Gans Research Advisor

A Codex skill for developing economics research ideas in a Joshua Gans-style workflow. It helps turn rough ideas, drafts, model sketches, and paper sections into a clearer research logic: phenomenon, conventional view, missing friction, minimal model, benchmark, proposition, intuition, implications, and extensions.

This is an independent research-assistant skill. It is not affiliated with, endorsed by, or maintained by Joshua Gans.

What It Does

• Assesses whether a research idea is paper-worthy and how to narrow it.
• Reconstructs rough ideas into economic research questions.
• Builds compact model setups with agents, timing, information, actions, payoffs, and equilibrium objects.
• Reviews drafts for missing mechanism, weak propositions, vague frictions, or unclear contribution.
• Rewrites abstracts, introductions, and proposition exposition around mechanism rather than generic polish.
• Searches a local case corpus to anchor advice in concrete Gans-style examples.

Directory Structure

.
├── SKILL.md
├── agents/
│   └── openai.yaml
├── references/
│   ├── core_25_casebook.md
│   ├── corpus_95_paper_examples.md
│   ├── deep_cases/
│   └── paper_notes_95/
└── scripts/
    └── select_cases.py

Installation

Clone or copy this repository into your Codex skills directory:

git clone https://github.com/YOUR_USER/gans-research-advisor.git ~/.codex/skills/gans-research-advisor

Restart Codex if the skill does not appear immediately.

Usage

Ask Codex to use the skill on a research idea, draft, or model:

Use $gans-research-advisor to evaluate this idea.

帮我用 Gans 风格把这个 idea 变成一个经济学模型。

Use $gans-research-advisor to rewrite this introduction.

The skill is case-driven. For substantive advice, it searches the local corpus and borrows specific modeling or writing moves from the closest cases.

Case Search

The case selector uses only the Python standard library:

python scripts/select_cases.py --top 8 "AI adoption in law firms with human review"

Use --core-only to restrict results to the selected core cases:

python scripts/select_cases.py --core-only "startup licensing patent uncertainty"

Publication Notes

The references/ directory contains original reading notes, structured summaries, and research-transfer templates based on papers and public materials. The underlying papers, titles, abstracts, and source materials remain the property of their respective authors and publishers.

Before publishing this repository publicly, review the reference notes for:

• long verbatim excerpts from copyrighted works;
• publisher-provided PDFs or paywalled material;
• private annotations or unpublished research notes;
• personal paths, credentials, or local configuration.

For personal syncing, a private repository is the safest default.

License

See LICENSE.md.