What problem does it solve? Academic papers are too long to read or analyze in one pass, and naive splitting breaks claims away from their evidence, equations, or figures. This Skill partitions an ingested paper into contiguous, ordered chunks that each contain one complete argument unit while guaranteeing no source block is lost, duplicated, or reordered. ## Core Features & Use Cases - Semantic Chunking: Walks source blocks in order and groups each claim with its direct justification, equation derivation, figure/table evidence, or qualification. - Coverage Validation: Flattens all chunk source-block lists and verifies they exactly match the source map IDs in order, rejecting omissions, duplicates, and orphaned headings. - Structured Output: Writes one Markdown file per chunk under 01 Chunks/ with frontmatter metadata (role, pages, dependencies, assets), plus chunk-map.json and a readable index. - Use Case: After ingesting a research paper into a reading workbench, run this Skill to rebuild poor chunks so each chunk focuses on a single argument unit before handing them to a reader for sequential study. ## Quick Start Use the segment-paper skill to partition the ingested paper into ordered semantic chunks and validate complete source-block coverage.
