git:20260307.7e1c6c8 to git:20260307.79c583f

---

name: research-repo

description: "Create and set up automated research repositories for curating academic papers in any domain. Use when user wants to start a new research tracking repository, migrate existing literature collection to AI-assisted workflow, or establish standardized workflows for paper discovery and documentation"

---

# Research Repository Setup

## Overview

This skill helps you create a fully-structured research repository for tracking academic papers in any domain. It provides templates and guidance for establishing an AI-assisted workflow for paper discovery, classification, and documentation.

## Quick Workflow

1. **Initialize repository structure** - Create directory layout and core files

2. **Configure domain specifics** - Customize priorities, tags, and templates

3. **Set up Claude Code Skills** - Create domain-specific search and add workflows

4. **Validate and commit** - Ensure all files are properly configured

## Step-by-Step Guide

### Step 1: Initialize Repository Structure

Create the following directory structure:

```bash

mkdir your-domain-repo && cd your-domain-repo

mkdir -p .claude/skills

mkdir -p paper_notes

touch README.md

touch README_zh.md # Optional: for Chinese support

touch TODO.md

touch CLAUDE.md

touch research-focus.md

touch classification-guide.md

touch .gitignore

touch LICENSE

```

### Step 2: Configure Domain Specifics

### Step 3: Create Domain-Specific Skills

Create two core skills in `.claude/skills/`:

#### A. search-[items] Skill (e.g., search-papers)

**Purpose**: Automated discovery workflow for your research items.

#### B. add-[item] Skill (e.g., add-paper)

**Purpose**: Standardized workflow for adding new items to repository.

### Step 4: Initialize README Files

```markdown

# [Domain Name] Research

Curated literature collection for [research area].

## 2026

### January

- **Paper Title**



> **Source:** Venue Year [[Link]](url)

```

### Step 5: Configure Git Workflow

**Files to commit**: Only README.md and README_zh.md

**Never commit**: CLAUDE.md, .gitignore, paper_notes/, reference files

**Commit message format**:

- Add: `Add paper: Title (Venue Year)`

- Update: `Update paper: Title - description`

- Fix: `Fix: issue description`

**First commit setup**:

```bash

git init

git add README.md README_zh.md TODO.md LICENSE

git commit -m "Initial commit: [Domain] research repository"

```

### Step 6: Validate Setup

- [ ] CLAUDE.md customized for domain

- [ ] research-focus.md has priorities and search queries

- [ ] classification-guide.md has complete tag system

- [ ] .gitignore excludes paper_notes/ and .claude/

- [ ] Git repository initialized

## Reference Materials

### Quick Reference Navigation

### Template Organization

Templates are split into focused modules for efficient loading:

- **skills/** - Skill-specific templates (search, add)

- **readme-entry.md** - Entry formatting guide

## Design Principles

### Why This Pattern Works

1. **Skill-based automation**: Complex workflows encoded as discoverable, executable commands

2. **Progressive disclosure**: Quick reference (CLAUDE.md) → deep dive (reference docs)

3. **Priority-driven focus**: Ensures attention on core research areas

4. **Local-global separation**: Personal insights (local) vs. polished repository (public)

5. **Git discipline**: Clean history, no accidental commits of notes or config

### Adaptation Examples

---

## Need Help?

See [faq.md](references/faq.md) for answers to common questions about:

- Language and localization

- Priority and classification

- Domain adaptation

- File management

- Tags and classification

- Git workflow

- Skills and workflow

- Quality and maintenance

- Collaboration and sharing

- Troubleshooting