title	PaperFlux
emoji	📚
colorFrom	blue
colorTo	indigo
sdk	streamlit
sdk_version	1.30.0
app_file	app.py
pinned	true

PaperFlux: AI Research Paper Insights

PaperFlux is a Streamlit app that fetches Hugging Face Daily Papers, then runs two Gemini agents: one that explains the PDF in depth, and one that produces 2–3 insights with diagrams.

Features

• Daily Updates: Automatically fetches and processes new papers every weekday at 8:00 AM UTC
• Analyst agent: Gemini Flash reads the native PDF and writes a technical breakdown
• Insights agent: A second Gemini key (separate quota) turns the paper + writeup into 2–3 visual diagrams (SVG or Mermaid)
• Paper Library: Browse processed papers in Streamlit
• Original PDFs: Direct arXiv download links

System Architecture

PaperFlux follows a robust architecture for fetching, processing, and displaying research papers:

   flowchart TD
      A[Scheduler] -->|Daily trigger| B[Paper Processor]
      B -->|Fetch papers| C[Hugging Face API]
      B -->|Download PDFs| D[arXiv]
      B -->|Analyze PDF| E[Analyst agent — Gemini Flash]
      E -->|Explanation + PDF| I[Insights agent — Gemini Flash]
      I -->|Store data| F[(MongoDB)]
      G[Streamlit UI] -->|Display papers| F
      H[User] -->|View papers| G

Loading

System Flow

1. Scheduled Polling: Every weekday at 8:00 AM UTC, the scheduler checks if papers need to be processed
2. Data Collection: The application fetches the latest papers from Hugging Face's API
3. PDF Processing: Papers are downloaded from arXiv and stored temporarily
4. Analyst agent: Each PDF is explained in depth using Gemini (native PDF input)
5. Insights agent: A second agent (separate API key) produces 2–3 insights and diagrams
6. Data Storage: Results are stored in MongoDB for quick access
7. User Interface: Users browse papers, insights, and full writeups in Streamlit

Installation

Prerequisites

• Python 3.8 or higher
• MongoDB database
• Google Gemini API keys (one for the analyst agent, one for the insights agent)
• Poetry (dependency management)

Local Setup with Poetry

1. Clone the repository:

   git clone https://github.com/yourusername/paperflux.git
   cd paperflux

2. Install dependencies using Poetry:

   # Install Poetry if you haven't already
   # curl -sSL https://install.python-poetry.org | python3 -
   
   # Install dependencies
   poetry install

3. Create a .env file with your credentials (copy from .env.example):

   cp .env.example .env
   # Edit .env with your credentials

4. Configure your environment variables:

   MONGODB_URI=mongodb+srv://username:password@cluster.mongodb.net/paperflux
   GEMINI_ANALYST_API_KEY=your_analyst_gemini_key
   GEMINI_INSIGHTS_API_KEY=your_insights_gemini_key
   GEMINI_ANALYST_MODEL=gemini-3.5-flash
   GEMINI_INSIGHTS_MODEL=gemini-3.5-flash
   

5. Run the Streamlit app with Poetry:

   poetry run streamlit run app.py

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.