This open-source tool analyzes your Zotero reference library to discover new arXiv preprints each day and delivers them as a personalized email digest. It runs on a scheduled workflow, such as free GitHub Actions, requiring no manual intervention to fetch, rank, and email paper recommendations.
To match new papers to your research interests, the tool uses embedding-based semantic similarity between your library’s entries and daily preprint titles and abstracts. You can extend its sources and ranking logic by decorating Python classes or functions, and opt into AI-generated paper summaries by connecting a configurable language model through a pluggable backend. All workflow parameters and secrets are controlled via environment variables, so the pipeline adapts without code changes.
The system pulls preprints from the arXiv API (and other supported servers), computes relevance scores, and sends a daily email containing links, author details, and code repositories. Email delivery uses SMTP or a transactional API, and the entire discovery and delivery cycle runs automatically on a cron schedule.
