ai4ss-skills

Source-grounded literature discovery and evidence-matrix skill for social-science research. Use when searching, screening, extracting, or updating literature tables from Google Scholar, Crossref, Semantic Scholar, journal pages, working-paper sites, Zotero libraries, PDFs, or user supplied bibliographies. Triggers: "literature matrix", "文献矩阵", "literature review search", "Zotero", "DOI", "screen papers", "extract identification strategy", "do not write prose yet", "source-grounded literature", "找文献", "文献综述前的检索", "文献检索", "引用核查", "建立文献矩阵".

Empirical methods review skill for auditing social-science research designs, scripts, model outputs, tables, figures, robustness checks, and reproducibility evidence. Use before submission, presentation, replication release, or reviewer response when the task is to find design, data, code, inference, reporting, or overclaiming issues. It reviews and structures problems; it does not decide substantive identification validity for the author. Triggers: "methods review", "audit empirical design", "review regression output", "check fixed effects", "cluster standard errors", "robustness audit", "reproducibility review", "submission checklist", "计量审查", "方法审查", "结果审查", "稳健性检查".

R 代码性能诊断与优化指南。帮助识别性能瓶颈、理解 R 底层机制、选择优化策略。 **触发场景：** - "这段代码跑得太慢了" - "帮我优化这个函数/循环的性能" - "如何加速这个包的某个函数" - "为什么这段 R 代码这么慢" - "如何在 HPC 上并行运行" - 用户询问 profvis、bench、future、并行计算相关问题

Social-science analysis execution skill for turning a study design brief and analysis-ready data into first-pass tables, figures, model outputs, coding outputs, logs, and an analysis run manifest. Use after study-design-builder and research-data-builder when the user asks to run the first analysis, produce baseline/descriptive outputs, execute an analysis plan, generate table or figure shells from data, or make results ready for methods review. Triggers: "run analysis", "first table", "baseline model", "descriptive table", "analysis runner", "execute analysis plan", "生成第一张表", "跑基准模型", "描述性统计", "结果清单".

Social-science data engineering skill for turning raw files into auditable analysis samples. Use when building or repairing data pipelines, panel datasets, merges, variable construction, sample-flow reports, missingness audits, text-to-structure extraction, or reproducible scripts for empirical research. Triggers: "build analysis sample", "data cleaning", "merge audit", "panel construction", "sample flow", "variable dictionary", "raw to analysis", "text extraction", "do not overwrite raw data", "reproducible data pipeline", "数据清洗", "样本构造", "合并检查", "变量构造", "样本流".