mdsearch-mcp

An MCP server that fetches any URL as clean markdown via markdown.new. Uses ~80% fewer tokens than raw HTML.

Built by Esteban Dalel R.

Setup

Claude Code

claude mcp add mdsearch-mcp -- node /path/to/mdsearch-mcp/dist/index.js

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "mdsearch-mcp": {
      "command": "node",
      "args": ["/path/to/mdsearch-mcp/dist/index.js"]
    }
  }
}

Build from source

npm install
npm run build

Tool

fetch_markdown

Fetches a URL and returns its content as clean markdown.

Parameters:

• url (string) — The URL to fetch

Example:

fetch_markdown({ url: "https://example.com" })

How it works

Proxies the given URL through https://markdown.new/<url>, which converts HTML to structured markdown — stripping nav, ads, scripts, and other noise. The result is a compact, LLM-friendly representation of the page content.

License

MIT