Skip to main content

Decodo LangChain Tools

The @decodo/langchain-ts package enables developers to use Decodo's Web Scraper API alongside their LangChain applications.

The Web Scraper API features:

  • Easy web data access. Simplified retrieval of information from websites and online sources.
  • Geographic flexibility. Access content regardless of regional restrictions.
  • Reliable scraping. Advanced techniques to avoid detection and blocks.

Features​

The @decodo/langchain-ts plugin features:

  • Web Scraping: Scrape any URL and retrieve Markdown content
  • Google Search: Search Google and retrieve structured results
  • Amazon Search: Search Amazon and retrieve structured product data
  • Geolocation selection for location-sensitive content
  • JavaScript rendering for targets that require a real browser
  • Markdown output for token-efficient connection to LLMs

Installation​

npm install @decodo/langchain-ts

Quick Start​

To use the tools in this project, you will need a Decodo Advanced Web Scraping API subscription. Free trials are available on the dashboard.

Examples​

A set of agentic examples for each tool.

Universal​

import dotenv from "dotenv";
import { ChatOpenAI } from "@langchain/openai";
import { createReactAgent } from "@langchain/langgraph/prebuilt";
import { DecodoUniversalTool } from "@decodo/langchain-ts";

dotenv.config();

const main = async () => {
  const username = process.env.SCRAPER_API_USERNAME!;
  const password = process.env.SCRAPER_API_PASSWORD!;

  const decodoUniversalTool = new DecodoUniversalTool({ username, password });

  const model = new ChatOpenAI({
    model: "gpt-4o-mini",
  });

  const agent = createReactAgent({
    llm: model,
    tools: [decodoUniversalTool],
  });

  const result = await agent.invoke({
    messages: [
      {
        role: "user",
        content:
          "scrape the wikipedia NBA 2025 season page and tell me who won in 2025?",
      },
    ],
  });

  console.log(result.messages[result.messages.length - 1].content);
};

if (require.main === module) {
  main();
}

For more examples, see the @decodo/langchain-ts on GitHub.

const main = async () => {
  const username = process.env.SCRAPER_API_USERNAME!;
  const password = process.env.SCRAPER_API_PASSWORD!;

  const decodoGoogleSearchTool = new DecodoGoogleSearchTool({
    username,
    password,
  });

  const model = new ChatOpenAI({
    model: "gpt-5-mini",
  });

  const agent = createReactAgent({
    llm: model,
    tools: [decodoGoogleSearchTool],
  });

  const prompt =
    "which mobile service provider appears first on Google in Germany?";

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

  console.log(result.messages[result.messages.length - 1].content);
};
const main = async () => {
  const username = process.env.SCRAPER_API_USERNAME!;
  const password = process.env.SCRAPER_API_PASSWORD!;

  const decodoAmazonSearchTool = new DecodoAmazonSearchTool({
    username,
    password,
  });

  const model = new ChatOpenAI({
    model: "gpt-5-mini",
  });

  const agent = createReactAgent({
    llm: model,
    tools: [decodoAmazonSearchTool],
  });

  const prompt =
    "What is the cheapest laptop with a GeForce RTX 5080 on Amazon in France?";

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

  console.log(result.messages[result.messages.length - 1].content);
};

Configuration​

All tools accept a DecodoConfig object:

type DecodoConfig = {
  username: string; // Your Web Advanced product username
  password: string; // Your Web Advanced product password
};

API Parameters​

See the Scraper API documentation for a list of available parameters

License​

MIT

Support​

For support, please visit Decodo's documentation or open an issue on GitHub.

Was this page helpful?


You can also leave detailed feedback on GitHub.