Get a daily digest of new papers and auto summaries of the ones you keep
Idea + starter prompt, not run by us yet
For: Researchers and analysts who follow new papers in a field
A Python script reads a topic file and checks arXiv daily and Google Scholar once a week, ignores anything it found before, and saves a Markdown digest with the date. The user skims it and saves good PDFs into topic folders. Overnight, another script launches one Claude Code agent per new PDF with a summary skill, which writes notes on method, sample, findings and limits.
Steps
- Write a topic file with the keywords you follow.
- Have Claude Code write a script that searches for new papers and skips ones already seen.
- Read the digest and save only the papers worth keeping, which keeps the queue small.
- Write a summary skill that asks for method, sample, findings and limits.
- Check a few summaries against the paper, since vague prompts just repeat the abstract.
Starter prompt
Write a Python script that reads my topics from [topic file], searches arXiv for new papers on them each day, skips any paper already listed in [seen file], and writes a dated Markdown digest to [folder]. Then write a second script that, for each new PDF in [saved folder], creates a summary note covering method, data, key findings, effect sizes and limitations. Test both on [small sample] and show me the output before scheduling anything.
Original post: chatprd.ai · found via agentusecases.ai. Written up in our own words. We haven't run it, so test it on made-up data first.