Strips HTML down to clean markdown and cuts your token usage by 70-80% in the process. This is what you want when you're scraping web pages for Claude but don't want to burn through your context window on navigation menus, footers, and inline styles. The compression happens by converting raw HTML into structured markdown, dropping all the rendering cruft while keeping the actual content intact. Reach for this when you're building agents that need to read documentation sites, blog posts, or any web content where you care about the text but not the presentation layer. Solid choice for RAG pipelines or any workflow where you're feeding web pages into an LLM.
claude mcp add --transport stdio io.github.robot-resources-scraper uvx scraper