Sort each week's new research papers into keep, skim or drop
Idea + starter prompt, not run by us yet
For: Researchers and analysts who follow new papers on a topic and fall behind
Every week a Codex job looks up the author's chosen keywords on arXiv and logs hits in a CSV, so each run only sorts fresh papers. Every paper is labelled, for example keep, skim or drop. Strong picks are sent to a Zotero inbox. The author still reads and checks each paper personally.
Steps
- Pick three to five search keywords for your topic.
- Paste the starter prompt into Codex and run it once by hand before you schedule it.
- Check the labels on ten papers yourself to see if the sorting matches your taste.
- Keep the CSV log so later runs skip papers already seen.
- Only add more automation once this simple loop saves you time.
Starter prompt
Search arXiv for new papers on [keywords] from the last [7] days. Check [papers.csv] first and skip any paper already listed there. For each new paper, add a row with the title, link, date and one label: keep, skim, drop, duplicate, off topic or weak method. Add one sentence explaining the label. Show me the new rows before you save them. Do not delete or rewrite old rows. Do not claim anything about a paper that is not in its abstract.
Original post: Substack · found via agentusecases.ai. Written up in our own words. We haven't run it, so test it on made-up data first.