sitemap

This Sitemap MCP Server provides tools to fetch, parse, and analyze website sitemaps. It exposes a set of MCP commands for working with sitemap data, including getting a full sitemap tree, listing sitemap pages with optional filtering and pagination, retrieving sitemap statistics, and parsing sitemap content directly from XML or text inputs. In Claude Desktop (via MCP), you’ll also find ready-to-use prompt templates that guide you through analyzing sitemap structure, checking health, extracting URLs, identifying missing content, and visualizing the sitemap as diagrams. To use the server, install it through the recommended transport (uvx) and then call the available tools by name with the required arguments (for example, a URL to analyze). The included prompts help you perform common sitemap tasks with minimal input: provide a URL and the server returns structured JSON representing the sitemap hierarchy, pages, and stats, which you can then visualize or further filter.

Prerequisites:

• Python 3.11+ (and uv installed)
• Access to uvx transport for MCP servers

Install and configure the server in Claude Desktop, Cursor, or Windsurf:

# In Claude Desktop/Cursor/Windsurf configuration (claude_desktop_config.json)
{
  "mcpServers": {
    "sitemap": {
      "command": "uvx",
      "args": ["sitemap-mcp-server"],
      "env": { "TRANSPORT": "stdio" }
    }
  }
}

Option 2: Install via Smithery (for automatic client setup):

npx -y @smithery/cli install @mugoosse/sitemap --client claude

MCP Inspector setup (optional):

# uv + stdio transport
npx @modelcontextprotocol/inspector env TRANSPORT=stdio uvx sitemap-mcp-server

Alternatively, for other transports you may use the provided snippets in the README to start with different configurations (e.g., SSE transport) but the standard local development path is to run via uvx sitemap-mcp-server after installing the package.

Local development and source code access: refer to DEVELOPERS.md in the repository for building and running from source.

Notes and tips:

• The server exposes these tools: get_sitemap_tree, get_sitemap_pages, get_sitemap_stats, parse_sitemap_content. Use the exact tool names when constructing MCP calls.
• When using MCP cursor-based pagination (get_sitemap_pages), include limit and optional cursor values as shown in the examples to iteratively fetch large sitemaps.
• For SSE transport, you can configure the MCP client with transport: sse and a url like http://localhost:8050/sse; adjust your environment accordingly.
• If you’re running locally, ensure TRANSPORT is set to the expected value in your client configuration to enable proper MCP communication.
• For Claude Desktop templates, look for prompts like Analyze Sitemap, Check Sitemap Health, Extract URLs from Sitemap, Find Missing Content in Sitemap, and Visualize Sitemap Structure to quickly run common analyses.