What problem does it solve? Before filing a patent, inventors need to know whether their idea is novel and what existing patents or papers might block it. Manually searching Google Patents, Espacenet, and academic databases is slow and easy to do incompletely. ## Core Features & Use Cases - Multi-database patent search: Queries Google Patents and Espacenet using extracted keywords and predicted IPC/CPC classification codes, in both English and the target jurisdiction language. - Academic literature coverage: Searches Google Scholar, arXiv, and Semantic Scholar for non-patent prior art relevant to novelty and inventive step. - Structured risk analysis: Classifies each reference by overlap risk (HIGH/MEDIUM/LOW) and produces a preliminary freedom-to-operate assessment. - Use Case: Given an invention brief describing a new neural network compression method, the Skill extracts key technical features, searches patents and papers from the last 10 years, and writes a PRIOR_ART_REPORT.md with a landscape analysis and claim-scope recommendations. ## Quick Start Run a prior art search on my invention described in INVENTION_BRIEF.md and generate the patent/PRIOR_ART_REPORT.md report.
