MarkGrab

This server wraps the markgrab Python library to turn arbitrary URLs into clean markdown that Claude can work with directly. It handles HTML articles with content density filtering, YouTube transcripts with timestamps, PDFs, and DOCX files. The extraction is async first, tries httpx for static pages, and falls back to Playwright when JavaScript rendering is needed. You'd reach for this when you want Claude to read web content without manually copying and pasting, or when you need to process documents from links in a conversation. The CLI and Python API support options like max character limits, forced browser rendering, and stealth mode for bot detection.