What problem does it solve? Deep reading of long academic papers often loses continuity: readers forget earlier claims, skip unresolved questions, and cannot resume a session without rereading. This Skill enforces a disciplined sequential reading loop that preserves a cumulative mental model across chunks and sessions. ## Core Features & Use Cases - One-Chunk-at-a-Time Reading: Reads exactly one ordered semantic chunk, explains its claim, evidence, and role, and never reads ahead or batches unread chunks. - Persistent Reader State: Maintains Reader State, a glossary, a question ledger with typed questions (clarification, mechanism, critique, etc.), and per-chunk notes so a fresh session can resume without hidden context. - Answer Integration and Chunk Gating: Integrates answers from a separate Answerer, classifies questions as resolved, deferred, open research, or unanswerable, and only marks a chunk complete when no blocking question remains. - Use Case: A researcher working through a dense machine learning paper uses the Skill to read section by section over multiple days, accumulating linked questions and a durable mental model without losing track of earlier derivations. ## Quick Start Use the read-paper-sequentially skill to continue the Reader session from the next pending chunk in my paper reading workspace.
