By jiutuhky
AI-powered Chinese patent application document writer using multi-agent orchestration
npx claudepluginhub jiutuhky/my-super-capsule --plugin patent-writerUse this agent when the patent abstract (说明书摘要) needs to be written. This agent writes concise, compliant abstracts under 300 words. <example> Context: Patent outline has been generated and abstract section needs to be written user: "撰写专利摘要" assistant: "I'll use the abstract-writer agent to write a concise patent abstract within 300 words." <commentary> Abstract writing is the fourth step, after outline generation, producing a compliant summary of the invention. </commentary> </example> <example> Context: The patent writing workflow has reached the content generation phase user: "写说明书摘要" assistant: "I'll use the abstract-writer agent to draft the abstract following Chinese patent standards." <commentary> The abstract must follow strict format requirements: under 300 words, no paragraphs, standard opening phrase. </commentary> </example>
Use this agent when patent claims (权利要求书) need to be written. This agent writes legally precise claims with maximum protection scope. <example> Context: Patent outline and abstract are ready, claims need to be drafted user: "撰写权利要求书" assistant: "I'll use the claims-writer agent to draft legally precise patent claims with multi-aspect protection." <commentary> Claims writing is the fifth step, requiring legal precision and broad protection scope strategy. </commentary> </example> <example> Context: The patent writing workflow needs claims covering method, apparatus, device, and medium user: "写独立权利要求和从属权利要求" assistant: "I'll use the claims-writer agent to write independent and dependent claims covering all protection aspects." <commentary> Claims must cover method, apparatus, electronic device, and storage medium for comprehensive protection. </commentary> </example>
Use this agent when the patent specification (说明书) including the detailed description section (具体实施方式 >10000 words) needs to be written. This agent produces comprehensive long-form technical writing. <example> Context: Patent outline, abstract, and claims are ready; the specification needs to be written user: "撰写说明书和具体实施方式" assistant: "I'll use the description-writer agent to write the full specification with detailed description exceeding 10000 words." <commentary> Description writing is the sixth step, producing the bulk of the patent document with multiple embodiments. </commentary> </example> <example> Context: The patent writing workflow needs the longest section of the document user: "写具体实施方式,要超过一万字" assistant: "I'll use the description-writer agent to generate comprehensive embodiments with sufficient technical detail." <commentary> The detailed description must exceed 10000 words and enable skilled persons to reproduce the invention. </commentary> </example>
Use this agent when all patent sections (abstract, claims, description) and diagrams have been generated, and need to be merged into a final Word (.docx) patent application file. This agent performs quality checks, analyzes content to determine diagram mapping, then calls scripts to produce the output. <example> Context: All patent text sections and diagrams have been generated user: "合并生成最终的 Word 专利申请文件" assistant: "I'll use the docx-merger agent to merge all content and diagrams into patent_application.docx." <commentary> The docx-merger agent performs quality checks, determines diagram mapping, and generates the final Word document. </commentary> </example> <example> Context: The patent writing workflow has completed text generation and diagram drawing user: "生成 Word 格式的专利申请文件" assistant: "I'll use the docx-merger agent to produce the final .docx patent application." <commentary> The agent reads all content files, verifies completeness, analyzes diagram references, and merges into a Word document. </commentary> </example>
Use this agent when the user needs to parse a technical disclosure document (技术交底书) for patent writing. This agent extracts structured information from input documents. <example> Context: User has provided a technical disclosure document and started the patent writing process user: "帮我解析这个技术交底书 data/技术交底书.docx" assistant: "I'll use the input-parser agent to extract structured information from the technical disclosure document." <commentary> The user wants to parse a technical disclosure document, which is the first step in the patent writing workflow. </commentary> </example> <example> Context: The patent writing command has started and needs to parse the input file user: "开始写专利,输入文件是 disclosure.docx" assistant: "First, I'll use the input-parser agent to extract key technical information from the disclosure document." <commentary> Parsing the input document is the mandatory first step before any patent content can be generated. </commentary> </example>
Use this agent when a patent application outline needs to be generated based on parsed technical information and prior art analysis. This agent designs the complete patent structure. <example> Context: Input has been parsed and similar patents have been searched user: "生成专利大纲" assistant: "I'll use the outline-generator agent to design a comprehensive patent application outline." <commentary> Outline generation is the third step, after input parsing and patent search, to plan the full document structure. </commentary> </example> <example> Context: The patent writing workflow needs a structured outline before content generation user: "设计专利文件的结构" assistant: "I'll use the outline-generator agent to create a structured outline with word count requirements for each section." <commentary> A well-designed outline ensures all sections meet Chinese patent law requirements. </commentary> </example>
Use this agent when similar patents need to be searched and analyzed for writing style reference during the patent writing process. This agent uses MCP tools to search patent databases. <example> Context: Input document has been parsed and keywords are available for patent search user: "搜索相关的专利文献" assistant: "I'll use the patent-searcher agent to search for similar patents and analyze their writing styles." <commentary> Patent search is the second step in the workflow, after input parsing, to find prior art and learn writing conventions. </commentary> </example> <example> Context: The patent writing workflow needs prior art research user: "查找类似专利,学习写作风格" assistant: "I'll use the patent-searcher agent to search Google Patents and technical databases for similar patents." <commentary> Searching for similar patents helps understand existing technology and learn professional patent writing style. </commentary> </example>
This skill should be used when the user asks to "生成专利附图PNG", "画专利插图", "generate patent diagram images", "专利附图绘制", "draw patent figures", "patent illustration", "专利示意图", "附图生成", or when the /patent-writer:write-patent command reaches its diagram generation step. Generates CNIPA-compliant black-and-white engineering-style patent diagrams (PNG), including method flowcharts, apparatus block diagrams, system architecture diagrams, and hardware cross-section diagrams.
This skill should be used when the user asks to "写专利", "撰写专利申请文件", "从技术交底书生成专利", "自动写专利", "generate patent from disclosure", "run patent writing workflow", "执行专利写作流程", or invokes the /patent-writer:write-patent command. Orchestrates 6 specialized sub-agents in strict sequence to produce patent application text content (Markdown) from a technical disclosure document. 附图 PNG 由 patent-diagram-drawing 技能在后续步骤中生成, 最终由 docx-merger agent 合并为 Word 专利申请文件。
个人 Claude Code 插件收录
| 插件 | 分类 | 说明 |
|---|---|---|
| paper-banana | 可视化 | 基于 Gemini 图像模型的学术论文插图生成器。通过 5 个子代理协作流水线,自动生成符合 NeurIPS/ICML 风格标准的科学示意图(架构图、流程图、方法图)和统计图表(柱状图、折线图、热力图等),支持多轮 Critic 迭代优化。 |
| patent-writer | 写作 | AI 驱动的中国专利申请文件自动撰写工具。从技术交底书(.docx)出发,经 8 个子代理按序协调,自动生成符合《专利法》和《专利审查指南》的完整申请文件(摘要、权利要求书、说明书、附图)。 |
在 Claude Code 会话中执行:
/plugin marketplace add jiutuhky/my-super-capsule
# 学术插图生成插件
/plugin install paper-banana@jiutuhky-plugins
# 专利撰写插件
/plugin install patent-writer@jiutuhky-plugins
/plugin
安装插件后,需根据所用插件配置对应的环境变量:
| 变量 | 所需插件 | 必填 | 说明 |
|---|---|---|---|
GOOGLE_API_KEY | 两者均需 | 是 | Google Gemini API,用于图像生成 |
GOOGLE_API_BASE_URL | 两者均需 | 否 | 自定义 Gemini API 端点 |
SERPAPI_API_KEY | patent-writer | 是 | Google Patents 专利检索 |
EXA_API_KEY | patent-writer | 是 | Exa 搜索引擎 |
pip install google-genai matplotlib Pillow markitdown
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
Requires secrets
Needs API keys or credentials to function
Uses power tools
Uses Bash, Write, or Edit tools
Share bugs, ideas, or general feedback.
Standalone image generation plugin using Nano Banana MCP server. Generates and edits images, icons, diagrams, patterns, and visual assets via Gemini image models. No Gemini CLI dependency required.
Ultra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Memory compression system for Claude Code - persist context across sessions
No description provided.
Streamline people operations — recruiting, onboarding, performance reviews, compensation analysis, and policy guidance. Maintain compliance and keep your team running smoothly.
No model invocation
Executes directly as bash, bypassing the AI model
No model invocation
Executes directly as bash, bypassing the AI model
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claim