patent-disclosure-skill: Automatically Generate Patent Disclosure Statements and Knowledge Graphs from Code and Design Documents

When developers input code and design documents, the system automatically completes draft patent disclosure statements for three types: invention, utility model, and design. It handles the entire process from project scanning and novelty search to document generation, eliminating the need to manually draw diagrams or craft wording as in traditional methods.

It extracts isometric projections, structural line drawings with part numbers, and design outlines from CAD models or product photos and inserts them into the disclosure statements. Generated documents support iterative updates with revision history, and final versions can be exported in Word format.

It interprets published patent PDFs and converts them into an Obsidian-based personal knowledge graph. It connects claims to specifications, visualizes term relationships, and searches public resources to supplement technical context. Accumulated data is used for similar patent comparisons and technology roadmap analysis.

It detects changes in the latest examination guidelines from the Korean Intellectual Property Office to update disclosure statement drafting strategies, and learns from past office actions and response data to draft responses to new office actions. It references similar cases via vector search, and all outputs are designed to undergo human review.