An AgentSkills skill that automatically generates Chinese patent technical disclosure documents from project documentation, covering patent point mining, novelty search, desensitization drafting, and self-checking closure. Suitable for development teams with patent needs. https://github.com/handsomestWei/patent-disclosure-skill… Helps you sort out patent points from project documents and code, directly outputting a deliverable technical disclosure document.
Process: scan project documentation → mine patent points → connect to CNIPA for novelty search (prioritize China Patent Publication Announcement website) → desensitize and draft disclosure (including mermaid system diagrams and flowcharts) → automatically output .md and .docx. Supports iteration: existing disclosure documents can be supplemented or corrected, each revision saves as a new file with revision history traceable.

handsomestWei/patent-disclosure-skill

Source: https://github.com/handsomestWei/patent-disclosure-skill

中国专利.skill

From project documentation to deliverable technical disclosure documents: patent point mining, novelty search prioritizing CNIPA Publication Announcement website, desensitization drafting and self-checking closure.

License: MIT

Python 3.9+ (https://www.python.org/) Node.js (https://nodejs.org/) AgentSkills (https://agentskills.io)

Have design documents and code, but haven’t sorted out patent points?
Need system block diagrams and flowcharts in the disclosure, and a Word document that agents can directly edit?
Need multiple rounds of supplementing materials and corrections after finalization, and want file modification traceability?
CNIPA publication website search, expecting successful crawling and accurate search every time?

This Skill structures the entire workflow according to AgentSkills conventions; SKILL.md + prompts/ are readable and iterable step by step.

Features · Installation · Usage · Project Structure · Examples · Running Results · Reference Documentation · Detailed Installation Guide · Skill Entry

Features

Capability Description

• Project scanning (reads documentation/code by priority; .docx/.pptx first converted to Markdown before scanning, see prompts/project_scan.md)
• Patent point candidate discussion and merging (patent_points_analyzer.md)
• Novelty search prioritizes CNIPA · China Patent Publication Announcement (tools/cnipa_epub_search.py); falls back to WebSearch (Google Scholar / Patents) on exception or no results. Bibliographic data and external links are written into Chapter 1 (prior_art_search.md)
• Disclosure drafting: desensitization template + mermaid system block diagrams and flowcharts; mermaid_render.py → PNG, default output as .docx
• Naming convention for all deliverables: {CaseName}_{YYYYMMDDHHmmss}.md and same-name .docx (disclosure_builder.md §7.3)
• Self-checking: logical closure, formula and parameter consistency (disclosure_self_check.md, not written into the body)
• Iteration merging / correction: new file saved; disclosure revision dialog record.md appended incrementally (iteration_context.md, iteration_dialog_log.py)

Office extraction: .docx / .pptx first converted to Markdown using the repo’s docx_to_md.py / pptx_to_md.py before scanning (see SKILL.md).

Python dependencies (per file):

• Basic (Office / disclosure conversion): root requirements.txt — pip install -r requirements.txt
• Novelty search (CNIPA Publication Announcement, optional): tools/requirements-cnipa.txt — pip install -r tools/requirements-cnipa.txt, then run python -m playwright install chromium. Not installed? Step 5 will fall back to WebSearch only, per prior_art_search.md. See INSTALL.md, tools/README.md.

Installation

Claude Code

Place this directory in the git repository root or global skills path so that SKILL.md is at the skill folder root (consistent with INSTALL.md).

# Example: install to the skills directory of the current project
mkdir -p .claude/skills
git clone <this repo URL> .claude/skills/patent-disclosure-skill

Cursor

Place the entire content of this repository into Cursor’s designated skills path (see table in INSTALL.md), restart, and confirm the skill is detected under Settings → Rules.

Dependencies

# Basic (Office conversion, disclosure‑related Python packages)
pip install -r requirements.txt

# Optional: CNIPA novelty search (epub.cnipa.gov.cn)
pip install -r tools/requirements-cnipa.txt
python -m playwright install chromium

Diagram rendering also requires Node.js; run npm install in tools/ or use npx mmdc (see tools/README.md).

Usage

Describe your needs in natural language to the Agent, for example:

• “patent mining”, “patent points”, “technical disclosure document”, “novelty search”, “prior art comparison”
• Slash commands (depending on host configuration): e.g., /patent-disclosure-skill, /交底书

It is recommended to also specify the project path or technical subject (consistent with the argument-hint in SKILL.md).

Novelty search (Step 5) will first query through the China Patent Publication Announcement (http://epub.cnipa.gov.cn/) for Chinese patent publication information, then supplement with other sources as needed; see prompts/prior_art_search.md.

When supplementing materials or correcting errors on an existing disclosure document, you don’t need to say “iterate” — the skill will handle it via merger.md / correction_handler.md; details in SKILL.md.

Project Structure

This repository follows AgentSkills (https://agentskills.io); the root directory is a single skill:

patent-disclosure-skill/
├── SKILL.md                          # Entry: triggers, tool list, steps and prompts reference
├── prompts/                          # Step templates (read and followed by the Agent)
│   ├── intake.md
│   ├── project_scan.md
│   ├── patent_points_analyzer.md
│   ├── prior_art_search.md
│   ├── disclosure_preview.md
│   ├── disclosure_builder.md
│   ├── disclosure_self_check.md
│   ├── iteration_context.md
│   ├── merger.md
│   ├── correction_handler.md
│   └── template_reference.md
├── tools/                            # mermaid_render, md_to_docx, docx_to_md, pptx_to_md; cnipa_epub_*(novelty search); iteration_dialog_log, etc.
├── docs/                             # PRD, repository structure description, running result screenshots (example-*.jpg)
├── examples/                         # Sample raw materials (e.g., example_batch_job_scheduler/knowledge/)
├── outputs/                          # User output, entire directory .gitignore
├── requirements.txt
├── LICENSE
├── INSTALL.md
└── .gitignore

Examples

See examples/README.md for fictional scanning raw materials (e.g., examples/example_batch_job_scheduler/knowledge/).

Complete outputs (patent points, novelty search notes, disclosure document, etc.) are generated into the local outputs/{CaseIdentifier}/ directory.

Running Results

First version generation (first deliverable on disk) First version generation: timestamped disclosure documents in outputs directory, mermaid diagram directory, etc.

Iterative update (delivery after merge/correction, multiple versions coexisting + dialog record) Iterative update: new timestamped file and disclosure revision dialog record.

Reference Documentation

• Skill Entry and Agent Workflow (triggers, prompts/ mapping, tool table)
• Detailed Installation Guide (Claude Code / Cursor paths)
• Diagram and Conversion Scripts (mermaid / mmdc, Word export, CNIPA epub novelty search tools)
• Example Case and Raw Material Description
• Product Workflow and Directory Conventions
• Engineering Structure Description
• Disclosure Template Details

Support the Author

If this Skill saved you time writing disclosure documents, feel free to buy me a coffee ☕. Your support is greatly appreciated 🙏🙏

MIT License © handsomestWei (https://github.com/handsomestWei/)