PDF Brain

Claude proposes split points using the same ANTHROPIC_API_KEY as document upload; you can preview and edit boundaries before running the full pipeline. Then: canonical text + stub summary per segment. Run as guest (IP rate limit) or sign in for a higher limit. See pdf-brain/ARCHITECTURE.md.

Files (PDF or images — multiple images may merge when implemented)

Use preview when one PDF may contain several documents — you decide where to split before extraction.