deBUGger404/Scrape-AI — reverse-engineered prompt

Build me a small Python library that can scrape a website based on a plain English request, kind of like an agent that reads the page and pulls out just the info I asked for. It should work on normal pages and dynamic sites that need a real browser, so use Selenium for page interaction, with an option to run headless or show the browser.

I want to be able to pass in a config with the target URL, my prompt, whether logging is verbose, and the LLM settings for providers like OpenAI, Anthropic, Azure, or Google. Then I should be able to create something like a WebScraper object and call one method to get back structured JSON with the answer. A good example is pointing it at a search page and asking for all library names plus their install commands.

Please make it feel reusable as a package, keep the design modular so different model providers can plug in cleanly, and include a basic test and README example. If anything is unclear, check current docs online.