Servers / Firecrawl

Install Firecrawl MCP in Cursor

Firecrawl · Local stdio · API key

Cursor mcp.json

Paste this into ~/.cursor/mcp.json (global) or.cursor/mcp.json (this repo). Merge it with servers you already have. Primer: how Cursor loads MCP.

{
  "mcpServers": {
    "firecrawl": {
      "command": "npx",
      "args": ["-y", "firecrawl-mcp"],
      "env": {
        "FIRECRAWL_API_KEY": "YOUR-API-KEY"
      }
    }
  }
}

Source: Firecrawl MCP.

What it is

Firecrawl turns a URL into markdown or structured data. Context7 is library docs. Firecrawl is everything else on the web: a pricing page, a changelog, a random README that is not in Context7.

The problem it solves

Cursor's model will happily summarize a page it cannot fetch. Firecrawl is the fetch.

Install steps

1. Get an API key. Create a key at firecrawl.dev. Put it in FIRECRAWL_API_KEY. Do not commit the key. Prefer the global mcp.json over a repo file for secrets.
2. Add the stdio server. command npx, args -y firecrawl-mcp, env with the key. Trust the dialog only if you meant to run that package.
3. Scrape one URL. Ask Cursor to scrape a docs URL and quote the install section.

A prompt that uses it

Scrape https://docs.that-library.dev/getting-started and extract the current install command. Do not use memory of an older version.

If it will not connect

A missing or invalid FIRECRAWL_API_KEY fails the tools even if the process starts. npx must reach the registry. If the package name in Firecrawl's current docs differs, use their snippet, not an old blog.

FAQ