firecrawl

Firecrawl MCP Server exposes Firecrawl's web scraping capabilities as an MCP-compatible service. It integrates cloud browser-based scraping, discovery, and content extraction, with features like automated retries, rate limiting, and support for both cloud and self-hosted deployments. The server can be launched via npx and is designed to be consumed by the MCP framework (e.g., through Cursor, Windsurf, or other MCP clients). Once running, you can describe your scraping needs to the Composer Agent and it will route requests to Firecrawl for tasks such as page discovery, data extraction, and bulk scraping, including support for cloud browser sessions and SSE transport.

You can operate it using the standard MCP tooling: start the server with npx, supply your Firecrawl API key, and then interact with it through MCP-compatible clients. The server supports configuring a custom API URL for self-hosted Firecrawl deployments via FIRECRAWL_API_URL, and it offers options for retries and credit monitoring to manage usage and reliability.

Prerequisites:

• Node.js (recommended latest LTS) and npm installed on your system
• Access to a Firecrawl API key (for cloud API)

1. Install the MCP server globally via npm (one-time):

npm install -g firecrawl-mcp

2. Run the MCP server using npx (supply your API key):

env FIRECRAWL_API_KEY=YOUR_API_KEY npx -y firecrawl-mcp

3. Alternative: run directly with npm and npx in your environment (without global install):

env FIRECRAWL_API_KEY=YOUR_API_KEY npx -y firecrawl-mcp

4. If you need to configure for a self-hosted Firecrawl instance, set the API URL and credentials accordingly:

env FIRECRAWL_API_URL=https://firecrawl.your-domain.com \
    FIRECRAWL_API_KEY=YOUR_API_KEY \
    npx -y firecrawl-mcp

5. Optional: run with Streamable HTTP local mode (local transport):

env HTTP_STREAMABLE_SERVER=true FIRECRAWL_API_KEY=YOUR_API_KEY npx -y firecrawl-mcp

6. For VS Code / Cursor / Windsurf setups, follow the configuration examples in the README to embed the same command and environment variables in your editor-specific config.

Environment variables and configuration:

• FIRECRAWL_API_KEY: Required for the cloud Firecrawl API. Use the key provided by your Firecrawl account.
• FIRECRAWL_API_URL: Optional. Use for self-hosted Firecrawl instances. If omitted, the cloud API is used.
• FIRECRAWL_RETRY_MAX_ATTEMPTS, FIRECRAWL_RETRY_INITIAL_DELAY, FIRECRAWL_RETRY_MAX_DELAY, FIRECRAWL_RETRY_BACKOFF_FACTOR: Retry configuration for failed requests (defaults shown in docs).
• FIRECRAWL_CREDIT_WARNING_THRESHOLD and FIRECRAWL_CREDIT_CRITICAL_THRESHOLD: Monitoring thresholds for credit usage.

Tips:

• Ensure the API key is kept secret and not committed to source control.
• When running in environments like Cursor or Windsurf, replicate the exact command and env block as shown in the README to avoid configuration drift.
• If you encounter Windows command issues, wrap the command with cmd /c and set the environment variable in the same shell, e.g., cmd /c "set FIRECRAWL_API_KEY=... && npx -y firecrawl-mcp".