Use FireScraper with Claude: Crawl and Query the Web in Claude Code & Desktop

June 29, 20264 min read

guide

mcp

claude

ai-agents

Claude is a strong reasoning partner, but on its own it can't fetch a live web page. The FireScraper Model Context Protocol (MCP) server gives Claude that ability: crawl a site, get clean Markdown, and ask questions grounded in the scraped content — directly in Claude Code or Claude Desktop.

This guide covers installing the server in both, the prompts you'll use, and what comes back.

What you'll be able to do

Ask Claude to crawl a URL and reason over the real page
Pull entire docs sites into clean, LLM-ready Markdown
Build a searchable corpus and ask it questions with cited sources
Keep it all inside your normal Claude workflow

Prerequisites

Node.js 18+ (the server runs via npx)
A FireScraper API key from the API keys page. Keys start with fsk_; new accounts include free crawl units.

Creating a FireScraper API key — Create a key on the API keys page — it's shown only once.

Install in Claude Code

One command:

claude mcp add firescraper --env FIRESCRAPER_API_KEY=fsk_your_key -- npx -y @firescraper/mcp

Restart the session and the FireScraper tools are available.

Adding the FireScraper MCP server in Claude Code, then a prompt and grounded answer — One claude mcp add command — then ask, and Claude crawls and answers with sources.

Install in Claude Desktop

Add the server to your MCP config (claude_desktop_config.json):

{
  "mcpServers": {
    "firescraper": {
      "command": "npx",
      "args": ["-y", "@firescraper/mcp"],
      "env": {
        "FIRESCRAPER_API_KEY": "fsk_your_key"
      }
    }
  }
}

Restart Claude Desktop. You'll see the FireScraper tools in the tools menu.

The tools Claude gets

Tool	What it does
firescraper_scrape_and_wait	Crawl URLs and return the text in one call — the easiest path.
firescraper_scrape	Start a large crawl and return a session ID immediately.
firescraper_get_session	Check crawl status and page counts.
firescraper_list_results	List the export files available for a finished crawl.
firescraper_get_results	Download output in a chosen format (markdown, json, chunks, vectors, …).
firescraper_ask_corpus	Ask a finished crawl a question; get an answer grounded in its content, with sources.

Prompts you'll actually use

You just ask in plain language — Claude picks the right tool.

Summarize a live page:

Crawl https://modelcontextprotocol.io/introduction and explain what MCP is
and when I'd build a server vs. a client.

Bring a whole site into the conversation:

Crawl https://docs.myvendor.com with depth 2 and give me the Markdown for
the authentication section.

Build and query a knowledge base:

Crawl https://help.acme.com at depth 2. When it's finished, ask the corpus:
"How do I rotate API keys, and what happens to the old key?"

Claude runs the crawl, waits for it to finish, then calls firescraper_ask_corpus and answers from the scraped pages with cited source URLs.

A grounded answer from a crawled corpus with cited sources — A grounded answer with sources, drawn from the crawl.

What you get back

Clean Markdown that keeps headings, lists, tables, and links
Structured exports (documents, chunks, vectors) for RAG pipelines
Grounded answers with citations from ask_corpus

Tips

Use firescraper_scrape_and_wait for quick lookups; for large sites, let Claude start the crawl and poll with firescraper_get_session.
Crawls cost credits per page and ask_corpus a few credits per question — watch your balance on big jobs.
Tune depth: 0 = seed URLs only, 2 = follow links two hops in.

Give Claude the live web

Create a free FireScraper API key and connect the MCP server in two minutes.

Get your API key Read the docs