arXiv Digest

Built

A small digest of recent arXiv papers in AI, machine learning and language, with tracked keywords ranked first.

Design

Python standard library only.
Public, keyless arXiv Atom API.
Injectable network layer, so tests are fully offline.
Transparent cache fallback on transport or parse failure.
Ranking favors keyword hits in titles, then abstracts, while preserving recency.

Rendered from the project's PROJECT.md, with setup instructions and anything private left out.