Tabstack for LangChain

First call

Start with a current question.

Search gives your model sources. It still has to build the answer. Tabstack Research completes that answer-building work and returns the cited result through the integration, so the agent receives a finished answer rather than a list of pages to read.

Why LangChain + Tabstack

LangChain’s built-in loaders fetch and parse in your process and need Playwright for JavaScript pages. The Tabstack adapters move that work to a hosted API exposed as native LangChain tools: schema-enforced output, server-side rendering of JS-heavy pages, and one key for extraction, research, generation, and automation. It’s an SDK call, not a loader coupled to your LangChain version.

Two officially maintained packages track the same tool surface:

TypeScript: @tabstack/langchain for LangChain.js
Python: langchain-tabstack for LangChain (Python)

Quickstart

Install the adapter and set your key:

# TypeScript
npm install @tabstack/langchain @langchain/core zod

# Python
pip install langchain-tabstack

export TABSTACK_API_KEY="your-key-here"

Drop the tool set into an agent. TypeScript:

import { createAgent } from "langchain";
import { tabstackTools } from "@tabstack/langchain";

const agent = createAgent({
  model: "openai:gpt-4o",
  tools: tabstackTools,
  systemPrompt: "You are a research assistant with web research and extraction tools.",
});

const result = await agent.invoke({
  messages: [{ role: "user", content: "What are Vercel's pricing plans?" }],
});

console.log(result.messages.at(-1)?.content);

The same set in Python:

from langchain.agents import create_agent
from langchain_tabstack import TABSTACK_TOOLS

agent = create_agent("openai:gpt-4o", tools=TABSTACK_TOOLS)

result = agent.invoke(
    {"messages": [{"role": "user", "content": "What are Vercel's pricing plans?"}]}
)
print(result["messages"][-1].content)

tabstackTools / TABSTACK_TOOLS read TABSTACK_API_KEY from the environment. Every tool is also exported individually, so you can hand a single one to a chain without an agent.

The tools

Tool	What it does
`extract_structured_data`	Pull specific fields from a URL into a JSON shape you define.
`extract_page_content`	Fetch a page as clean Markdown.
`research_question`	Synthesized answer with cited sources across multiple pages.
`generate_structured_data`	Fetch a page, then AI-transform it into derived or reshaped JSON.
`automate_browser_task`	Run a multi-step, natural-language browser task.

Common use cases

Give a research agent cited, multi-source answers instead of a single fragile page fetch.
Extract typed records (pricing, listings, docs) from a public URL straight into your chain state.
Reshape a live page into derived JSON with generate_structured_data.
Run natural-language browser tasks from inside an agent step.

Next steps

Note

Good to know.

Automate is public only. The hosted task does not sign in to accounts, in this integration or any other.

Want the endpoint rather than the adapter? The Research page explains what happens inside the call.
Want to read the adapter itself? The package is on npm.
Building somewhere else? All integrations.

Return finished web results inside LangChain.