Integration
Claude Code + Firecrawl
Firecrawl's official MCP gives your agent real scraping: it converts pages to clean markdown, crawls whole sites and extracts structured data with a schema you define.
Updated 2026-07-27
What you can do
| Action | What it is for |
|---|---|
| Extract a URL | Convert a page to clean markdown, without the HTML noise. |
| Crawl a site | Go through many pages of a domain in a single task. |
| Map URLs | List a site's structure before deciding what to extract. |
| Search the web | Bring back results with their content, not just links. |
| Extract with a schema | Pull specific fields (price, services, contact) in a structured format. |
How to connect it
- 01Create your Firecrawl account and generate the API key.
- 02Add Firecrawl's official MCP server to your client following its documentation.
- 03Pass the API key as an environment variable of the server.
- 04Restart and test by extracting a single URL before launching a full crawl.
Agency use case
Quick audit of a client's competition
- 1.You give the agent 5 competitor sites of the client.
- 2.Firecrawl maps and extracts the services and pricing pages.
- 3.The agent compares offers, promises and calls to action.
- 4.You get a positioning table ready for the meeting.
What you will trip over
- Respect the terms of use and robots rules of the site you crawl. Extracting is not permission to republish.
- A full crawl burns through credits fast: start with map and then extract only what matters.
- For market research, web search is often enough; a full crawl is for when you need the entire site.
Other integrations
FAQ
Frequently asked questions
Firecrawl or Exa?+
Firecrawl shines at extracting and crawling specific sites. Exa shines at broad semantic search. For deep research they complement each other.