Draco: A Rust-based, self-hostable web scraper alternative to Firecrawl

Introduction to Draco

Draco is a newly released web scraper developed in Rust, designed as a self-hostable alternative to existing solutions like Firecrawl and Browserbase. It focuses on providing a fast and efficient method for extracting web content.

Core Functionality

The primary function of Draco is to scrape a given URL and return its main content as clean Markdown, along with associated metadata. This includes handling complex client-rendered Single Page Applications (SPAs) by rendering them before conversion to Markdown, but without the overhead of booting a full headless browser for each request.

Draco employs browser-faithful TLS/JA3 fingerprinting to access pages that might block standard scraping clients. For typical HTML pages, the process involves a single fingerprinted fetch and parse, completing in approximately 300 milliseconds.

Output and Features

The output includes the page's main content as Markdown, with elements like headings, links, lists, and code blocks, while stripping boilerplate content such as navigation, headers, and ads. Metadata extracted includes title, description, language, canonical URL, favicon, and Open Graph/Twitter tags.

Users can install Draco via a shell script for Linux and macOS, or build it from source. It supports outputting to standard output or as a JSON object containing Markdown, metadata, and a trace of the scraping process. Additional features include proxy support and configurable delays for politeness.