Repolode
Submit
claesbackman

claesbackman/AI-research-feedback

A set of Claude Code skills that review academic papers, pre-analysis plans, grant proposals and analysis code, and turn papers into briefs or slide decks.

View on GitHub
Stars
529
Forks
101
Open issues
0
Last commit
10 days ago

What is it?

AI-research-feedback is a collection of Claude Code skills for academic researchers. Each skill is a folder with a SKILL.md file that you install into ~/.claude/skills and invoke with a slash command such as /review-paper. The review skills launch several specialised agents in parallel and write a prioritised report into a reviews/ folder in your project. They cover full referee-style paper reviews, quick contribution and identification checks, mechanical consistency checks, pre-analysis plan reviews and grant proposal reviews.

The repository also includes code-focused skills. review-paper-code checks whether analysis scripts reproduce the claims in a LaTeX paper, audit-analysis adversarially audits changed Stata, R or Python code in an isolated subagent, and explain-diff writes an offline HTML explainer of a code change with a quiz. For communication, paper-version turns a LaTeX paper into a policy brief or summary, and three deck skills build explorable-style presentations in Quarto reveal.js, LaTeX Beamer and PowerPoint. A pdf-to-markdown utility converts PDFs for use with the other skills.

The skills target economics and finance venues and funders by name, and are written for projects that use LaTeX papers and Stata, R or Python code. Installation requires Claude Code. Some skills need extra tools such as Quarto, TeX Live or Node.js.

Key features

  • 01Multi-agent review skills for full papers, quick checks, pre-analysis plans and grant proposals
  • 02Paper-to-code reproducibility review for LaTeX papers with Stata, R or Python scripts
  • 03Adversarial audit of changed analysis code, with findings tied to file, line and quoted excerpt
  • 04Policy brief, one-page and five-page summaries generated from a LaTeX paper
  • 05Slide deck builders for Quarto reveal.js, LaTeX Beamer and PowerPoint
  • 06Journal and funder targets such as AER, QJE, NSF and ERC that tune review standards

Use cases

  • Getting a referee-style critique of a draft paper before submitting to a journal
  • Checking that analysis code actually produces the tables and coefficients reported in a paper
  • Reviewing a pre-analysis plan or grant proposal before filing
  • Turning a finished paper into a policy brief or a presentation for non-specialists

Quick start

  1. 1

    Create the skills folder and copy all skills into it using the one-line terminal install

    git clone --depth 1 https://github.com/claesbackman/AI-research-feedback.git /tmp/airf && mkdir -p ~/.claude/skills && cp -R /tmp/airf/Skills/. ~/.claude/skills/ && rm -rf /tmp/airf
  2. 2

    Restart Claude Code, then run a review skill such as /review-paper inside a project with a LaTeX paper

Things to know

  • Review skills need Claude Code with access to the general-purpose subagent.
  • The paper reviews expect a LaTeX paper, and the code-review skills need Stata, R, or Python analysis code.
  • Installation is done by copying files from git, with no package-manager install.

In collections

Claude Code tools, skills & plugins

Topics

#claude-code#academic-research#paper-review#llm-agents#economics#reproducibility#presentations#latex

FAQ

What is AI-research-feedback?
A set of Claude Code skills that review academic papers, pre-analysis plans, grant proposals and analysis code, and turn papers into briefs or slide decks. AI-research-feedback is a collection of Claude Code skills for academic researchers. Each skill is a folder with a SKILL.md file that you install into ~/.claude/skills and invoke with a slash command such as /review-paper. The review skills launch several specialised agents in parallel and write a prioritised report into a reviews/ folder in your project. They cover full referee-style paper reviews, quick contribution and identification checks, mechanical consistency checks, pre-analysis plan reviews and grant proposal reviews.
Is AI-research-feedback free to use? What license does it have?
AI-research-feedback is open source under the MIT license. Check the license terms before using it in commercial products.
Is AI-research-feedback still maintained?
Actively maintained. The last commit to AI-research-feedback was on Oct 28, 2026.
What can I use AI-research-feedback for?
Getting a referee-style critique of a draft paper before submitting to a journal Checking that analysis code actually produces the tables and coefficients reported in a paper Reviewing a pre-analysis plan or grant proposal before filing Turning a finished paper into a policy brief or a presentation for non-specialists
How do I install AI-research-feedback?
Create the skills folder and copy all skills into it using the one-line terminal install: git clone --depth 1 https://github.com/claesbackman/AI-research-feedback.git /tmp/airf && mkdir -p ~/.claude/skills && cp -R /tmp/airf/Skills/. ~/.claude/skills/ && rm -rf /tmp/airf Restart Claude Code, then run a review skill such as /review-paper inside a project with a LaTeX paper
Where can I get AI-research-feedback?
The source code and installation instructions are on GitHub: https://github.com/claesbackman/AI-research-feedback

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects