zytedata/claude-skills — reverse-engineered prompt

Build me a Claude Code plugin for Zyte that lets someone type a normal scraping request with a website URL and get back a working Scrapy spider.

The main flow should feel guided. It should explore the site, find pages with repeating structured content like products, jobs, or recipes, suggest the fields it can extract, ask me to approve or adjust the schema, then generate the project. I want it to create the page objects, spider, fixtures, dependencies, and basic tests automatically, then run a smoke test so I know the scraper actually works before it says it’s done.

The generated scraper should be a normal Scrapy project that can run locally with uv, not something locked inside Claude Code. Please include a simple install and usage flow for the plugin, a slash command like scrape, and an optional path for sending the finished spider to Scrapy Cloud for scheduled runs and monitoring.