- Fix: title extraction now reads h2 text directly (h2 has no child <a>)
- Fix: link extraction uses a.sf-search-ad-link overlay (was falling through to carousel)
- Fix: skip skeleton/placeholder articles
- Add: --verdict flag with buy_used/buy_new/wait/decent_deal recommendation
- Add: verdict compares median used price vs lowest new price with savings %
- Add: verdict works standalone (no --compare needed)
- Rewrite: README + USAGE docs for the 'go-to' workflow
- Refactor: shared _parse_dba_article() for requests + selenium paths
14828ddd0f