Takes a Deepwiki URL via MCP, crawls all relevant pages, converts them to Markdown, and returns either one document or a list by page.