Skills · Community project

patent-disclosure-skill

China Patent Skill organizes invention mining, disclosures, application drafts, patent reading, and search into separate workflows, with Word documents, diagrams, and optional Obsidian notes. Formal use requires professional review.

9.6k StarsMITUpdated today
01

Project overview

China Patent Skill (patent-disclosure-skill) is a toolkit with a root router for disclosures, application drafts, dockets, search, reading, maps, office-action assistance, and policy briefs. It helps turn existing technical material into reviewable drafts. It does not establish patentability or replace professional review or formal filing.

02

Core capabilities

01

Organize disclosure material

Extracts technical points from documents and code, routes invention, utility-model, and design disclosures separately, and supports diagrams and iterative versions.

02

Read patents and organize knowledge

Turns public patents into plain-language notes and graphs, optionally using Obsidian. Maps and other extensions require an explicit request.

03

Assist applications and responses

Existing disclosures can feed application drafts. Office-action assistance has a separate workflow and optional case library, not an automatic follow-up to reading.

Author example showing scheduler materials, patent search, and drafting a technical disclosure
Disclosure workflow example from the repository author, not an AgentMatter case test or a granted patent.View repository image
03

Installation and usage

Place the complete repository in the client's skill directory, retaining root SKILL.md and all subpackages. Run the dependency and browser-probe commands from the Skill repository root. Verify a small disclosure or public-patent reading task without enabling unrelated services.

AI AGENT INSTALL

Let an AI Agent install it

Send this prompt to Codex, Claude Code, or another AI agent that can work with your local environment.

Help me install patent-disclosure-skill from https://github.com/handsomestWei/patent-disclosure-skill. Read README.md and INSTALL.md first, confirm my agent and install destination, and use the documented commands. Place the complete repository in the client's skill directory, retaining root SKILL.md and all subpackages. Run the dependency and browser-probe commands from the Skill repository root. Verify a small disclosure or public-patent reading task without enabling unrelated services. Confirm discovery of patent-disclosure-skill and an ok=true browser probe. Run a clearly scoped task with non-sensitive material and inspect output paths, facts, diagrams, and editable files. A successful probe does not establish filing readiness. Ask before providing credentials, adding permissions or dependencies, or overwriting files. Keep secrets out of logs and do not automatically perform external writes.
01Before you start
  • Python 3.9+, pip, and a client that can read the Skill.
  • Local Chrome or Edge, or documented Playwright Chromium setup if missing.
  • Authorized technical material; start with a non-sensitive sample.
02Copy the install command or configuration
pip install -r requirements.txt
python skills/patent-disclosure/tools/browser.py --probe
03Complete the setup steps
  1. 1
    Place the complete Skill directory

    Download and extract the full repository. For project-scoped Claude Code use .claude/skills/patent-disclosure-skill; for Cursor use .cursor/skills/patent-disclosure-skill. Retain root SKILL.md and skills/patent-* and reload the client.

  2. 2
    Install dependencies and probe the browser

    From the Skill repository root, run pip install -r requirements.txt and python skills/patent-disclosure/tools/browser.py --probe. Install Chromium according to INSTALL.md only when no suitable browser is available.

  3. 3
    Choose one small task

    Provide authorized material and request only a disclosure or only a public-patent explanation. Inspect factual support and open questions. Request applications, dockets, or office-action assistance separately.

  4. 4
    Review the deliverables

    Open Markdown, Word, and diagrams to check content and formatting. Have a patent professional assess legal scope, claims, search completeness, and filing suitability; do not submit automatically.

How to verify the setup

Confirm discovery of patent-disclosure-skill and an ok=true browser probe. Run a clearly scoped task with non-sensitive material and inspect output paths, facts, diagrams, and editable files. A successful probe does not establish filing readiness.

Before using it
  • Run these commands from the Skill root containing requirements.txt.
  • CAD and office-action vector libraries are optional dependencies configured separately.
  • A Markdown fallback may be available without a browser, but it does not verify diagram or Word generation.
04

Use cases

SCENARIO 01

Prepare R&D disclosures

Organize existing documents and code into a disclosure draft with explicit unresolved technical facts.

SCENARIO 02

Read public patents

Use a public PDF or publication number to understand terminology, structure, and claims before human analysis.

05

Assessment

The documentation separates disclosure, application, and reading workflows and adds checks around materials and outputs. Its scripts and optional services require more setup than a simple prompt. AgentMatter has not installed the toolkit or tested case outputs and makes no legal-validity assessment.

Why it may be useful

  • Separates patent tasks to limit unintended scope expansion.
  • Supports editable documents, diagrams, and versioned materials.

What to know first

  • Outputs are assistance materials requiring patent-professional review.
  • Search, diagram generation, and Word output have not been tested by AgentMatter.
06

README

patent-disclosure-skill


Overview

China Patent Skill organizes invention mining, disclosures, application drafts, patent reading, and search into separate workflows, with Word documents, diagrams, and optional Obsidian notes. Formal use requires professional review. China Patent Skill organizes invention mining, disclosures, application drafts, patent reading, and search into separate workflows, with Word documents, diagrams, and optional Obsidian notes. Formal use requires professional review.

Getting started

  • Place the complete repository in the client's skill directory, retaining root SKILL.md and all subpackages. Run the dependency and browser-probe commands from the Skill repository root. Verify a small disclosure or public-patent reading task without enabling unrelated services.
  • The root SKILL.md routes to task-specific skills rather than executing every capability at once.
  • The default disclosure route targets inventions and generally requires Python, dependencies, and Chrome or Edge.
  • Confirm discovery of patent-disclosure-skill and an ok=true browser probe. Run a clearly scoped task with non-sensitive material and inspect output paths, facts, diagrams, and editable files. A successful probe does not establish filing readiness.

Configuration

pip install -r requirements.txt
python skills/patent-disclosure/tools/browser.py --probe
Read the complete README on GitHub