CyberScraper-2077

Why we included this project

If you regularly pull data off the web, this tool replaces hand-written selectors with a conversation: you give it a URL, describe what you want, and an LLM (OpenAI, Gemini, or a local Ollama model) finds and structures the content before handing it back as JSON, CSV, SQL, Excel, or HTML. Teams doing repeated extraction for market research, price tracking, or lead building get a Streamlit interface that needs no code, while developers can run it as a Docker service and fit it into their own pipelines. It also handles paginated sites with page-range and multi-page scraping, caches requests to keep LLM API costs down, and offers optional Tor and stealth routing for sites that push back on scrapers. Expect some prompt tuning, especially with local models, but for turning messy pages into clean, labeled data it works out of the box.