Review Paper Agent

A Streamlit-based review paper writing assistant that guides you through a structured workflow — from high-level overview → detailed structure → draft → final paper — with user review at each stage.

Features

• 5-Stage Workflow: Topic → Overview → Structure → Draft → Finalize
• 3 Writing Modes: Adjustable question depth and automation level based on user experience
• LLM Integration (Optional): Enable AI assistance with an OpenAI or Anthropic API key
• Manual Mode: Write everything yourself without an API key
• Interactive Chat: Ask questions during the writing process
• Export: Download as Markdown (.md) or Word (.docx)

Writing Modes

Mode	Description	Best For
Quick Start	Enter a topic and the AI auto-generates the rest	Rapid prototyping / first drafts
Standard	User makes key decisions; AI handles detailed writing	General review paper writing
Expert	In-depth workshop-style questions at every stage	Publication-quality academic papers

Project Structure

ResearchRA/
├── app.py                         # Streamlit main app
├── requirements.txt               # Python dependencies
├── .streamlit/config.toml         # Streamlit theme config
├── src/
│   ├── llm_client.py              # OpenAI / Anthropic API client
│   ├── paper_state.py             # Paper state & mode management
│   └── prompts.py                 # Mode-specific LLM prompt templates
└── components/
    ├── sidebar.py                 # Sidebar (mode selection, progress, LLM settings)
    ├── export.py                  # Markdown / Word export
    ├── stage_topic.py             # Stage 1: Topic setup
    ├── stage_overview.py          # Stage 2: High-level overview
    ├── stage_structure.py         # Stage 3: Detailed structure design
    ├── stage_draft.py             # Stage 4: Draft writing
    └── stage_finalize.py          # Stage 5: Finalization

Getting Started

pip install -r requirements.txt
streamlit run app.py

LLM API Setup

After launching the app, open LLM API Settings in the left sidebar:

1. Select a provider (OpenAI or Anthropic)
2. Choose a model
3. Enter your API key
4. Adjust temperature
5. Click Save Settings

The app is fully functional without an API key — all content can be written manually.