Ask Scraper Finder from your agent.

Add https://scraperfinder.com/mcp to an MCP client and your agent gets the same answer a person gets on a website page: the official API route, the ranked tools, what each costs per 1,000 results and which fields it returns.

In one line

A free, read-only MCP server (streamable HTTP, no sign-in) with three tools over 172 websites (170 with ranked tools) and the published output fields of 415 tools.

Checked 2026-11-13. The answers are the site's published data, refreshed with every publish.

Add it to a client

One URL, no headers. If a client asks about authentication, choose none.

Claude Code

claude mcp add --transport http scraper-finder https://scraperfinder.com/mcp

Claude Desktop and claude.ai

Customize > Connectors > Add custom connector
Name: Scraper Finder
URL:  https://scraperfinder.com/mcp

Cursor (~/.cursor/mcp.json)

{
  "mcpServers": {
    "scraper-finder": {
      "url": "https://scraperfinder.com/mcp"
    }
  }
}

VS Code (.vscode/mcp.json)

{
  "servers": {
    "scraper-finder": {
      "type": "http",
      "url": "https://scraperfinder.com/mcp"
    }
  }
}

Codex CLI (~/.codex/config.toml)

[mcp_servers.scraper_finder]
url = "https://scraperfinder.com/mcp"

Gemini CLI

gemini mcp add -t http scraper-finder https://scraperfinder.com/mcp

The three tools

Tool	What it answers	Try asking
`which_tool`	One website's routes: the official API (status, price and limits notes, docs) beside the ranked tools with score, price per 1,000 and core-field coverage, plus robots.txt and terms facts.	What's the cheapest way to get Zillow listings with bedrooms and price?
`find_by_field`	Tools whose published output includes the fields you name, optionally for one website.	Which tools return a salary field for job posts?
`list_websites`	The 172 websites mapped, filterable by name or by official API status.	Which websites have no open API at all?

Tools are ranked by the formula on How we rank. Access facts come from What it takes: one plain request per website, which says nothing about CAPTCHAs, rate limits or login walls.

Technical details

Endpoint https://scraperfinder.com/mcp, MCP streamable HTTP. Opened in a browser, it shows a short description.
Protocol versions 2024-11-05 to 2025-11-25 (with initialize) and the stateless 2026-07-28 revision (server/discover). No sessions.
Read-only tools with input schemas and readOnlyHint. No prompts, resources or sampling.
Fair use: about two requests a second per address, short bursts allowed; over the limit the server answers HTTP 429.
Registry name com.scraperfinder/scraper-finder. Manifest and README: github.com/bolvein-firex/scraper-finder-mcp.

Questions people ask

Does it scrape anything?

No. It answers from the data this site publishes: official API facts, the ranking and published output fields. Running a tool happens on the tool's own platform with your account.

Is it free? Do I need a key?

Free and open, no key or sign-in. It is read-only, with a fair-use rate limit per address.

How current are the answers?

They come from the same data file the site publishes, so they change when the pages do. Each answer carries the check dates of its facts.

Page updated 2026-11-13