visualizevalue/context-score — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a simple command line tool that can scan any website URL and tell me how well the site explains itself to people, search engines, and AI tools.

I want to run it with something like a URL in the terminal and get a clean score out of 100, with a letter grade and five sections, identity, structure, depth, connectivity, and discoverability. It should crawl pages, check things like sitemap, robots.txt, meta descriptions, about page, schema, internal links, orphan pages, word count, speakable markup, social links, and AI discovery files.

Make the output easy to read, with progress bars and a short list of the biggest opportunities to improve the site. Add a verbose mode that shows page by page details, orphan pages, and how sections link to each other.

Keep it lightweight and usable with npx. Look up current docs online if you need to.

Want more depth? Deep Reverse