servo-fetch

servo-fetch is a self-contained browser engine built on Servo. This skill tells your agent when and how to use it: for pages that come back empty from a plain HTTP fetch, screenshots without a GPU, and running JavaScript in a page. - **Fetch** a page as clean Markdown, JSON, HTML, text or an accessibility tree. JavaScript runs, layout is computed, and navbars, sidebars, footers and cookie banners are stripped. - **Batch fetch** up to 20 URLs in parallel. - **Crawl** a site's same-site links breadth-first, respecting robots.txt, with include and exclude patterns. - **Screenshot** a page, or its full height, with a software renderer. - **Run JavaScript** after the page loads and get the result with its console output. - **PDFs** are detected and extracted directly. It's read-only: it doesn't click or fill in forms. Your agent uses it through the servo-fetch CLI or its built-in MCP server (`servo-fetch mcp`). Install the CLI with the script in the README, `cargo binstall servo-fetch-cli` or a release download. No API key. MIT or Apache-2.0 licensed.