zhuangwenhui

Analyze symptom patterns and potential disease associations. Use when user describes multiple symptoms and needs pattern analysis or differential diagnosis suggestions.

Assess symptom risk level (low/medium/high/emergency). Use when user describes symptoms and needs risk evaluation to determine urgency of medical attention.

Search clinical practice guidelines and expert consensus documents. Use when user needs authoritative treatment standards or evidence-based medical recommendations.

Conduct deep research over a local medical knowledge base (Milvus RAG) and PubMed literature with evidence synthesis. Use for complex medical questions requiring literature review.

查询疾病的 ICD-10 编码与分类。基于本地 ICD-10 中文码表（1586 条三位类目码）确定性查表，支持标准病名与常见口语（如冠心病、乙肝、慢阻肺）；命中多条并列返回，无匹配时明确告知未收录、绝不臆造编码。Use when the user asks for a disease's ICD-10 code or classification.

给出疾病的生活方式建议（饮食/运动/作息/戒烟限酒/监测）。基于本地确定性处方表（高血压、2型糖尿病、血脂异常/冠心病、慢阻肺、慢性肾病、感冒、通用健康），支持常见别名；未收录则明确告知，不返回无关泛化文案。Use when the user asks for lifestyle/diet/exercise advice for a condition.

Search current session conversation history (short-term memory). Use when user asks "我之前问过什么", "历史对话", "上次说了什么", or needs context from earlier in the same conversation.

Search medical knowledge base. Use when user asks about medical information, disease details, or general health knowledge. Fast semantic search powered by Milvus vector database.

Use when writing, editing, reviewing, or comparing C++/C code and task-local style decisions matter, especially for formatting, naming, headers and includes, comments, error handling, logging messages, or C++20-specific syntax choices. Primary target is the ShapeReconstruction codebase (mvrmesh).

Use when the user explicitly asks to run the incremental change cycle after adding code or changing function behavior, especially for long-running or high-token work where checklist continuity and repeated review checkpoints matter.

Use this skill when you writing commands, hooks, skills for Agent, or prompts for sub agents or any other LLM interaction, including optimizing prompts, improving LLM outputs, or designing production prompt templates.

Use when writing, editing, reviewing, or comparing Python code and task-local style decisions matter, especially for formatting, naming, docstrings, logging, resource handling, or Python 3.12-specific syntax choices.

维护一个"个人面试问答知识库"的可复用流程：从公开面经里构建业界问题分布，再据此核对个人项目逐字稿是否落在 "个人真实项目 ∩ 业界会问" 的交集里。当用户粘贴面经 / 面试经验 / Q&A 让你整理、要求更新问答字典或问题频率统计、 按章节拆/合子文件、做"标准答案¦自己项目回答"双轨、把项目逐字稿与业界问题分布做交集核对、或为某条目缺图/缺公式做复原时触发。 关键词：面经、面试问题、Q&A、问答知识库、问题频率、去重归纳、逐字稿、双轨、缺图复原、双向同步。

Synchronize recruiting job collections, JD dictionaries, technical-stack classification files, resume-version mappings, and application tracker workbooks. Use when Codex needs to ingest jobs from recruiting sites or company portals, deduplicate job postings, assign D-style inventory IDs, classify roles by JD technical requirements and available resumes, update Markdown JD/classification files, repair or validate Excel/XLSX application trackers, normalize workbook formatting, rebuild hyperlinks, or prepare reusable recruiting workflows for internship, campus, or full-time job applications.