MCP Web Search & Scrape in Crush

​

Requirements

• Crush 0.98 or newer.
• A Firecrawl API key, exported as FIRECRAWL_API_KEY.

​

Quick setup

1. Export your API key in the shell you start Crush from:

export FIRECRAWL_API_KEY="fc-YOUR-API-KEY"

2. Add the server to crush.json in your project, or to ~/.config/crush/crush.json for every project:

{
  "$schema": "https://charm.land/crush.json",
  "mcp": {
    "firecrawl": {
      "type": "http",
      "url": "https://mcp.firecrawl.dev/v2/mcp",
      "headers": {
        "Authorization": "Bearer $FIRECRAWL_API_KEY"
      }
    }
  }
}

3. Start Crush and ask it to search or scrape.

mcp add firecrawl \
  --type http \
  --url "https://mcp.firecrawl.dev/v2/mcp" \
  --header Authorization "Bearer $FIRECRAWL_API_KEY"

​

Try it

Search the web for the latest Rust release notes and summarize the sources.

Scrape https://firecrawl.dev and explain what it does.

​

What you can do

​

Troubleshooting

• The server does not start: check that FIRECRAWL_API_KEY is exported in the shell that starts Crush, and that the URL is exactly https://mcp.firecrawl.dev/v2/mcp.
• The header is dropped: Crush removes a header whose value resolves to an empty string. An unset FIRECRAWL_API_KEY causes this.
• Tools are not used: some models call MCP tools more reliably than others. Try a different model.