name: professional-patent-agents version: 1.0.2 description: |- 📜 專利專業代理 - Patent Professional Agents

一個專業的多代理專利撰寫與最佳化技能套件，覆蓋專利申請全流程。

🎯 核心功能：
• 場景一：使用者想法 → 技術挖掘 → 檢索分析 → 專利撰寫 → 質量稽核
• 場景二：使用者初稿 → 問題分析 → 最佳化建議 → 強化權利要求
• 場景三：代理機構反饋 → 風險評估 → 最佳化/取消建議

🤖 9個專業代理：
tech-miner (技術挖掘)、prior-art-researcher (現有技術檢索)、inventiveness-evaluator (創造性評估)、patent-drafter (專利撰寫)、claims-architect (權利要求架構)、patent-analyst (專利分析)、patent-auditor (專利稽核)、patent-value-appraiser (價值評估)、patent-converter (文件轉換)

✨ 特色：
• 中英文雙語支援
• 7章節標準專利模板
• 授權率預判
• 自動Word文件轉換
• 持續學習能力

🔍 搜尋關鍵詞：專利, patent, 專利撰寫, 專利最佳化, 現有技術檢索, 智慧財產權, IP, 權利要求, 技術交底書, 創造性評估, 授權率, patent drafting, prior art search, claims, patent prosecution

dependencies: skills: - tavily-search - aminer-open-academic python: - requests>=2.28.0 - python-docx>=0.8.11 system: - pandoc>=2.0 (Markdown轉Word) - node>=18 (mermaid-cli依賴) - mmdc/mermaid-cli (Mermaid圖表渲染) - chromium (Puppeteer瀏覽器，可選)

Professional Patent Agents Suite

License

MIT License

Copyright (c) 2026 BigPiPiHua

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Core Positioning

Boundary: Does not handle Office Action (OA) responses - leave that to professional patent agencies.

Dependencies

Required Skills

Python Dependencies

pip install requests python-docx

Agent List (9 Core Agents)

Workflows

Scenario 1: User Idea → Drafting + Search

flowchart TB
    subgraph S1[Phase 1: Tech Mining]
        M1[tech-miner<br/>Understand idea] ~~~ M2[Extract innovations] ~~~ M3[Disclosure framework]
    end

    subgraph S2[Phase 2: Search]
        R1[prior-art-researcher<br/>Keyword strategy] ~~~ R2[Multi-source search] ~~~ R3[Analyze results]
    end

    subgraph S3[Phase 3: Evaluation]
        E1[inventiveness-evaluator<br/>Inventiveness analysis] ~~~ E2[Risk scoring] ~~~ E3[Differentiation advice]
    end

    subgraph S4[Phase 4: Drafting]
        D1[patent-drafter<br/>Draft specification] ~~~ C1[claims-architect<br/>Design claims]
    end

    subgraph S5[Phase 5: Review]
        A1[patent-auditor<br/>Full review] ~~~ A2[Grant rate prediction]
    end

    S1 --> S2 --> S3 --> S4 --> S5

Trigger:

"Help me write a patent: [technical idea]"
"I have an idea and want to apply for a patent"

Output Files: - TECH_DISCLOSURE.md - Technical disclosure framework - KEYWORD_STRATEGY.md - Keyword strategy - PATENT_SEARCH_REPORT.md - Search report - INVENTIVENESS_REPORT.md - Inventiveness evaluation report - Patent-*.md - Complete patent document (7-section standard format) - PATENT_AUDIT_REPORT.md - Audit report (with grant rate prediction) - Patent-*.docx - Word document (auto-converted)

Scenario 2: User Draft → Optimization

flowchart TB
    subgraph P1[Phase 1: Analysis]
        A1[patent-analyst<br/>Parse draft] ~~~ A2[Identify issues] ~~~ A3[Extract keywords]
    end

    subgraph P2[Phase 2: Search]
        R1[prior-art-researcher<br/>Targeted search] ~~~ R2[Analyze results]
    end

    subgraph P3[Phase 3: Evaluation]
        E1[inventiveness-evaluator<br/>Inventiveness risk] ~~~ E2[Enhancement suggestions]
    end

    subgraph P4[Phase 4: Optimization]
        D1[patent-drafter<br/>Optimize specification] ~~~ C1[claims-architect<br/>Strengthen claims]
    end

    subgraph P5[Phase 5: Review]
        Q1[patent-auditor<br/>Full review] ~~~ Q2[Grant rate prediction]
    end

    P1 --> P2 --> P3 --> P4 --> P5

Trigger:

"Help me optimize this patent: /path/to/patent.md"
"Review this patent and provide optimization suggestions"

Output Files: - PATENT_ANALYSIS_REPORT.md - Analysis report - PATENT_SEARCH_REPORT.md - Search report - INVENTIVENESS_REPORT.md - Inventiveness evaluation report - PATENT_OPTIMIZATION_SUGGESTIONS.md - Optimization suggestions - Patent-*.md - Optimized patent document - PATENT_AUDIT_REPORT.md - Audit report

Scenario 3: Agency Feedback → Evaluation

Use Case: User has submitted to a patent agency, received a search report or prior art, needs to evaluate whether to continue optimizing or cancel the application.

flowchart TB
    subgraph F1[Phase 1: Read Feedback]
        R1[Read search report] ~~~ R2[Read prior art] ~~~ R3[Read original draft]
    end

    subgraph F2[Phase 2: Comparative Analysis]
        A1[inventiveness-evaluator<br/>Comparison] ~~~ A2[Identify conflicts] ~~~ A3[Assess differentiation space]
    end

    subgraph F3[Phase 3: Decision]
        D1{Inventiveness space?}
        D2[🟢 Sufficient<br/>Recommend optimization]
        D3[🟡 Limited<br/>User confirmation needed]
        D4[🔴 No space<br/>Recommend cancellation]
        D1 --> D2
        D1 --> D3
        D1 --> D4
    end

    subgraph F4[Phase 4: Optimization]
        O1[patent-drafter<br/>Targeted optimization] ~~~ O2[Strengthen differences] ~~~ O3[patent-auditor<br/>Final review]
    end

    F1 --> F2 --> F3
    D2 --> F4
    D3 -->|User confirms| F4
    D4 --> X[Output cancellation report]

Trigger:

"The agency gave me a search report, help me see if I can optimize: /path/to/report.pdf"
"Here's the prior art, help me evaluate if I need to modify the patent"
"The agency says there's risk, should I continue or cancel?"

Output Files: - AGENCY_FEEDBACK_ANALYSIS.md - Agency feedback analysis - DECISION_RECOMMENDATION.md - Decision recommendation (continue/cancel) - PATENT_OPTIMIZATION_SUGGESTIONS.md - Targeted optimization suggestions

Decision Criteria:

Agent Details

Tech Miner (Technology Mining Expert)

Identity: Dr. Li, 15 years of technology assessment and innovation mining experience

Trigger: When user provides vague idea

Output: Technical disclosure framework

Use tech-miner to analyze the technical idea:
[User description]

Prior Art Researcher (Prior Art Search Expert)

Identity: Dr. Chen, 15 years of patent search experience

Trigger: When search is needed

Output: Search report + analysis

Use prior-art-researcher to search:
Keywords: [keywords]
Technical field: [field]

Search Channels (Default): | Priority | Channel | Tool | Purpose | |----------|---------|------|---------| | 1 | Tavily | tavily-search | Quick search, technical overview | | 2 | AMiner | aminer-open-academic | Academic papers + patent database | | 3 | Google Patents | web_fetch | Global patent full text | | 4 | GitHub | tavily site: | Open source projects | | 5 | Tech blogs | tavily site: | Technical articles |

⚠️ Patent Database APIs Recommended:

Default channels may not be sufficient for accurate patent prior art search. For professional patent search, recommend users to connect patent database APIs:

ClawHub Skill Discovery:

# Always check ClawHub for patent search skills
clawhub search patent
clawhub search "prior art"

Inventiveness Evaluator (Inventiveness Evaluation Expert)

Identity: Dr. Zhao, former examiner, 12 years of evaluation experience

Trigger: After search completion

Output: Inventiveness evaluation report (with risk score)

Use inventiveness-evaluator to evaluate inventiveness:
Patent document: /path/to/patent.md
Search report: PATENT_SEARCH_REPORT.md

Patent Drafter (Patent Drafting Expert)

Identity: Patricia, 12 years of drafting experience, 92% grant rate

Trigger: After inventiveness evaluation passes

Output: Complete patent document (7 sections)

Use patent-drafter to draft patent:
Patent title: [title]
Technical disclosure: TECH_DISCLOSURE.md
Search report: PATENT_SEARCH_REPORT.md
Inventiveness evaluation: INVENTIVENESS_REPORT.md

Claims Architect (Claims Architect)

Identity: Claude, 1000+ claims experience

Trigger: Parallel participation during drafting phase

Output: Claims document

Use claims-architect to design claims:
Patent document: /path/to/patent.md
Core inventive points: [points]

Patent Analyst (Patent Analyst)

Identity: Dr. Zhang, 12 years of patent analysis experience

Trigger: First step in optimization scenario

Output: Analysis report + optimization suggestions

Use patent-analyst to analyze patent:
Patent document: /path/to/patent.md

Patent Auditor (Patent Audit Expert)

Identity: Judge Wu, former examiner, reviewed 3000+ applications

Trigger: After drafting/optimization completion

Output: Audit report (with grant rate prediction)

Use patent-auditor to audit patent:
Patent document: /path/to/patent.md

Patent Value Appraiser (Patent Value Appraiser)

Identity: Ms. Lin, 10 years of patent valuation experience, certified IP asset appraiser

Trigger: User needs to evaluate existing patent value (transfer, pledge, financing)

Output: Patent value assessment report (5-dimension radar chart + market value range)

Use patent-value-appraiser to evaluate patent value:
Patent title: [title]
Or patent number: [number]
Purpose: pledge financing / licensing negotiation / technology transfer / M&A transaction

5 Evaluation Dimensions: | Dimension | Weight | Content | |-----------|--------|---------| | Technical Value | 25% | Innovation degree, technical complexity, substitution difficulty | | Legal Value | 25% | Claim breadth, stability, invalidation resistance | | Market Value | 25% | Application scenarios, market size, competitive alternatives | | Economic Value | 15% | Cost savings, revenue potential, licensing income | | Strategic Value | 10% | Supply chain position, barrier strength, negotiation leverage |

Patent Converter (Document Conversion Expert)

Identity: Alex, document conversion expert, proficient in Markdown → Word conversion

Trigger: Auto-triggered after patent-auditor review passes

Output: Word document (.docx), with embedded Mermaid diagrams

Conversion Flow: 1. Parse 7 sections of patent Markdown 2. Extract Mermaid code blocks and render to PNG 3. Use Pandoc to convert section content 4. Fill into Word template at corresponding positions 5. Embed images into document 6. Output to same directory as source file

Dependencies: | Tool | Purpose | |------|---------| | Pandoc | Markdown → Word conversion | | mmdc (mermaid-cli) | Mermaid → PNG rendering | | python-docx | Word document operations |

Standard Patent Template (7 Sections)

# [Patent Title]

## 1. Related Prior Art and Their Defects or Deficiencies
### 1.1 Description of Prior Art
### 1.2 Defects or Deficiencies of Prior Art

## 2. Technical Improvements to Overcome the Above Defects

## 3. Alternative Solutions for Technical Improvements

## 4. Detailed Embodiments of the Technical Solution
### 4.1 System Architecture
### 4.2 Signal Logic Relationships
### 4.3 Implemented Functions
### 4.4 Specific Implementation Steps
### 4.5 Embodiment 1
(Describe included components/modules and their connections, signal logic; implemented functions and specific implementation steps)
## 5. Advantages of This Proposal Over Prior Art
(Achieved technical effects)
## 6. Related Drawings
(Structure diagrams with labeled component/module names; flowcharts with clear steps and process directions)
## 7. Claims
(Optional)

Language Adaptation

The agents automatically detect and use the user's language for output.

Chinese 7-Section Template (中文七章節模板)

# [專利名稱]

## 一、相關的現有技術及現有技術的缺陷或不足
### 1.1 現有技術描述
### 1.2 現有技術的缺陷或不足

## 二、為克服上述缺陷本提案的技術改進點

## 三、技術改進點的其他替代方案

## 四、詳細的技術方案具體實施例
### 4.1 系統架構
### 4.2 訊號邏輯關係
### 4.3 實現功能
### 4.4 具體實現步驟
### 4.5 實施例一

## 五、本提案相對現有技術的優點

## 六、相關附圖

## 七、權利要求書

Key Rules for All Languages

1. 7-Section Structure — Must follow the standard template regardless of language
2. No Executable Code — Use pseudocode or flowcharts instead
3. Quantified Technical Effects — Always quantify improvements (e.g., "30% efficiency increase")
4. Comparison Table — Include comparison with prior art
5. Mermaid Diagrams — Use flowcharts and architecture diagrams

Output Files Summary

這個技能的最新版本可以在7w4.net小蔥技能站找到。

Scenario 1: User Idea → Drafting

Scenario 2: User Draft → Optimization

Scenario 3: Agency Feedback → Evaluation

Patent Value Assessment

Installation Location

skills/
└── professional-patent-agents/
    ├── SKILL.md
    ├── agents/
    │   ├── tech-miner/
    │   ├── prior-art-researcher/
    │   ├── inventiveness-evaluator/
    │   ├── patent-drafter/
    │   ├── claims-architect/
    │   ├── patent-analyst/
    │   ├── patent-auditor/
    │   ├── patent-value-appraiser/
    │   └── patent-converter/
    └── skills/
        └── continuous-learning/

Auto-Conversion Flow

Trigger Conditions

Auto-invoke patent-converter to convert Markdown to Word when:

Conversion Flow

flowchart TB
    A[patent-auditor<br/>Review complete] --> B{Grant rate ≥ 60%?}
    B -->|Yes| C[Auto-invoke<br/>patent-converter]
    B -->|No| D[Prompt user confirmation]
    D -->|Confirm convert| C
    D -->|Don't convert| E[Output Markdown only]
    C --> F[Output .docx file]
    F --> G[Notify user<br/>Ready for agency submission]

Output Location

Source directory/
├── Patent-1-xxx.md      # Source Markdown
├── Patent-1-xxx.docx    # Converted output (same directory)
├── Patent-2-xxx.md
├── Patent-2-xxx.docx
└── ...

Innovation Mining & Notifications

Work Record Source

memory/
├── 2026-03-16.md    # Daily work records
├── 2026-03-17.md
├── 2026-03-18.md
└── ...

Innovation Mining Rules

1. Extract technical points from work records: Code commits, architecture designs, problem solutions
2. Combine with industry trends: Search for latest technical developments
3. Evaluate grant rate: Prior art search + Inventiveness evaluation
4. Selection criteria: Grant rate ≥ 65%, Low/Medium risk level

Notification Format

📢 Innovation Mining Report

📊 X potential innovations discovered this week:

1. 【Highly Recommended】A method for XXX
   - Grant rate prediction: 70-80%
   - Risk level: Low
   - Innovation: XXX

2. 【Recommended】A system for XXX
   - Grant rate prediction: 65-75%
   - Risk level: Medium
   - Innovation: XXX

📁 Detailed documents: /patent/weekly/

💡 Suggestion: Prioritize applying for patent #1

⚠️ Security & Installation Guide

System Dependencies

# 1. Install Pandoc (Markdown to Word conversion)
apt install pandoc

# 2. Install Node.js and mermaid-cli (Mermaid diagram rendering)
curl -fsSL https://deb.nodesource.com/setup_18.x | bash -
apt install nodejs
npm install -g @mermaid-js/mermaid-cli

# 3. Install Chromium (optional, for Playwright/Puppeteer)
playwright install chromium

Security Considerations

Recommended Usage

# Option 1: Non-root user (recommended)
python convert_patents.py /path/to/patents

# Option 2: Docker container (isolated environment)
docker run -v /path/to/patents:/data your-image python convert_patents.py /data

Credential Management

This skill depends on other skills that may require API keys or cookies:

Configure credentials securely before use.

Professional Patent Agents Suite v1.0.2 Author: BigPiPiHua Mail: 775262592@qq.com License: MIT Updated: 2026-03-31

質量較高，是一個非常專業的專利撰寫輔助工具。優點在於多代理協作流暢、流程覆蓋全面、雙語支援友好、持續學習能力強，能有效提升專利申請質量並預測授權可能性。不足是依賴較多外部工具可能導致安裝複雜，部分代理描述完整度待驗證。適合需要頻繁撰寫或最佳化專利的使用者使用。