessage: string) {     const agent = await getAgent();     const result = await agent.invoke({         messages : [            { role: "user", content : userMessage },         ]     })     return result; } const result = await runAgent("send a text message to +1234567890 saying 'Hello from ClickSend via LangChain!'"); console.log(result);

Advanced Usage

Direct MCP Client Access

If you need more control, you can access the underlying MCP client:

import { createClickSendMCPClient } from "@clicksend/langchain-clicksend-mcp"; const client = createClickSendMCPClient({   username: process.env.CLICKSEND_USERNAME!,   apiKey: process.env.CLICKSEND_API_KEY!, }); // Get tools const tools = await client.getTools(); // Access the raw MCP client // ... additional MCP operations


FAQS

Does LangChain need a separate host?

No - LangChain itself is the host. No separate server needed.

How do I install LangChain MCP?

Install it on your code tool (Eg. Visual Studio code) and follow the LangChain docs here.

Can I connect to any AI model?

Yes - Since ClickSend MCP Server is in Github, you can connect to any AI model of your choice

Did this answer your question?