• New extractors for LinkedIn, Threads, Bluesky, Discourse, Medium
• Footnotes refactor with sidenote support and more patterns
• Content boundary detection and eyebrow removals
• H1 fallback, title normalization
• Code blocks: fix duplicate language name (#235)
• Metadata: rel=author fallback, date deduplication, author name cleanup
• Keep content grids with lots of content
• Audio/video source parity, empty video placeholder removal
• X extractor refactored to use comments for replies