This project provides an automated workflow for generating patent technical disclosure documents from project documentation. It targets engineers, researchers, and patent professionals, aiming to streamline the complex process of patent point identification, prior art searching, and document drafting, ultimately accelerating patent application preparation.

How It Works

The skill operates as an AgentSkill, processing input project documents (including .docx and .pptx converted to Markdown) to identify potential patentable points. It then performs prior art searches, prioritizing China's National Intellectual Property Administration (CNIPA) official publication site (epub.cnipa.gov.cn) via a dedicated Python script and Playwright, with a fallback to general web search. The core output is a technical disclosure document, incorporating Mermaid-rendered system and flow diagrams, and optionally exported to .docx format. The process includes self-checking for logical consistency and supports iterative refinement with version tracking.

Quick Start & Requirements

• Primary Install: Clone the repository into your AgentSkills directory (e.g., .claude/skills/patent-disclosure-skill).
• Dependencies: Python 3.9+ is required. Install core Python dependencies with pip install -r requirements.txt. For CNIPA prior art search, install additional packages via pip install -r tools/requirements-cnipa.txt and run python -m playwright install chromium. Node.js is required for diagram rendering.
• Links: Detailed installation instructions are available in INSTALL.md. The skill entry point and prompt mappings are defined in SKILL.md.

Highlighted Details

• Automated conversion of Office documents (.docx, .pptx) to Markdown for processing.
• Prioritized prior art search targeting epub.cnipa.gov.cn, with robust fallback mechanisms.
• Generation of disclosure documents featuring Mermaid-based system and flow diagrams, with .docx export capability.
• Built-in self-checking for logical consistency and parameter alignment within the disclosure.
• Iterative correction and merging functionality, maintaining dialogue history and versioned outputs.

Maintenance & Community

The README does not specify details regarding active maintenance, notable contributors, or dedicated community channels (e.g., Discord, Slack). Integration is framed within the AgentSkills ecosystem.

Licensing & Compatibility

The project is licensed under the MIT License. This license is permissive and generally allows for commercial use, modification, and distribution, including integration into closed-source projects.

Limitations & Caveats

The effectiveness of the CNIPA search is dependent on the successful installation of Playwright and associated dependencies; failure results in a less precise web search fallback. Diagram rendering requires a separate Node.js installation. The quality of the generated disclosure heavily relies on the clarity and completeness of the input project documentation. The skill is designed for integration into specific AI agent frameworks (e.g., Claude Code, Cursor).