Skip to main content

​
Prerequisites

Before you can configure and use Node.js with Connect AI Embed, you must first do the following:
  • Generate an OAuth JWT bearer token. Copy this down, as it acts as your password during authentication.
  • Install Node.js 18 or later (required for built-in fetch).
  • If you plan to use the LangChain.js / LangGraph example below, obtain an OpenAI API key: https://platform.openai.com/.

​
Connect through LangChain.js and LangGraph

LangChain.js and LangGraph can call the Connect AI Embed MCP server through the @langchain/mcp-adapters package. The example below uses @langchain/langgraph to build a ReAct-style agent on top of the MCP tools — the agent discovers the MCP tools and uses them to answer prompts.
1
Create a folder for the project and initialize it:
mkdir connect-ai-mcp
cd connect-ai-mcp
npm init -y
2
Install the LangChain.js and MCP adapter packages:
npm install @langchain/mcp-adapters @langchain/openai @langchain/langgraph
3
Create a file called langchain.mjs and paste the following. Replace OAUTH_JWT_TOKEN with the token from the prerequisites and YOUR_OPENAI_API_KEY with your OpenAI API key:
import { MultiServerMCPClient } from "@langchain/mcp-adapters";
import { ChatOpenAI } from "@langchain/openai";
import { createReactAgent } from "@langchain/langgraph/prebuilt";

const mcpClient = new MultiServerMCPClient({
  mcpServers: {
    default: {
      transport: "streamable_http",
      url: "https://mcp.cloud.cdata.com/mcp",
      headers: {
        Authorization: "Bearer OAUTH_JWT_TOKEN", // Replace with your OAuth JWT token
      },
    },
  },
});

// Load remote MCP tools exposed by the Connect AI Embed server
const tools = await mcpClient.getTools();
console.log("Discovered MCP tools:", tools.map((t) => t.name));

// Create and run the ReAct-style agent
const llm = new ChatOpenAI({
  model: "gpt-4o",
  temperature: 0.2,
  apiKey: "YOUR_OPENAI_API_KEY", // Replace with your OpenAI API key
});

const agent = createReactAgent({ llm, tools });

const userPrompt = "Tell me how many sales I had in Q1 for the current fiscal year.";
console.log(`\nUser prompt: ${userPrompt}`);

const response = await agent.invoke({
  messages: [{ role: "user", content: userPrompt }],
});

console.log("Agent final response:", response.messages.at(-1).content);
4
Run the script:
node langchain.mjs
The script discovers the Connect AI Embed MCP tools and uses them to answer the prompt.

​
Connect Directly to the Connect AI Embed REST API

If you do not want to go through the MCP server, you can call the Connect AI Embed REST API directly from Node.js with fetch. The API accepts a JSON body with a query field and the JWT bearer token in the Authorization header. See REST API for the full response format.
1
Create a file called connect-ai-client.mjs and paste the following. Replace OAUTH_JWT_TOKEN with the token from the prerequisites:
const API_URL = "https://cloud.cdata.com/api/query";
const JWT = "OAUTH_JWT_TOKEN"; // Replace with your OAuth JWT token

async function runQuery(query) {
  const response = await fetch(API_URL, {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "Authorization": `Bearer ${JWT}`,
    },
    body: JSON.stringify({ query }),
  });

  if (!response.ok) {
    throw new Error(`Request failed: ${response.status} ${await response.text()}`);
  }

  const data = await response.json();

  // The API returns a 200 even when a partial-result error occurs, so always check for error.
  if (data.error) {
    throw new Error(`${data.error.code}: ${data.error.message}`);
  }

  return data.results[0];
}

const result = await runQuery("SELECT * FROM Salesforce1.Salesforce.Account LIMIT 5");
console.log("Columns:", result.schema.map((c) => c.columnName));
console.log("Rows:", result.rows);
2
Run the script:
node connect-ai-client.mjs