Staying current without drowning

1 · Learn the move · Tool choice + routing

Staying current fails in two directions: the unread pile that becomes guilt furniture, and the feed you gave up on. The fix is routing, not effort. Name your sources deliberately - two or three journal alerts, one preprint server query, maybe a curated list - and give each a route: everything lands in one weekly triage, processed once. The model's job is the digest: you supply the week's titles and abstracts, it sorts them against your standing interests into READ (the few), SKIM (the maybe), SKIP (the most, with one honest line why). Rules that keep the digest trustworthy: it processes only what you feed it, never implies knowledge beyond an abstract, and preprints are labeled preprints every time - unreviewed work routes differently than published work. The digest triages; the reading, as ever, is yours.

Build my weekly literature digest from what I paste - nothing else. My standing interests, ranked: [interest 1], [interest 2], [interest 3]. Sources this week: [journal alerts, preprint queries]. Sort every item into READ (directly on a ranked interest - cap at [N]), SKIM (adjacent or method-relevant), SKIP (one honest line why). Rules: abstracts only, and the digest says so; every preprint labeled PREPRINT - NOT PEER REVIEWED at the item, not in a footnote; no item summarized beyond what its abstract states; anything matching [standing red-flag topics, e.g. competitor's system] flagged URGENT regardless of pile. Close with counts and a one-line check: which interest got zero hits this week.