Repository avatar
AI Tools
v0.1.3
active

scrapewise

ai.scrapewise/scrapewise

Clean, matched, AI-ready product data from any website

Documentation

ScrapeWise MCP Server

ScrapeWise turns any website into clean, matched, AI-ready product data.

Clean, matched, AI-ready product data from any website, served by a hosted Model Context Protocol server. Point your MCP client at one address and add your API key. Your agent can then read clean product and price data from any shop or from Amazon, see the same product matched across shops, search Google and Amazon, fetch web pages, read the rows saved in your account and, with a Full key, build and run scrapers.

Categories: Web Scraping, Price Monitoring, Product Data, E-commerce

Add to Cursor Add to VS Code

After the one-click install, replace YOUR_SCRAPEWISE_API_KEY with your own key.

  • Endpoint: https://mcp.scrapewise.ai/mcp
  • Transport: Streamable HTTP
  • Auth: Authorization: Bearer <your API key>
  • Registry name: ai.scrapewise/scrapewise (official MCP registry)
  • Docs: scrapewise.ai/mcp

This repository holds the connection manifest, the per-client config for that server, and a small optional stdio bridge (stdio/server.mjs) for clients and tools that only run local servers. The server itself is hosted, so most clients need nothing from here except the address.

Get an API key

Sign up at scrapewise.ai, then go to Settings > API Keys. Pricing is pay-as-you-go.

Your key decides what your agent can do. There is one address for every key.

  • Read key (LLM_READ) can read your groups, scrapers, runs and saved rows, preview changes, and clean rows you already have. It cannot spend credits or change anything.
  • Full key (LLM_FULL) can also run the paid data tools and build, run and change scrapers.

Connect

Replace YOUR_SCRAPEWISE_API_KEY in every snippet below.

Claude Code

claude mcp add scrapewise --transport http https://mcp.scrapewise.ai/mcp \
  --header 'Authorization: Bearer YOUR_SCRAPEWISE_API_KEY'

Use with Claude Code

This repository is also a Claude Code plugin. It adds the scrapewise MCP server and keeps your key out of the config file.

  1. Set your key in the shell that starts Claude Code (Read or Full, from Settings > API Keys):

    export SCRAPEWISE_API_KEY=YOUR_SCRAPEWISE_API_KEY
    
  2. In Claude Code, add this repository as a marketplace source and install the plugin with /plugin:

    /plugin marketplace add BEBOTECH-OU/scrapewise-mcp
    /plugin install scrapewise
    

    To try it from a local clone without installing, run claude --plugin-dir ./scrapewise-mcp.

The plugin sends Authorization: Bearer ${SCRAPEWISE_API_KEY} to https://mcp.scrapewise.ai/mcp. The claude mcp add command above works too if you prefer not to use a plugin.

Claude Desktop

claude_desktop_config.json:

{
  "mcpServers": {
    "scrapewise": {
      "url": "https://mcp.scrapewise.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_SCRAPEWISE_API_KEY"
      }
    }
  }
}

Cursor

Use the Add to Cursor button above, or put the same block in ~/.cursor/mcp.json:

{
  "mcpServers": {
    "scrapewise": {
      "url": "https://mcp.scrapewise.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_SCRAPEWISE_API_KEY"
      }
    }
  }
}

VS Code

Use the Add to VS Code button above, or open the Command Palette, run MCP: Open User Configuration and add:

{
  "servers": {
    "scrapewise": {
      "type": "http",
      "url": "https://mcp.scrapewise.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_SCRAPEWISE_API_KEY"
      }
    }
  }
}

VS Code uses servers, not mcpServers, and needs "type": "http". If either is wrong, VS Code skips the entry without an error.

Gemini CLI

gemini mcp add --transport http scrapewise https://mcp.scrapewise.ai/mcp \
  --header "Authorization: Bearer YOUR_SCRAPEWISE_API_KEY"

Or install this repository as an extension. It asks for your key and keeps it out of the config file:

gemini extensions install https://github.com/BEBOTECH-OU/scrapewise-mcp

LM Studio

Open Program > Edit mcp.json and add:

{
  "mcpServers": {
    "scrapewise": {
      "url": "https://mcp.scrapewise.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_SCRAPEWISE_API_KEY"
      }
    }
  }
}

Any other client

Any client that speaks Streamable HTTP works. Give it the endpoint, the Authorization header and the name scrapewise.

Local stdio bridge (Docker or Node)

For clients that start a local process, this repository has a small stdio server that forwards every tool call to https://mcp.scrapewise.ai/mcp with your key. It lists the tools without a key, but running a tool needs SCRAPEWISE_API_KEY. Without a key, each call returns a clear error and nothing runs.

# Node 20+
git clone https://github.com/BEBOTECH-OU/scrapewise-mcp && cd scrapewise-mcp
npm ci --omit=dev
SCRAPEWISE_API_KEY=YOUR_SCRAPEWISE_API_KEY node stdio/server.mjs

# or Docker (Dockerfile in this repository)
docker run -i --rm -e SCRAPEWISE_API_KEY=YOUR_SCRAPEWISE_API_KEY <image built from this repo>

Optional: SCRAPEWISE_MCP_URL (default https://mcp.scrapewise.ai/mcp) and SCRAPEWISE_CATALOG_URL (default <host>/.well-known/mcp/tools.json).

Public tool list

Anyone can read the tool list, with no key needed, at https://mcp.scrapewise.ai/.well-known/mcp/tools.json. It shows what a Full key sees. The MCP Server Card is at https://mcp.scrapewise.ai/mcp/server-card. The /mcp endpoint itself still asks for a key or an OAuth sign-in.

What your agent can do

AreaWhat it coversKey
ProductsProduct rows from any shop: name, price, currency, availability, brand, SKU/GTIN/EAN, images, description, rating. Amazon product pages, optionally with every seller's priceFull
FindLocal businesses from Google Maps: name, address, phone, website, rating (up to 20 per call). Amazon seller leads. idealo shop offersFull
SearchGoogle web, news and shopping results; Amazon search results; Google Trends over timeFull
FetchThe text or HTML of one public web pageFull
CleanTidy rows you already have: EUR prices, price per kg, litre or piece, trimmed text, no duplicates. Free, nothing is fetchedRead
FollowLong jobs: progress and rows (Read), cancel (Full)Read
ReadYour groups, scrapers, runs, saved rows, match results and settingsRead
Build and runPreview a page, create scrapers from a URL or a cURL command, add product links, run and stop scrapers and groupsFull
Rules and monitoringGroup rules (currency, price per unit), price monitoring, schedules, retention, matching setupFull

Every paid tool states its price and takes quote_only: true, which returns the price and your wallet balance without spending anything.

Tools

The server sends its own tool list when you connect, so that list is always right. On 9 November 2026 a Read key saw 48 tools and a Full key saw 114.

Data tools (Full key): get_product_data · get_amazon_product · find_local_businesses · amazon_seller_leads · idealo_offers · google_search · search_amazon · google_trends · fetch_page · cancel_job

Free tools (both keys): clean_product_data · get_results and every scrapewise_get_*, scrapewise_list_*, scrapewise_search_* and scrapewise_preview_* tool.

Setup tools (Full key): every scrapewise_create_*, scrapewise_update_*, scrapewise_run_*, scrapewise_stop_*, scrapewise_cancel_* and scrapewise_delete_* tool. Each delete has a _preview tool that shows what would go, without deleting anything.

Full documentation: docs.scrapewise.ai.

Manifest

server.json is a mirror. The authority is https://scrapewise.ai/server.json, which is served from the ScrapeWise site repo and checked by its test suite. If the two ever disagree, the hosted one is right.

Namespace ownership for ai.scrapewise/* is proved by the Ed25519 public key published at https://scrapewise.ai/.well-known/mcp-registry-auth.

Support

Licence

MIT, see LICENSE. Covers this repository's contents only, not the hosted service.

No installation packages available.

Remote