nature-paper-to-patent

Convert research papers and project materials into evidence-grounded Chinese patent drafts.

Updated Aug 26, 2026
One-click install
npx skills add https://github.com/Yjj0333/skills-manager-backup --skill nature-paper-to-patent-yjj0333
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: nature-paper-to-patent
Source: https://github.com/Yjj0333/skills-manager-backup/tree/main/nature-paper-to-patent
Command: npx skills add https://github.com/Yjj0333/skills-manager-backup --skill nature-paper-to-patent-yjj0333

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires latex2mathml, mammoth, matplotlib, Pillow, pypdf, python-docx, python-pptx, and includes scripts (resource) and references (resource) components.

What problem does it solve? Turning a scientific paper or technical project into a Chinese invention patent application requires mapping every claim feature back to source evidence, preserving formulas, and producing formal Chinese documents—a slow, error-prone manual process. ## Core Features & Use Cases - Evidence-Constrained Drafting: Builds stable source IDs and an evidence ledger so every claim feature traces back to paper pages, equations, figures, or code; unsupported features are excluded rather than invented. - Full Patent Package Generation: Produces claims, specification, abstract, abstract figure, and claim-aligned flowcharts as separate Chinese DOCX files, with core formulas rendered as editable Office Math. - Technical Disclosure & Prior-Art Workflow: Mines patent points from project documents, runs CNIPA-oriented prior-art comparison, and generates timestamped Chinese technical disclosure (技术交底书) Markdown and Word files with Mermaid diagrams. - Use Case: Given a PDF paper on a defect-detection model, generate a complete Chinese patent draft where each claim limitation maps to a paper page, formulas are editable in Word, and validation scripts flag any structural or traceability errors. ## Quick Start Use the nature-paper-to-patent skill to convert this paper PDF into a Chinese invention patent disclosure and claim draft with evidence mapping.

Frequently Asked Questions about nature-paper-to-patent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I convert a research paper into a Chinese patent draft?▼

Provide the paper PDF plus any code, figures, or inventor notes, and specify the protection target such as method, system, or device. The skill builds a source map and evidence ledger first, then drafts claims, specification, abstract, and figures as separate Chinese DOCX files.

How does the skill ensure patent claims are supported by the paper?▼

Every material claim feature must map to stable source IDs (paper text, equations, figures, or code) with a support status of explicit, inherent, needs-confirmation, or unsupported. Unsupported features are excluded from formal claims, and validation scripts report unmapped features as errors.

Can it handle scanned PDFs or unclear formulas?▼

Yes, a scanned-pdf source mode detects OCR-limited input and flags uncertain blocks. Unclear formula symbols are marked as inventor questions rather than guessed, and uncertain features stay out of formal claims.

Does it generate Word documents with editable equations?▼

Yes, core formulas are stored with LaTeX source and converted to native editable Office Math (OMML) in the DOCX output via latex2mathml. Plain LaTeX strings are not used as the visible formula in delivered documents.

What are the limitations of this patent drafting skill?▼

It is a drafting aid, not a patentability opinion, legal advice, or filing guarantee, and it does not replace a patent attorney. It never invents embodiments, experimental data, or prior-art conclusions, and it does not determine novelty, inventorship, or ownership.

Can it compare a paper against an existing patent?▼

Yes, the paper-patent audit mode builds separate source maps for both documents and produces a bidirectional feature table labeling paper-only, patent-only, broader, or narrower content. It never assumes patent-only text is supported by the paper.