# ExtractHub

> ExtractHub is a paid MCP server for web search and scraping. Remote MCP server that runs versioned web scrapers on a URL and returns JSON. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Paid
- **Platforms:** API
- **Website:** https://extracthub.io/mcp
- **MCP Registry name:** io.extracthub/extracthub
- **Categories:** web search and scraping
- **Listing updated:** 2026-11-14

## About

Scrapers here are called extractors, and each one is pinned to a version such as @user/slug@1. An agent searches for an extractor that fits a site, checks its output schema and per-call price, then extracts a single URL or queues a batch. When the queue is busy, a single extract becomes an async job you poll.

The server also lets an agent write its own extractors. It can validate a definition against sample HTML without spending credit, save a private draft, test it on real pages with the @draft pin, and release it as version 1.0.0. Multi-step pipelines chain extractors, for example a list page followed by every detail page it links to, with a cost estimate before anything is saved.

It is a hosted Streamable HTTP endpoint that takes your ExtractHub API key as a bearer or X-API-Key header. Calls are billed, rate limited and logged the same way as the REST API, and creating or running things needs a live key rather than a test key.

## Features

- **Versioned extractors:** Search public and private extractors and pin an exact version before extracting.
- **Single and batch extraction:** Extract one URL or queue many in the background, then page through results with a cursor.
- **Agent-built extractors:** Validate, draft, test and release new extractors without opening the Studio.
- **Multi-step pipelines:** Chain extractors from a list page to its detail pages and estimate cost before saving.
- **Account and cost checks:** Read credit balance, key type, rate limits and per-call prices.
- **Async job control:** Check or cancel jobs, batches and pipeline runs.

## Good for

- Pull structured product data from a list of URLs
- Have an agent build a scraper for a site with no extractor
- Scrape a list page and all its detail pages in one pipeline
- Check which fields an extractor returns before paying for a run
- Run a few hundred URLs as a batch and summarise the results

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http extracthub https://api.extracthub.io/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://api.extracthub.io/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "extracthub": {
      "url": "https://api.extracthub.io/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "extracthub": {
      "type": "http",
      "url": "https://api.extracthub.io/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "extracthub": {
      "serverUrl": "https://api.extracthub.io/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "extracthub": {
      "type": "http",
      "url": "https://api.extracthub.io/mcp"
    }
  }
}
```

## Tools

- `search_extractors`: Find extractors for a site or kind of data
- `get_extractor`: Versions, output schema, sample output and price per call
- `get_account`: Credit balance, key type, rate limits and prices
- `extract`: Extract one URL with a pinned extractor
- `get_job`: Status and result of an async extract job
- `cancel_job`: Cancel an async job that has not started
- `create_batch`: Extract many URLs in the background
- `get_batch`: Batch progress, counts and cost so far
- `get_batch_results`: Page through batch results with a cursor
- `cancel_batch`: Cancel a batch
- `validate_extractor_config`: Check an extractor definition, optionally dry-run on sample HTML
- `create_extractor`: Create a private extractor draft
- `update_extractor`: Change the draft of your extractor
- `release_extractor`: Release the draft as a new version
- `set_sample_output`: Replace a released version's sample output
- `set_preview_page`: Replace a released version's preview page
- `delete_extractor`: Delete a private, never released draft
- `validate_pipeline`: Check a multi-step pipeline and estimate its cost
- `create_pipeline`: Save a pipeline of pinned extractor steps
- `update_pipeline`: Change a pipeline's steps, inputs, limits, schedule or webhook
- `get_pipeline`: Show one pipeline's steps, limits and run stats
- `list_pipelines`: List your pipelines with runs and cost
- `run_pipeline`: Run a pipeline in the background
- `get_pipeline_job`: Run progress per step, cost and results preview
- `cancel_pipeline_job`: Cancel a queued or running pipeline run

## FAQ

### How do I authenticate?

Send your ExtractHub API key as Authorization: Bearer <key> or as an X-API-Key header to https://api.extracthub.io/mcp.

### Which clients are covered by the setup guide?

Claude Code, Cursor and Claude Desktop (through the mcp-remote bridge), plus any client with Streamable HTTP and custom headers.

### What does it cost to use?

Extract and batch calls are charged at the extractor's price per successful URL; searching extractors is free, and creating or releasing your own extractors is free with a live key.

### Can a test key do everything?

No. A test key can only validate extractor configs; creating, updating, releasing and running need a live key.

### Is ExtractHub free?

No. ExtractHub is a paid product.

### How do I add ExtractHub to Claude Code?

Run this command in your terminal: claude mcp add --transport http extracthub https://api.extracthub.io/mcp

### What are the best alternatives to ExtractHub?

The closest alternatives to ExtractHub are Scrapling MCP Server, Firecrawl MCP Server, SearXNG Search, NeuralVerge and AI-Powered directory.

## Alternatives

- [Scrapling MCP Server](https://ai-powered.com/mcp/scrapling-mcp-server): Scrapling MCP Server is an open-source MCP server for web search and scraping.
- [Firecrawl MCP Server](https://ai-powered.com/mcp/firecrawl-mcp-server): Firecrawl MCP Server is a freemium MCP server for web search and scraping.
- [SearXNG Search](https://ai-powered.com/mcp/searxng-search): SearXNG Search is an open-source MCP server for web search and scraping.
- [NeuralVerge](https://ai-powered.com/mcp/neuralverge): NeuralVerge is an MCP server for web search and scraping.
- [AI-Powered directory](https://ai-powered.com/mcp/ai-powered-directory): AI-Powered directory is a free MCP server for developer tools.
- [Shopify MCP Server (any store)](https://ai-powered.com/mcp/shopify-mcp-server-any-store): Shopify MCP Server (any store) is a paid MCP server for e-commerce.
- [Vidtrawl: YouTube Search Scraper & Trending Videos (Official API)](https://ai-powered.com/mcp/vidtrawl-youtube-search-scraper-and-trending-videos-official): Vidtrawl: YouTube Search Scraper & Trending Videos (Official API) is a paid MCP server for design and media.
- [Chanroll: YouTube Channel Video Scraper (Official API)](https://ai-powered.com/mcp/chanroll-youtube-channel-video-scraper-official-api): Chanroll: YouTube Channel Video Scraper (Official API) is a paid MCP server for design and media.

---
Source: https://ai-powered.com/mcp/extracthub
