Getting Started

Set up ScholaRAG in 3 minutes with a single copy-paste prompt. No manual commands required—Claude Code handles everything automatically.

✨ Conversation-First Setup

ScholaRAG is designed for conversation-driven automation. You paste prompts, Claude Code executes commands, and you answer questions about your research. That's it.

Prerequisites

You need two things installed before starting:

1. VS Code

Download and install Visual Studio Code

Free, works on macOS/Windows/Linux

2. Claude Code Extension

Install Claude Code in VS Code

Requires Claude Pro subscription ($20/month)

How to Install Claude Code Extension

1. Open VS Code
2. Open Extensions: Press Cmd/Ctrl + Shift + X
3. Search: Type "Claude Code" by Anthropic
4. Install: Click the Install button
5. Sign In: Click Claude icon → Sign in with Claude Pro account

Extension link: marketplace.visualstudio.com/claude-code

One-Click Automated Setup

Once Claude Code is installed, copy this prompt and paste it into the Claude Code chat panel. That's all you need to do.

📋 Copy this entire prompt → Paste to Claude Code

Please set up ScholaRAG for me:

1. Clone https://github.com/HosungYou/ScholaRAG.git
2. Navigate into ScholaRAG directory
3. Create Python virtual environment (venv)
4. Activate the virtual environment
5. Install CLI dependencies: pip install click pyyaml
6. Initialize my first project: python scholarag_cli.py init
7. Guide me through Stage 1 of the systematic review workflow

My research topic: [Describe your research question here]

✨ What Claude Code Will Do Automatically:

✅ Clone the ScholaRAG repository from GitHub
✅ Create and activate Python virtual environment
✅ Install all required dependencies
✅ Run CLI initialization to create your project folder
✅ Guide you through research domain setup (Stage 1)

Estimated time: 2-3 minutes for setup. Claude will prompt you for project name and research question during initialization.

How ScholaRAG Works

Understanding the repository structure helps you see how the conversation-driven workflow operates:

ScholaRAG/
├── prompts/                ← 7 stage-specific prompts (you copy-paste these)
│   ├── 01_research_domain_setup.md
│   ├── 02_query_strategy.md
│   ├── 03_prisma_configuration.md
│   ├── 04_rag_design.md
│   ├── 05_execution_plan.md
│   ├── 06_research_conversation.md
│   └── 07_documentation_writing.md
│
├── scripts/                ← Python scripts (Claude Code runs these)
│   ├── 01_fetch_papers.py
│   ├── 02_deduplicate.py
│   ├── 03_screen_papers.py
│   ├── 04_download_pdfs.py
│   ├── 05_build_rag.py
│   ├── 06_query_rag.py
│   └── 07_generate_prisma.py
│
├── templates/              ← config.yaml templates
├── scholarag_cli.py        ← CLI tool for project initialization
└── requirements.txt        ← Python dependencies

The 7-Stage Workflow

Each stage has a dedicated prompt in the prompts/ folder. You copy-paste each prompt to Claude Code, have a conversation, and Claude automatically runs the corresponding script when ready.

Stage 1: Research Domain Setup (15 min)

Define scope, research questions, constraints

Stage 2: Query Strategy (10 min)

Design Boolean queries, select databases

Stage 3: PRISMA Configuration (20 min)

Configure AI-PRISMA rubric with evaluation criteria

Stage 4: RAG Design (15 min)

Plan vector database and chunking strategy

Stage 5: Execution Plan (1-3 hrs, automated)

Fetch papers → Deduplicate → Screen → Download PDFs → Build RAG

Stage 6: Research Conversation (ongoing)

Query your RAG system for insights and analysis

Stage 7: Documentation & Writing (ongoing)

Generate PRISMA diagrams and prepare publication materials

💡 Key Insight

You never manually run python scripts/01_fetch_papers.py. Instead, Claude Code reads the prompts, understands which stage you're in, and executes the appropriate scripts automatically when needed.

What Gets Created

After running python scholarag_cli.py init, you'll have a project folder:

projects/2025-10-24_AI-Healthcare-Adoption/
├── config.yaml             ← Project settings
├── data/
│   ├── open_access/        ← Database search results
│   ├── combined/           ← Deduplicated papers
│   ├── prisma/             ← AI-PRISMA evaluation
│   └── pdfs/               ← Downloaded PDFs
├── rag/
│   └── chroma_db/          ← Vector database
└── outputs/                ← PRISMA diagrams, reports

Next Steps

Once setup is complete, proceed to Chapter 3: Core Concepts to understand PRISMA, RAG architecture, and how the stages connect.

📖 Core Concepts

Learn about PRISMA, RAG architecture, and vector databases

💡 Prompt Library

Browse 7 specialized prompts for research conversations

Advanced: Manual CLI Mode

For users who prefer direct command-line control

If you want to run commands yourself without Claude Code automation:

# Clone the repository
git clone https://github.com/HosungYou/ScholaRAG.git
cd ScholaRAG

# Create virtual environment
python3 -m venv venv

# Activate virtual environment
# On macOS/Linux:
source venv/bin/activate
# On Windows:
venv\Scripts\activate

# Install CLI dependencies
pip install --upgrade pip
pip install click pyyaml

# Initialize your first project
python scholarag_cli.py init

Note: Manual mode requires you to run Stage scripts yourself. Conversation mode is recommended for seamless automation.

Troubleshooting

Claude Code not responding

Solution:

Press Cmd/Ctrl + Shift + P → "Reload Window"
Re-authenticate: Claude icon → Sign out → Sign in
Update extension to latest version

Python not found

Error: command not found: python3

Solution: Install Python from python.org/downloads

Claude Code will usually prompt you to install Python if it's missing

Introduction

Core Concepts