Auto-Empirical Research Skills

Install

Manual install

claude plugin marketplace add brycewang-stanford/Auto-Empirical-Research-Skills
claude plugin install aer-skills@auto-empirical-research-skills
claude plugin install empirical-analysis-python@auto-empirical-research-skills
claude plugin install empirical-analysis-stata@auto-empirical-research-skills
claude plugin install empirical-analysis-r@auto-empirical-research-skills

This is third-party code. Review the repository files before installing.

What it does

The project installs into Claude Code and Codex as a plugin marketplace for empirical research. The authors' own core plugins are aer-skills, a nine-skill stack for preparing an economics manuscript for top journals, from topic selection and identification strategy to robustness, tables and revision letters; plus three empirical-analysis pipelines in Python, Stata and R with steps for data cleaning, variable construction, model estimation, robustness and table and figure output. Identification methods cover difference-in-differences, instrumental variables, regression discontinuity, synthetic control and double machine learning. Around these first-party pipelines sits a curated catalog of third-party skills you can attach by copying into the skills directory or via --plugin-dir. The project ships benchmarks and eval scenarios so the agent's work can be re-checked.

Who it is for. For economists and social science researchers who run empirical work and prepare papers for publication.

Example request

Run a Callaway and Sant'Anna event study on this panel data and build the paper tables

Limitations

The analysis pipelines require installed environments: Python with econometrics packages, Stata or R with Quarto. The project targets economics and social science first, so methods need adaptation for other fields. Part of the catalog is third-party skills attached as submodules, with their own quality and licenses. Marketplace plugin installation needs a recent Claude Code version. The agent helps with method and formatting, but responsibility for the correctness of conclusions and data rests with the researcher.

How to disable. Remove the installed plugins via /plugin and drop the auto-empirical-research-skills marketplace. Delete any manually copied folders from ~/.claude/skills/ or the project's .claude/skills/.

Security check

• Runs statistical code in Python, Stata and R and reads project data
• Part of the catalog is attached as third-party submodules with their own licenses

README in short

The README presents this Stanford REAP and CoPaper.AI project as an end-to-end empirical research process: from topic framing and literature review to data collection, identification strategy, estimation, robustness, table and figure output, writing and reviewer responses. Documentation is multilingual, Chinese by default with an English version. Installation goes through the Claude Code plugin marketplace, and individual skills can be copied into the skills directory or attached via --plugin-dir. The project keeps benchmarks and eval scenarios so results can be re-checked with the same scorer. First-party plugins and parts of the catalog are under MIT and CC-BY-SA.

Related