LangChain

Use Lightcone with LangChain as a document loader to extract web page content, or as a tool that gives LangChain agents the ability to browse the web.

Install

pip install langchain-tzafon tzafon

npm install @langchain/tzafon @tzafon/lightcone

Document loader

Load web page content into your LangChain pipeline:

from langchain_tzafon import TzafonLoader

loader = TzafonLoader(
    urls=["https://example.com", "https://example.com/about"],
    kind="browser",
)

documents = loader.load()
for doc in documents:
    print(doc.page_content[:200])
    print(doc.metadata["url"])

import { TzafonLoader } from "@langchain/tzafon";

const loader = new TzafonLoader({
  urls: ["https://example.com", "https://example.com/about"],
  kind: "browser",
});

const documents = await loader.load();
for (const doc of documents) {
  console.log(doc.pageContent.slice(0, 200));
  console.log(doc.metadata.url);
}

Browser tool for agents

Give a LangChain agent the ability to navigate and interact with web pages:

from langchain_tzafon import TzafonBrowserTool
from langchain.agents import initialize_agent, AgentType
from langchain_openai import ChatOpenAI

tools = [TzafonBrowserTool()]

agent = initialize_agent(
    tools,
    ChatOpenAI(model="gpt-4o"),
    agent=AgentType.STRUCTURED_CHAT_ZERO_SHOT_REACT_DESCRIPTION,
)

result = agent.run("Go to news.ycombinator.com and tell me the top 3 stories")
print(result)

How it works

• Document loader: Creates a Lightcone browser session, navigates to each URL, extracts the HTML content, and returns it as LangChain Document objects with metadata (URL, title).
• Browser tool: Wraps Lightcone actions (navigate, click, type, screenshot, and more) as LangChain tools that an agent can invoke during its reasoning loop.

The document loader uses Lightcone’s stealth browser, so it can access pages that block simple HTTP requests or require JavaScript rendering.

Resources

• LangChain Python integration docs
• LangChain JS integration docs

See also

• CrewAI — multi-agent workflows with browser tools
• Web scraping — scraping patterns that work with document loaders
• Computers — direct browser control without a framework