This project provides an open-source, zero-server platform for discovering and reading AI research papers from arXiv and OpenReview. It targets researchers and students by automating paper discovery, providing personalized recommendation feeds, enabling contextual reading, and facilitating AI-powered Q&A. The core benefit is a streamlined, cost-effective workflow deployable entirely via GitHub Actions and Pages.

How It Works

The platform uses a serverless architecture, leveraging GitHub Actions for automated updates and deployment to GitHub Pages. It features a recommendation engine using keywords, interests, and intent queries, incorporating vector search and LLM refinement. A recent shift to a "pure frontend" approach moves configuration and token management client-side. This design eliminates dedicated servers, prioritizing ease of deployment and cost reduction.

Quick Start & Requirements

Setup is a "Fork-and-Run" process, estimated at 5 minutes.

• Action: Fork the repository.
• Prerequisites:
  • A Large Model API Key (e.g., Platypus API Platform).
  • A GitHub Personal Access Token (PAT) with repo, workflow, gist permissions.
• Configuration: Enable GitHub Actions and GitHub Pages in repository settings.

Highlighted Details

• Automated daily aggregation of new papers from arXiv and OpenReview.
• Personalized recommendation feeds based on user-defined keywords and interests.
• Integrated AI-powered Q&A for papers.
• "Read in Context" view: abstracts, full text, and AI summaries in one interface.
• Zero-server deployment via GitHub Actions and GitHub Pages.
• Zotero integration for saving papers.

Maintenance & Community

Active development is evident through frequent updates. A community exists via QQ group (ID: 583867967) with over 1100 members, indicating user engagement.

Licensing & Compatibility

The provided README does not specify a software license. This creates ambiguity for usage rights, modification, and distribution, especially for commercial applications. Compatibility is primarily within the GitHub ecosystem.

Limitations & Caveats

The primary caveat is the lack of a declared software license, posing adoption risks. Functionality depends on external LLM APIs, potentially incurring costs. Initial setup requires obtaining and configuring API keys and GitHub PATs. The "pure frontend" architecture means sensitive credentials are client-side, requiring user diligence. Performance may be constrained by GitHub Actions/Pages limitations.