r2

AI-powered research environment for academic papers. Multi-agent orchestration, RAG retrieval, and autonomous debugging — built on Claude Code.

Open Repository

Last Updated: 2026-03-22

Install

pip install "r2-research @ git+https://github.com/shusuke-ioku/r2.git"

Quick Start

r2 init my-paper/
cd my-paper/
cp .env.example .env
claude

What You Get

  • 8 agents: manuscript-writer, analyst, researcher, reviewer, theorist, source-acquirer, slides-writer, proofreader
  • 14 skills: writing, analysis, deep-research, reading, formal-modeling, review, debugging, verification, and more
  • RAG system: local PDF indexing, Self-RAG, DeepRAG, citation snowballing, 3-database search (Semantic Scholar, OpenAlex, Scopus)
  • Zotero integration: one-command download → Zotero → RAG index

r2-typst

LaTeX-like minimalist Typst template for academic papers and presentation slides. One style file powers both.

Open Repository

Last Updated: 2026-03-22

Quick Start

#import "./style.typ": paper, theorem, proof, prop, lem, asp

#show: doc => paper(
  title: [My Paper],
  authors: ((name: [Author], affiliation: [University]),),
  abstract: [Your abstract here.],
  doc,
)

Features

  • Paper layout with title block, abstract, numbered sections, bibliography
  • Polylux slides with matching style from the same file
  • Theorem environments (Proposition, Lemma, Assumption, Proof)
  • Publication-ready table helpers (caption-with-note, table-note)
  • Customizable design tokens for fonts, colors, spacing