Say you follow a topic with a Discovery Agent. In March it surfaces a paper that looks relevant. You skim it, leave it in the feed, and move on. In June the journal retracts it.
Until recently, Folio would not have told you. A paper was checked for retraction once, at the moment it was first stored. If it was clean then, it stayed unmarked, even after the retraction notice went out. The only way it picked up a warning was if a later run happened to surface the same paper again, and a paper that has dropped out of the agent's results doesn't come back.
That was a real gap for anyone tracking a topic over months, and it is now closed.
What changed
The nightly retraction sweep used to cover only papers in your library. It now has a third pass over the items your Discovery Agents have already surfaced. Any item with a DOI that has no retraction flag yet is looked at again. If the paper has since been flagged, it gets the same retraction badge you'd see on a library source, with the notice link and reason when they are on record.
The badge appears on Radar and on the agent's page. A flagged paper is also no longer a candidate for "Check against my draft."
A few details, so you know what you're relying on:
- It runs once a night, after the library checks, so it picks up verdicts those checks just recorded.
- It reads what Folio already knows. The pass doesn't make a fresh call to CrossRef or OpenAlex for each item. It looks up the retraction verdicts Folio has already recorded for that DOI when it checked libraries. This keeps a nightly sweep across every agent cheap, and it means an item is flagged on exactly the same terms it would have been at the moment it was surfaced.
- It never overwrites. An item that already carries a flag is left alone, and running the pass twice gives the same result.
- It has a ceiling. Each run walks up to 5,000 items (50 pages of 100) within a time limit. If there is more than that to do, the rest waits for the next night.
What it doesn't cover
The limits are worth knowing now rather than discovering later.
Items without a DOI aren't re-checked. Matching is by DOI only. A title match across different people's libraries isn't enough to put "retracted" next to someone's search result, so Folio doesn't guess.
A paper nobody has checked yet stays unmarked. Because the sweep reads verdicts Folio has already recorded, a DOI that has never appeared in any checked library has no verdict to find. It isn't marked clean, it simply isn't marked. The more a paper is in circulation among Folio libraries, the sooner a retraction gets caught.
It is not a substitute for reading the notice. A badge tells you something happened. What it means for your argument depends on why the paper was retracted, which is why Folio links the notice instead of summarizing it.
No alert is sent. For library sources you get an email when something is flagged. Agent items don't. The badge is the signal, so you will see it the next time you open the feed.
If you cite from your feed
Treat the badge as one more reason to open the DOI before a paper goes into your reference list. A retraction is a fact about the paper, not about who found it, so the flag follows the paper wherever Folio holds it.
If you track a topic for months, the practical difference is that a paper you saw in March is no longer frozen at its March status.