handsomestWei/patent-disclosure-skill

What it solves

This project simplifies the complex process of patent application and analysis. It helps engineers and researchers who have the technical solution but struggle to write formal patent disclosure documents, understand dense legal language in existing patents, track changing patent examination policies, and draft responses to official examination notices.

How it works

The system operates as an AI agent skill with several specialized modes:

• Disclosure Writing: Scans project documentation and code (including .docx, .pptx, and CAD files) to identify patentable points. It generates structured disclosure documents for inventions, utility models, or designs, incorporating Mermaid diagrams or extracted line art/projections from CAD models, and can export to .docx.
• Plain-Language Interpretation: Extracts content from patent PDFs or public numbers to create simplified notes, terminology lists, and visual knowledge graphs (integrated with Obsidian) to make complex claims easier to understand.
• Policy Tracking: Uses web search to monitor official patent office (CNIPA) updates and examination trends, providing a reference list to ensure disclosure styles remain current.
• Examination Response: Implements a RAG-based workflow where historical responses and notices are stored in Obsidian. It uses tags or vector similarity search to retrieve relevant past cases to help draft responses to new examination notices.

Who it’s for

• Technical Inventors: Engineers and developers who need to convert their technical work into formal patent disclosures without manual drafting struggle.
• Patent Analysts: Professionals who need to quickly digest and map out the technical essence of existing patents.
• Legal Teams: Those managing patent responses who want to leverage historical case data for consistent drafting.

Highlights

• Multi-type Support: Tailored workflows for Invention, Utility Model, and Design patents.
• CAD Integration: Ability to extract isometric projections and line art from engineering models.
• Obsidian Ecosystem: Deep integration with Obsidian for knowledge graph visualization and private patent databases.
• Automated Prior Art Search: Prioritizes searches via the CNIPA public announcement system.
• RAG-powered Responses: Uses a local knowledge base of past cases to improve the accuracy of official response drafts.