What problem does it solve? When surveying dozens of papers for an LLM research project, notes end up scattered and inconsistent, making it hard to compare methods, datasets, and metrics or to spot coverage gaps and contradictory claims. This Skill consolidates deep-reading notes into a fixed schema and generates a unified comparison matrix. ## Core Features & Use Cases - Fixed paper note schema: Enforces a structured template (Problem, Method, Dataset, Metric, Claim, Limitation, Replicability Checklist) so every paper note is machine-parseable. - MATRIX.md generation: Parses all notes and produces a Markdown comparison table plus coverage analysis (dataset/method histograms) and contradiction detection across papers. - Parallel-safe coordination: Uses a claim column in papers.jsonl as a mutex so multiple paper-deep-reader agents can work concurrently without duplicating effort. - Use Case: During Phase 2 of an auto-research survey, after several agents finish deep-reading arXiv papers, run this Skill to merge their notes into a single comparison matrix revealing that GSM8K × DPO has only one supporting paper. ## Quick Start Use the research.literature.matrix skill to aggregate the paper notes in .research/<slug>/02_SURVEY/notes into MATRIX.md and update papers.jsonl.
