v0.1.1 — Vector PDF Figure Rasterization + NSA Paper Validation

v0.1.1 — Vector PDF Figure Rasterization

New

• render_figures.sh: PyMuPDF-based PDF figure → PNG conversion
• Auto-detects raster vs vector PDF figures
• Vector PDFs rasterized at 200 DPI (no external tools needed)
• Resizes oversized images to 1200px max width

Validated

• Successfully tested on NSA paper (ACL 2025 Best Paper, arXiv 2502.11089)
• Produced 16-page annotated deep-read with 8 figures + 8 formulas
• All 8 PDF figures correctly converted (4 were vector-only)

Documentation

• SOP.md: Phase 2.5 Figure Rasterization step added
• SKILL.md: render_figures.sh added to directory structure
• Known Limitations updated: vector PDF + PyMuPDF note

Commentary Note

• Current commentary depth is adequate but could be richer
• Future version will strengthen web-research integration for deeper insights