Firecrawl turns pages into text an agent can use: search, scrape, and the remote URL the listing now shows.

Firecrawl is a web server for agents. A person can open a page. An agent needs the text, the links, and sometimes a search result, without the chrome. Firecrawl's MCP is how that work shows up as tools.
The README on the listing describes a set of tools, not one button. Search finds pages. Scrape turns a URL into content. Other tools cover crawling a site, mapping links, and pulling structured extracts. The PlugMCP tools tab lists the names and the one-line descriptions from that README, twenty-two of them on the current crawl.
The remote endpoint we took from the README is https://mcp.firecrawl.dev/v2/mcp. You use Firecrawl's key. The listing does not rewrite that URL onto our gateway.
A browser MCP drives a real tab: clicks, logins, the page you are already signed into. Firecrawl fetches and cleans public pages at a distance. Use Firecrawl when the agent needs content from the open web. Use a browser server when the work only exists inside a session.
On PlugMCP the Firecrawl card is the publisher's server. Hosting a copy is a separate action, for when you want our logs on a server you operate. It is not the Firecrawl install.