Data as of

Firecrawl MCP Server for Claude Desktop: Setup & Config

Firecrawl MCP connects Claude Desktop to web search, scraping, crawling, and live browser interaction.

What you can do with Firecrawl in Claude Desktop

Combine web discovery, structured extraction, live interaction, and recurring monitoring for agents.

  • Extract selected fields from a known web page into JSON.
  • Discover site URLs before choosing pages to scrape.
  • Research a question across multiple sources asynchronously.

Before you start: Firecrawl in Claude Desktop

  • An MCP-compatible client
  • npx for the local package
  • Firecrawl credentials for full cloud API access

The command describes local stdio operation. The remote URL offers a limited keyless tier. Set FIRECRAWL_API_URL for a self-hosted backend, where authentication may be optional.

Firecrawl Claude Desktop configuration

  1. Choose Claude Desktop configuration scope

    For Firecrawl MCP Server, open macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%\Claude\claude_desktop_config.json. Merge under mcpServers via Settings → Developer → Edit Config when available.

  2. Add Firecrawl MCP Server to Claude Desktop

    Use the firecrawl-mcp-server entry below in mcpServers. Keep Firecrawl MCP Server's documented stdio connection fields and your other Claude Desktop entries.

  3. Check Firecrawl MCP Server in Claude Desktop

    Firecrawl MCP Server: Fully quit and reopen Claude Desktop; check developer diagnostics.

Config file · macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %APPDATA%\Claude\claude_desktop_config.json

JSON
{
  "mcpServers": {
    "firecrawl-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "firecrawl-mcp"
      ],
      "env": {
        "FIRECRAWL_API_KEY": "<your Firecrawl API key>"
      }
    }
  }
}

Firecrawl tools available to Claude Desktop

Firecrawl's stored tool list documents 27 tools; compare it with Claude Desktop's connected tool list, because the stored list is not a live check.

firecrawl_scrape

When Claude Desktop requests firecrawl_scrape from Firecrawl, review the action before approving it.

Extract one page into chosen formats, including schema-shaped JSON.

firecrawl_map

When Claude Desktop requests firecrawl_map from Firecrawl, review the action before approving it.

Discover indexed URLs belonging to a website.

firecrawl_search

When Claude Desktop requests firecrawl_search from Firecrawl, review the action before approving it.

Find ranked web, news, or image results with search operators and filters.

firecrawl_search_feedback

When Claude Desktop requests firecrawl_search_feedback from Firecrawl, review the action before approving it.

Record validated quality feedback for an earlier search result set.

firecrawl_feedback

When Claude Desktop requests firecrawl_feedback from Firecrawl, review the action before approving it.

Send compact quality feedback about a completed collection job.

firecrawl_crawl

When Claude Desktop requests firecrawl_crawl from Firecrawl, review the action before approving it.

Collect multiple site pages within configured scope and return final crawl results.

firecrawl_check_crawl_status

When Claude Desktop requests firecrawl_check_crawl_status from Firecrawl, review the action before approving it.

Read progress and available results for an existing crawl.

firecrawl_extract

When Claude Desktop requests firecrawl_extract from Firecrawl, review the action before approving it.

Deprecated extraction entry point; use JSON scraping for known URLs instead.

firecrawl_agent

When Claude Desktop requests firecrawl_agent from Firecrawl, review the action before approving it.

Launch asynchronous multi-source research from a prompt and optional schema.

firecrawl_agent_status

When Claude Desktop requests firecrawl_agent_status from Firecrawl, review the action before approving it.

Check whether an agent research task is still running or has final results.

firecrawl_interact

When Claude Desktop requests firecrawl_interact from Firecrawl, review the action before approving it.

Use a live browser session for navigation, actions, or executable browser work.

firecrawl_interact_stop

When Claude Desktop requests firecrawl_interact_stop from Firecrawl, review the action before approving it.

Release resources by closing an interaction session.

firecrawl_parse

When Claude Desktop requests firecrawl_parse from Firecrawl, review the action before approving it.

Convert supported documents to readable content or schema-shaped output.

firecrawl_monitor_create

When Claude Desktop requests firecrawl_monitor_create from Firecrawl, review the action before approving it.

Schedule recurring collection checks with change comparison.

firecrawl_monitor_list

When Claude Desktop requests firecrawl_monitor_list from Firecrawl, review the action before approving it.

firecrawl_monitor_get

When Claude Desktop requests firecrawl_monitor_get from Firecrawl, review the action before approving it.

firecrawl_monitor_update

When Claude Desktop requests firecrawl_monitor_update from Firecrawl, review the action before approving it.

firecrawl_monitor_delete

When Claude Desktop requests firecrawl_monitor_delete from Firecrawl, review the action before approving it.

firecrawl_monitor_run

When Claude Desktop requests firecrawl_monitor_run from Firecrawl, review the action before approving it.

firecrawl_monitor_checks

When Claude Desktop requests firecrawl_monitor_checks from Firecrawl, review the action before approving it.

firecrawl_monitor_check

When Claude Desktop requests firecrawl_monitor_check from Firecrawl, review the action before approving it.

firecrawl_research_search_papers

When Claude Desktop requests firecrawl_research_search_papers from Firecrawl, review the action before approving it.

firecrawl_research_inspect_paper

When Claude Desktop requests firecrawl_research_inspect_paper from Firecrawl, review the action before approving it.

firecrawl_research_related_papers

When Claude Desktop requests firecrawl_research_related_papers from Firecrawl, review the action before approving it.

firecrawl_research_read_paper

When Claude Desktop requests firecrawl_research_read_paper from Firecrawl, review the action before approving it.

firecrawl_research_search_github

When Claude Desktop requests firecrawl_research_search_github from Firecrawl, review the action before approving it.

firecrawl_developer_search

When Claude Desktop requests firecrawl_developer_search from Firecrawl, review the action before approving it.

Firecrawl: full stored tool descriptions

Troubleshooting Firecrawl in Claude Desktop

  • Fully quit and reopen Claude Desktop for Firecrawl MCP Server. If npx firecrawl-mcp (Node.js/npm) is missing, check the desktop process's PATH, which may differ from your terminal.
  • In Claude Desktop's Firecrawl MCP Server entry, check FIRECRAWL_API_KEY. Keep secrets out of shared screenshots and repositories.
  • Firecrawl MCP Server uses stdio here: Claude Desktop must keep the configured process running; protocol output belongs on stdout and diagnostics on stderr.
  • Known limitation: The keyless tier is rate-limited and excludes several tools.
  • Known limitation: Large crawls and full-page markdown can consume substantial model context.
  • Known limitation: The extract compatibility tool is deprecated in favor of schema-based scraping.
  • General Claude Desktop troubleshooting lives in the /clients/claude-desktop guide.

Firecrawl MCP Server — details. Claude Desktop guide

Frequently asked questions

Claude Desktop loads Firecrawl MCP Server from claude_desktop_config.json: ~/Library/Application Support/Claude/ on macOS, or %APPDATA%\Claude\ on Windows. Open it via Settings → Developer → Edit Config when available.

Claude Desktop's firecrawl-mcp-server example uses stdio.

Firecrawl MCP Server's Claude Desktop configuration invokes npx firecrawl-mcp (Node.js/npm).

Claude Desktop's Firecrawl MCP Server example references FIRECRAWL_API_KEY. Check these names if startup or authentication fails; the example alone does not distinguish required from optional values.

Firecrawl MCP Server's stored list contains 27 tools, including firecrawl_scrape, firecrawl_map, firecrawl_search. Compare those names with Claude Desktop's connected tool list; the stored list is not a live check.

The hosted keyless tier permits rate-limited scrape, search, and interact calls. Capabilities such as crawl, map, and agent need a key.

Yes. FIRECRAWL_API_URL selects a custom backend; provide a key if that instance requires authentication.