MCP server
Firecrawl
Fetch public web pages through a hosted extraction service without confusing extraction with verification or preservation.
Published by the company that makes the product.
What it can access
Sends target URLs and queries to Firecrawl. Depending on configuration, crawling can fetch many linked pages and consume credits.
Security note
URLs can reveal an investigation. Use only public links, set narrow crawl limits, and archive evidence separately. Firecrawl extraction is not a preservation record.
Do you need it?
Connect only for repeated public-page extraction or a bounded crawl. For one page, open it directly or use the assistant’s built-in browser.
Capability and permissions
Firecrawl receives the URLs and search queries you provide, fetches pages on its servers, and returns extracted content. Crawls can follow many links and consume credits.
Least-privilege setup
As of 2026-09-01, use Firecrawl’s official hosted connection or repository. Create a project-specific low-limit key, set a one-page or very small crawl limit, and use only public URLs that reveal nothing sensitive about your investigation.
Harmless success test
Fetch one public page from your organisation’s website. Compare the title, date, links, and body with the browser page. Confirm no linked pages were crawled unexpectedly.
Disable, revoke, remove
Disconnect the MCP, revoke the Firecrawl key or OAuth connection, review usage logs, and delete the test project if it is no longer needed.
When not to connect
Do not submit secret, tokenised, source-provided, intranet, or logged-in URLs. Do not cite Firecrawl output as an archived copy or proof that a page existed at a date.