ltczding-gif/ref-downloader

ref-downloader solves a very specific pain for academic researchers: when you read a paper and want to download all of its cited references as PDFs, doing so by hand means clicking through dozens of publisher websites one at a time, often hitting login walls. This tool automates that whole process. You give it a DOI, a unique identifier for an academic paper, or a PDF file, and it automatically fetches the reference list through Crossref, a scholarly metadata service, then attempts to download each referenced paper as a PDF. What makes it different from generic scrapers is that it drives your actual Microsoft Edge browser session. If your university already has a subscription to a journal, your login credentials stored in Edge carry through automatically, so papers you legitimately have access to can be downloaded without any extra setup. It is explicit that this is not a paywall bypass, references your institution does not subscribe to become manual_pending for you to chase by hand. It handles 17 or more publisher specific download paths, because each publisher such as Wiley, Elsevier, or AIP structures their download pages differently. After each run, it produces a report showing which papers downloaded successfully, which ones hit a paywall your institution does not cover, and which ones failed due to a technical issue. The tool also remembers progress per project, so rerunning it after a crash or dropped connection skips already downloaded references and retries only the failures. You would use this during a literature review when you need to read all the papers cited in a key publication. It runs on Python 3.11 or newer, currently only verified on Windows with Microsoft Edge, and it can also be installed as a skill for AI coding agents like Claude Code or Codex CLI.