Instructions file
Project instructions for stata-ai, a Chinese-language course and example repository about Stata and AI-assisted empirical research. The project is published as a Quarto book, which turns source files into a structured website or book.
Skill Claude CodeCodex
Use this skill to build a quick, structured Chinese overview of an empirical (economics/social-science) paper — its research background, question, approach, main findings, and literature position. For fast top-journal reading, relevance screening, and mapping a paper's place in the literature. Keeps figure/table…
Skill Claude CodeCodex
Use this skill to break down an empirical paper's identification strategy — its research question, setting, data, econometric method, and identification/endogeneity handling (a Problem-Setting-Data-Method-Identification decomposition). Explains what each method is for and why the author uses it, and maps key…
Skill Claude CodeCodex
Use this skill to navigate a paper's replication package. Given the README and directory listing, it produces a run roadmap (entry script to final outputs), explains what each folder holds and how the master script calls sub-scripts, describes the raw/intermediate/analysis/output data layers, and judges how far the…
Skill Claude CodeCodex
Use this skill when planning or reviewing data cleaning for empirical research. Given a variable list and data dictionary, it drafts a cleaning plan; checks whether a merge/join is sound (1:1 vs m:1, key name/case/type/leading-zero consistency, expected match rate); explains the differences among missing-value…
Skill Claude CodeCodex
Use this skill when reading, interpreting, or writing up regression results. Given a regression table and variable definitions, it explains each coefficient in plain language with correct direction, magnitude and units; reads log-level, log-log and standardized coefficients correctly (semi-elasticity / elasticity /…
Skill Claude CodeCodex
Use this skill to turn a data-analysis run into reproducible records. Given the scripts you ran, the data you used and the outputs you got, it drafts a reproduction log (input data → processing script → output), a variable-construction note, a sample-selection log (row counts at each filter), a results explanation and…
Skill Claude CodeCodex
Use this skill when an AI agent needs to actually run Stata code locally to verify results, check whether a do-file runs, locate errors from Stata output, or execute repetitive data-processing steps. This is a reference-type entry pointing to the Stata AI Skill bundled with the "Stata All in One" VS Code extension; it…
Skill Claude CodeCodex
Use this skill to audit and cross-validate a data-cleaning result before trusting it, especially work produced by an AI/agent. Given the cleaning code plus before/after data, it checks the empirical red lines: is every change in sample size accounted for (a sample-selection log); is the merge match rate normal (no…