Search Only MCP Endpoint

The Firecrawl MCP Server at https://mcp.firecrawl.dev/v2/mcp is the default endpoint and provides the complete Firecrawl tool set.

Endpoint

https://mcp.firecrawl.dev/v2/mcp-search

Every request requires authentication: connect with OAuth using the endpoint directly, or pass a Firecrawl API key in the Authorization header.

{
  "mcpServers": {
    "firecrawl-search": {
      "url": "https://mcp.firecrawl.dev/v2/mcp-search",
      "headers": {
        "Authorization": "Bearer fc-YOUR_API_KEY"
      }
    }
  }
}

Available tools

The search-only endpoint exposes exactly these six read-only tools:

Tool	Purpose
`firecrawl_search`	Search the web and return ranked results
`firecrawl_research_search_papers`	Search research papers
`firecrawl_research_inspect_paper`	Get metadata for a paper
`firecrawl_research_related_papers`	Find related papers through citation links
`firecrawl_research_read_paper`	Read full-text passages from a paper
`firecrawl_research_search_github`	Search public repository history and README files

Differences from the full MCP endpoint

The tool set is fixed and read-only.
Scrape, map, crawl, extract, agent, interact, parse, monitor, and feedback tools are not available.
firecrawl_search does not accept scrapeOptions and cannot request page-content fetching.

Get Started

Core Endpoints

More

Quickstarts

Developer Guides

Webhooks

Use Cases

Other

Contributing

Endpoint

Available tools

Differences from the full MCP endpoint

​Endpoint

​Available tools

​Differences from the full MCP endpoint

Endpoint

Available tools

Differences from the full MCP endpoint