By clicking “Accept”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.
DenyAccept
Coming to Claude Marketplace
Hosted MCP · Available now

Bring web scraping to Claude ChatGPT Cursor Windsurf VS Code Codex

The Import.io Web Scraper MCP server brings hosted scraping, browser automation, proxy routing, captcha handling, and structured data extraction into your AI tools. Point Claude, ChatGPT, Cursor, or any MCP client at our hosted endpoint. Import.io runs the scraping engine, you just connect.
Connect any MCP client
Coming soon
Get on Claude Marketplace
View ChatGPT setup
View Cursor setup
View Windsurf setup
View VS Code setup
View Codex setup
Connect today with any MCP client. One-click Claude Marketplace install - coming soon.
Connect ChatGPT to the hosted Import.io MCP endpoint using OAuth.
Add the hosted Import.io endpoint to Cursor and start using web data in your coding workflows.
Connect Windsurf to Import.io through the hosted MCP endpoint.
Connect VS Code to Import.io through a compatible MCP configuration.
Connect Codex to Import.io through the hosted MCP endpoint.
Works with:
Claude
ChatGPT
Cursor
Windsurf
VSCode
Codex

Built-in browsing gets you an answer. Import.io gets you structured web data.

Built-in browsing is useful for one-off page access. Repeatable extraction needs browser automation, proxy routing, structured outputs, and reliable handling of dynamic websites.

Dynamic pages need a browser

JavaScript, infinite scroll, and interactive controls can hide data from a basic HTTP request. Import.io renders the page in a real browser before extraction.

Location changes the answer

Prices, availability, and content vary by country. Target the geography your workflow needs so the result reflects the market you actually care about.

Raw pages waste model context

Sending full HTML to an LLM burns tokens and adds noise. Ask for the fields you need and return them as clean structured data.

Proxy-aware scraping

Route requests through Import.io’s scraping engine using the proxy type your workload requires—from data-centre to residential.

Country targeting

Select a target country when geography changes the page, so your agent works with the right regional version.

CAPTCHA support

Built-in CAPTCHA handling helps agents continue through blocking cases that would otherwise interrupt extraction.

How it works: Three steps to structured web data.

Works today with any compatible MCP client. No scraping infrastructure to run.
01

Connect your MCP client

Add Import.io’s hosted endpoint to your client’s MCP configuration. Start with 10,000 free calls. No credit card required.
https://mcp.import.io/mcp
02

Authenticate securely

Most compatible clients open the OAuth flow automatically. For scripts or clients without OAuth, use an Import.io API key as a bearer token.
Authorization: Bearer mcp_live_...
03

Ask for the data you need

Describe the page and fields you want. Your client invokes Import.io to render the page, perform browser actions, and return structured data.
Scrape the top stories from news.ycombinator.com
Claude
ChatGPT
Cursor
Any MCP client
// Add the import.io marketplace, then install
$ claude plugin marketplace add import-io/mcp-plugin
$ claude plugin install importio-web-scraper
> Import.io Web Scraper installed

// Claude now has web scraping tools
importio_render
   *
importio_extract_data
   *
importio_get_html
   *
Copy
// ChatGPT → Settings → Connectors → Create
//
//   Name   Import.io Web Scraper
//   URL    https://mcp.import.io/mcp
//   Auth   OAuth
//
// Enable the connector in a chat, then just ask:
"Scrape the top stories from news.ycombinator.com"
> importio_render    
 30 results extracted
‍
Copy
// Add to ~/.cursor/mcp.json
{
 "mcpServers": {
   "importio": {
     "url": "https://mcp.import.io/mcp"
   }
 }
}

// Cursor's agent can now scrape the web
importio_render
   *
importio_action_screen_capture
   *
Copy
// Streamable-HTTP MCP endpoint — point any client here
> https://mcp.import.io/mcp

// Authenticate with OAuth (most clients handle this for
// you), or create a key at https://mcp.import.io/app/keys
Authorization: Bearer mcp_live_...

// Full client walkthroughs
> https://mcp.import.io/docs
Copy

How it fits together: One endpoint between your AI and a hosted scraping engine

Your MCP client invokes Import.io tools. The hosted engine handles the browser workflow and returns the requested result.

ChatGPT

Research and workflow agents

Claude

Analysis and browser tasks

Cursor

Developer and coding agents

Import.io Web Scraper MCP

The hosted scraping engine
importio_render
browser actions
importio_extract_data
importio_get_html

Structured data

Defined fields for the agent

Rendered HTML

Page context and inspection

Screenshots

Visual source evidence
Works with Claude, ChatGPT, Cursor, Windsurf, VS Code, and other MCP-compatible clients.

What can you do with it? Ask in plain language. Get structured data back.

Describe the information you need. Your AI client selects the appropriate Import.io tools and returns only the fields you asked for.
You ask Claude
“Scrape the top five stories from news.ycombinator.com and return title, points, and link.”
→
Structured result
[
 {
    "title": "Example story title",
    "points": 342,    "link":     "https://example.com/story"
   },
   {
    "title": "Another story",
    "points": 271,
    "link":"https://example.com/another"
 }
]

How to get access. Production web scraping without infrastructure to manage.

OAuth authentication for compatible clients.
Bearer API keys for scripts or clients without OAuth.
Usage metered per successful tool call.
Set your own monthly spending limit.
Paid usage stops when the configured limit is reached.

MCP client Connect

Sign up for an Import.io MCP account, add the hosted endpoint to your MCP client, authenticate, and start scraping. You get 10,000 free successful calls to begin.
Start free
Available now

Agent Marketplace

Use Import.io through client-specific integrations as they become available.
Until then, connect any supported client through the hosted MCP endpoint.

Start free... Then pay only for successful calls.

Get 10,000 successful tool calls free on signup. After that, usage costs $0.0002 per successful tool call. Malformed requests, blocked requests, and failed upstream calls are not billed. Set a monthly spending limit you control.
$0.0002
per successful tool call
Your free allocation is available on signup. Paid usage begins after the free calls are used and billing is active.
Proxy routing, country targeting, and CAPTCHA support included.
Set a monthly spending limit from billing.
Paid usage stops when the configured limit is reached.
Start with 10,000 free callsSee pricing

What counts as a billable call

One agent request can use multiple tool calls-for example rendering a page, performing browser actions, taking a screenshot, and extracting structured data.
Successful render or action
Billable
Successful screenshot or extraction
Billable
Malformed or failed upstream request
Not billed
Blocked request
Not billed
Contact sales

What does Import.io do?

add here

Import.io turns any website into structured, decision-ready data. It provides pricing intelligence for retail and brand teams (competitor price monitoring, MAP compliance, availability and assortment), managed and self-service web data extraction, and Aperture, its pricing intelligence platform. Teams use it to track markets in real time and feed clean data into their own dashboards, APIs or warehouses.

What is pricing intelligence?

add here

Pricing intelligence is the practice of tracking competitor prices, promotions, availability and assortment across retailers and marketplaces, then turning that data into decisions. It goes beyond simple price tracking by structuring and normalising the data so teams can compare like-for-like, spot MAP violations, and respond to market changes quickly.

How is this different from the browsing my AI tool already has?

add here

Built-in browsing reads a page or two and struggles the moment a site defends itself. The MCP puts a real scraping engine underneath, so your agent can get past blocks, ask for a specific country, and pull clean structured data across many pages instead of just one.

What can the tools actually do?

add here

Import.io takes an AI-first approach to every project. Instead of starting with manual scoping or guesswork, you can engage directly with our platform using natural language to define your requirements. The AI helps refine the scope, shape the schema, and even build your first proof-of-concept automatically. This way, you see value faster, and the POC already reflects the exact business outcomes you care about.

How do I connect it?

add here

Add the hosted endpoint at https://mcp.import.io/mcp as a connector in your client. Most clients start the sign-in flow automatically. If your client uses bearer tokens instead, create an API key and send it in the authorization header. It takes about two minutes. For step-by-step setup, see the quickstart guide.

What does it cost?

add here

You get 10,000 free successful calls when you sign up, with no payment method needed to begin. After that it is a flat $0.0002 per successful tool call, billed monthly. You only pay for calls that return data, and you set a monthly spend limit you control. See the full pricing details.

What counts as a billable call?

add here

Only successful tool calls. Setup requests, malformed requests, blocked requests, and calls that fail upstream are never billed. You pay for results, not attempts. See the pricing page for the full breakdown.

Give your AI agent
a real web-scraping engine.

Connect any compatible MCP client to Import.io’s hosted endpoint and start with 10,000 free successful calls.
Connect any MCP client
View documentation
Get connected in minutes. No credit card required.
No credit card required