What changed in v1.1.0 (optimized with Claude Fable 5)

Three areas have been improved:

More reliable data collection

• The extractor used to be all-or-nothing: if one part failed, everything failed. Now each section runs independently — if button detection breaks on a weird site, you still get colors, typography, and everything else. The report tells you which parts degraded so you know what to trust.

Faster runs, lower cost

• Tall pages (like Stripe at 16× the viewport) used to generate a massive full-page screenshot that burned tokens without adding much insight. Now we check the page height first and only take targeted shots where they matter — hero, mid-page, and footer.

• The data pipeline went from three separate passes over the page to one, cutting redundant work by ~2/3.

• When asking setup questions (export target + crawl scope), we now ask once instead of twice.

Better analysis quality

• The extractor now captures data it previously missed: how often each font size actually appears, which spacing values dominate, button styling patterns, and how much area each background color covers. This gives the AI more to work with when identifying what makes a design distinctive.

• The "anti-slop" filter (which catches generic phrases like "clean and modern") now covers more filler words, so the final output stays specific and evidence-based.

• Each analysis step has output caps to keep things focused — fewer words, more signal.