Can a prompt replace Screaming Frog SEO Spider?

Build a local site crawler for technical SEO audits, scoped to sites you own or have explicit permission to crawl — require that acknowledgement before a crawl starts, and don't skip it. Use Python with FastAPI, Playwright for rendering (needed for accurate results on JavaScript-heavy sites), and SQLite for storing crawl results. Respect robots.txt and a configurable rate limit and URL cap before crawling starts. For each crawled page, extract: HTTP status, title, meta description, H1/H2 headings, canonical URL, robots meta directives, internal and external links, image alt text, and basic structured-data presence. From the crawl, compute and flag: duplicate titles or descriptions across pages, broken internal links, redirect chains longer than one hop, missing or conflicting canonical tags, and pages with no incoming internal links (orphans). Show every flagged issue with the specific affected URLs, the evidence (e.g. the actual duplicate title text), and a concrete one-line fix suggestion — vague warnings without evidence aren't useful. Export the full crawl data to CSV and a self-contained HTML report that doesn't need the app running to view. Do not build crawling of sites without permission, scheduled or continuous monitoring, or backlink/keyword datasets — those are out of scope; this is a one-shot local audit tool for your own site. No API key or account needed; everything runs locally.