Firecrawl
The web, turned into clean data
Turn any website into clean data your AI can read
Official Firecrawl server: scrape a page, search the web, or crawl a whole site.
Firecrawl turns messy web pages into clean, ready-to-use content — the kind of text an AI can actually work with. Its official connector, per the README, is an "MCP server for Firecrawl — search, scrape, and interact with the web", with tools to scrape a single page, search the web, crawl an entire site, map all its URLs, and extract structured data from many pages at once.
Ask "read this article and summarize it for me" with a link, and it fetches the page as clean text and answers. Ask "find the pricing pages of these five competitors" and it searches, opens each, and pulls out what matters — no copy-pasting, no broken formatting from a normal browser copy.
What you can ask
- "Read https://example.com/blog/post and give me the three main points in plain language."
- "Search the web for 'best CRM for small law firms 2026' and summarize what the top results agree on."
- "Go through this competitor's site and list every product they sell with its price."
What token you need
You need a Firecrawl API key — it starts with fc-.
- Go to firecrawl.dev/app/api-keys and create a free account if you don't have one.
- Copy your key (it looks like
fc-…). - Paste it when the Lab asks for
FIRECRAWL_API_KEY. It's stored encrypted in your Keychain.
The free tier is enough to try it. Heavy crawling of large sites uses more of your Firecrawl credits, so keep an eye on your usage in their dashboard.
We open the app and set it up in Claude Code (and your other Macs).
How it works with Terminal Sync
Set up the connector once on one machine. Terminal Sync keeps your claude_desktop_config.json encrypted in your Drive, so on any other machine where you open Claude Code, the connector is already there.