Give Claude web search, page reading and research workflows for finding and citing external information.
Firecrawl for Claude connects Claude to Firecrawl’s hosted tools for searching the web, reading pages, researching developer documentation and scientific papers, and retrieving structured provider data. Its workflows help with tasks such as researching prospects, collecting a company’s brand assets and checking a website’s SEO.
Describe your task and Claude can select a relevant skill automatically. In Claude Code, you can also invoke dedicated commands for searches, page reads and investigations. The current plugin uses a hosted MCP connection; installing or signing in to the Firecrawl CLI is unnecessary.
Research a question across relevant websites and return a cited answer or report. For programming questions, look up official documentation and release notes for the version you use; for research questions, find papers and read the passages supporting their findings.
Find professionals matching your target roles and companies, or enrich people already supplied in a list or CSV. Results include source attribution and unresolved matches; organizational inboxes have a separate workflow, and these workflows do not send outreach.
Collect an organization’s existing logos, colors, fonts and homepage messaging for a presentation. For a website implementation, extract a reference site’s visual system into DESIGN.md, distinguishing observed values from estimates and proposed choices.
Inspect selected pages for issues with titles, descriptions, headings, canonicals, structured data and internal links. Get findings tied to affected URLs, with prioritized fixes and verification suggestions; scrape output alone does not establish traffic, rankings or indexing status.
Search for jobs matching your role, location and requirements, then check employer or official application pages before including openings in the verified shortlist. Results retain application links, stated compensation and eligibility conditions.
Research player news, usage and matchups under your league’s scoring rules to compare starters, pickups or trades. Recommendations remain conditional where availability or status is uncertain; the workflow does not change your roster.
Search external sources, read URLs and choose a focused research workflow.
Collect existing logos, colors, fonts and messaging for documents.
Find published organizational inboxes, business phone numbers and contact pages.
Document a reference website’s visual system in DESIGN.md.
Add sourced professional details to supplied people or lead CSVs.
Research fantasy decisions using player news, matchups and league scoring.
Discover and rank professionals against your prospect criteria.
Investigate questions across sources and produce cited answers, reports or tables.
Find matching jobs and check whether employer postings accept applications.
Answer version-specific development questions from documentation and repository evidence.
Find scientific papers, read relevant passages and follow citation relationships.
Inspect website SEO issues and return evidence with prioritized fixes.
/firecrawl:agentResearch a question across sources and return a cited answer.
Choose agent from the Firecrawl commands in Claude Code’s slash menu and supply your question or research task. It uses the deep-research workflow or a more focused workflow when one fits. Command reference.
/firecrawl:crawlRead relevant site pages individually; coverage may be incomplete.
Choose crawl from the Firecrawl commands in Claude Code’s slash menu and supply a site URL, optionally with a topic. It finds relevant pages and reads them individually, with confirmation before reading more than ten. Reads are billed per URL, and the combined result may miss pages. Command reference.
/firecrawl:mapFind site URLs through search and links; coverage may be incomplete.
Choose map from the Firecrawl commands in Claude Code’s slash menu and supply a site URL, optionally with a topic. It returns URLs found through site-restricted search and available page links. This connection has no full-site mapping tool, so the list may be incomplete. Command reference.
/firecrawl:scrapeRead a webpage and return its content or requested details.
Choose scrape from the Firecrawl commands in Claude Code’s slash menu and supply a webpage URL. Add what you want extracted if you need particular details rather than the page content. Command reference.
/firecrawl:searchSearch the web and summarize findings with source citations.
Choose search from the Firecrawl commands in Claude Code’s slash menu and supply a query, including any date, site or regional constraints. It reads result pages when excerpts do not provide enough context. Command reference.
/firecrawl:setupCheck the account connection and guide sign-in when needed.
Run this command in Claude Code to check tool availability and make a free discovery call. If authentication fails, it directs you to the MCP connection manager to select Firecrawl and sign in through your browser. You do not need to paste an API key into the conversation.
/firecrawl:setupConnect to hosted tools for web search, page reading and research.
Install the whole plugin from Firecrawl’s own repository marketplace. Use Claude Code on your computer; locally installed plugins do not load in Claude Code cloud sessions. If you are switching from an official-marketplace installation of Firecrawl, disable that installation first.
In a Claude Code session, add Firecrawl’s marketplace.
/plugin marketplace add firecrawl/firecrawl-claude-pluginOpen the Firecrawl installation panel, choose the scope you want and install the plugin.
/plugin install firecrawl@firecrawlFollow the installation summary. If it reports that the plugin is active, continue; if a reload remains pending, apply it in the session.
/reload-pluginsOpen the MCP connection manager, select firecrawl and complete account sign-in in your browser. This plugin does not require a Firecrawl CLI installation or an API key pasted into the conversation.
/mcpCheck the connection. The setup command checks tool availability and makes a free discovery call; a passing check confirms that the account connection responds.
/firecrawl:setupUse the release offered by Claude Code’s official marketplace, which may differ from the repository release. Choose one Firecrawl installation. This route uses Claude Code on your computer; locally installed plugins do not load in cloud sessions.
Run the installation command in your computer’s terminal.
claude plugin install firecrawl@claude-plugins-officialStart Claude Code to load the plugin. If a session is already open, reload its plugins instead.
/reload-pluginsIn Claude Code, open the MCP connection manager, select firecrawl and complete account sign-in in your browser. The current hosted plugin does not need the Firecrawl CLI.
/mcpRun the setup check. It checks that Firecrawl’s tools are available and makes a free discovery call to confirm the account connection.
/firecrawl:setupUse Claude Desktop or Cowork with a paid Claude plan that supports skills. Turn on Code execution and file creation so the plugin’s skills can run. Installing the plugin and connecting its Firecrawl tools are separate actions.
Open Customize > Plugins, choose Add > Add marketplace, and add the Firecrawl repository: https://github.com/firecrawl/firecrawl-claude-plugin. Select firecrawl from that marketplace and install it.
Open the installed plugin’s Connectors tab, connect Firecrawl and complete the sign-in flow.
Check Customize > Skills > Your skills for the plugin’s skills, and confirm Firecrawl appears under Customize > Connectors > Your connectors with status Connected. In your conversation, open the message box’s Connectors menu and enable Firecrawl.