What is this?
Auto-Empirical Research Skills (AERS) is a curated library of 1,096 AI agent skills across 76 collections for doing empirical social-science research (economics, political science, sociology, psychology, etc.), maintained by Stanford REAP and CoPaper.AI.
Auto-Empirical Research Skills (AERS) is a curated library of 1,096 AI agent skills across 76 collections for doing empirical social-science research (economics, political science, sociology, psychology, etc.), maintained by Stanford REAP and CoPaper.AI.Is this a coding framework or a set of ready-made skill files?It is a skills library: each collection is a folder containing a SKILL.md that you drop straight into an AI coding agent's skills directory, not code you write against.
It is a skills library: each collection is a folder containing a SKILL.md that you drop straight into an AI coding agent's skills directory, not code you write against.How do I install these skills into my agent?Easiest way: paste the repo URL to Claude Code or Codex and tell it whether to install at the directory, project, or global scope and it handles the rest; alternatively use the Claude Code v2.1+ plugin marketplace or manually copy a skill folder (with its SKILL.md) into .claude/skills/.
Easiest way: paste the repo URL to Claude Code or Codex and tell it whether to install at the directory, project, or global scope and it handles the rest; alternatively use the Claude Code v2.1+ plugin marketplace or manually copy a skill folder (with its SKILL.md) into .claude/skills/.Which coding agents can use these skills?Claude Code (plugin marketplace needs v2.1+) and Codex are directly supported, with extra install paths documented for whole-repo import into CodeBuddy.
Claude Code (plugin marketplace needs v2.1+) and Codex are directly supported, with extra install paths documented for whole-repo import into CodeBuddy.Are all 1,096 skills equally trustworthy?No — only the 7 flagship skills are built and benchmark-locked by the Stanford REAP/CoPaper.AI team (17 numeric benchmarks plus behavioral evals); the rest are community skills vendored from upstream repos with license/provenance audits but not the same rigor guarantee.
No — only the 7 flagship skills are built and benchmark-locked by the Stanford REAP/CoPaper.AI team (17 numeric benchmarks plus behavioral evals); the rest are community skills vendored from upstream repos with license/provenance audits but not the same rigor guarantee.Who is this aimed at?Researchers and students doing quantitative empirical work in economics, political science, sociology, psychology, public administration, and international relations who use an AI coding agent as a research assistant.
Researchers and students doing quantitative empirical work in economics, political science, sociology, psychology, public administration, and international relations who use an AI coding agent as a research assistant.What is StatsPAI?StatsPAI is the project's causal-inference engine: an agent-native Python DSL with 900+ functions where a call like sp.causal(...) runs a full DID/RD/IV/SCM/DML workflow in one go.
StatsPAI is the project's causal-inference engine: an agent-native Python DSL with 900+ functions where a call like sp.causal(...) runs a full DID/RD/IV/SCM/DML workflow in one go.What does the 9-stage research pipeline cover?It chains idea refinement, literature review, data acquisition, identification strategy, estimation, robustness auditing, publication-grade tables/figures, writing/peer review, and de-AIGC/submission, orchestrated end-to-end by the Paper-WorkFlow skill.
It chains idea refinement, literature review, data acquisition, identification strategy, estimation, robustness auditing, publication-grade tables/figures, writing/peer review, and de-AIGC/submission, orchestrated end-to-end by the Paper-WorkFlow skill.Can I benchmark my own agent against the project's scoring suite?Yes - install with pip install -e . then run the aers-score command to get a score, which can be published on the public external scoreboard.
Yes - install with pip install -e . then run the aers-score command to get a score, which can be published on the public external scoreboard.Does it help pass Chinese plagiarism/AI-detection checks?Yes, the de-AIGC-skills collection offers bilingual Chinese/English AI-trace removal targeting Turnitin AI, GPTZero, and Chinese platforms like Zhiwang and Wanfang, including hidden-character and docx metadata cleanup.
Yes, the de-AIGC-skills collection offers bilingual Chinese/English AI-trace removal targeting Turnitin AI, GPTZero, and Chinese platforms like Zhiwang and Wanfang, including hidden-character and docx metadata cleanup.What are similar projects?open-science, codex-claude-academic-skills, modex-mh-agent, awesome-ai-research-writing, PaperSpine, awesome-persona-skills, goutoujunshi, Mind-Expanding-Books
open-science, codex-claude-academic-skills, modex-mh-agent, awesome-ai-research-writing, PaperSpine, awesome-persona-skills, goutoujunshi, Mind-Expanding-BooksWhich topics is it in?economics, reproducible-research, r, communication, latex, academic-research, data-science, skills, plugin, awesome-list, list, agent-skills
economics, reproducible-research, r, communication, latex, academic-research, data-science, skills, plugin, awesome-list, list, agent-skills