Radar and Zotero

Both live on tabs of your Library.

Radar

Radar watches each of your research questions and surfaces newly published papers so you can triage them in seconds.

What it needs: at least one research question. Without one the tab explains how to add one.

When it runs: every day at 12:00 UTC, across arXiv, CrossRef and OpenAlex, with each question on the schedule of its stake level (set in the question’s details: Details on its group header in the Library): Frontier every day and deeper, Live — and any question without a level — every Monday, Bedrock every four weeks and shallower. Each question’s section shows its schedule and next sweep. Click Run now to sweep all your questions immediately.

What it searches for: each question’s watch terms — set them in the question’s section header. With no terms, the question statement itself is used. You can also add a term from the Search tab: run a search, then Get alerts for this search → Watch on a question.

Searched for evidence against

Radar also looks for papers that could argue against each question: null results, failed replications, an opposite effect, or a rival explanation. What it finds is listed under Searched for evidence against in the question’s section, apart from the watch results, and you triage it the same way. Radar does not call these papers conflicting or challenging. It found them by searching against your question; you decide what they mean.

• The terms: Radar writes two search terms for each question from its statement and working hypothesis, the first time it runs this search for the question. Edit them in the section like watch terms. They are rewritten when you change the statement or the working hypothesis. Remove them all to stop this search for the question.
• When it runs: for a Frontier or Bedrock question, with each sweep. For a Live question, and any question without a level, once a week on a weekday between Tuesday and Friday (each question keeps its own day). Run now sweeps the watch terms only.
• How much: Frontier and Live questions search two terms, Bedrock questions one. Each search keeps fewer new papers than a watch sweep, so the counter-search does not crowd the tab.
• The weekly email lists them under each question, apart from the other papers, under Searched for evidence against.

Triage

The Radar tab lists new papers under each question, with a note on why each one might matter. Click a paper (or press J / K to move), then:

• A — Add it to your Library, attached to the question as unscreened.
• D — Dismiss it. Dismissed papers never come back for that question.

The header shows how many questions Radar is watching, how many papers are waiting, and your progress through today’s queue. Radar only surfaces papers — it never includes, excludes or decides anything for you.

Where you’ll hear about new papers

• the Radar tab, with a count badge;
• a card in Pulse with an Open Library button;
• a weekly email digest on Monday afternoons, listing up to five papers per question that Radar found since the previous digest (and up to five more it found searching for evidence against). Once your questions have levels, it is grouped by them: Frontier questions first, then the papers already sent to you as Frontier alerts (listed, not counted again), then Live questions and those without a level, then Bedrock questions as one line each. To stop the emails, turn off Radar digest under Settings → Notifications. The in-app trail stays.

Frontier alerts. For a question marked Frontier, you don’t wait for Monday. When a sweep finds a paper that Radar’s relevance check rates directly relevant (it answers, tests or could scoop the question), you get:

• a card in Pulse, one per question per day (later finds that day join it);
• a notice in the app, if you have CoChat open;
• an email listing the papers, at most one a day;
• a push notification with the first paper’s title, at most one a day, on each device where push is on. Tapping it opens Radar.

Frontier paper alerts under Settings → Notifications has a switch for the email and a switch for the push; the Pulse card and the notice stay either way.

Live and Bedrock questions, and questions without a level, never send alerts. Neither do papers from Searched for evidence against.

Zotero

One connection, set up with a Zotero API key. Your references, notes and collections sync both ways through Zotero’s own service; once connected you can also choose where your PDFs are kept. There is no separate “file storage” connection to set up first — the key is the connection.

Why one key: Zotero syncs in two channels. Everything except attachment files (items, notes, tags, collections) goes through zotero.org and is only reachable with an API key. Files go to either Zotero Storage or a WebDAV server. So without the key CoChat can store your PDFs but cannot know which paper any of them belongs to.

The tab shows what the connection does right now — References, Notes and PDFs — each marked Both ways, Into CoChat, Into Zotero or Off. Notes mirror both ways through the Zotero API, following the direction and permissions the strip shows (a read-only key brings notes in but sends none). Tags come into CoChat with the import and are not sent back. PDFs travel whichever storage you chose: through Zotero’s own storage and API, or through your CoChat storage. A paper CoChat sends to Zotero takes its PDF along; with Zotero storage that counts against your Zotero quota, and a full quota shows up in the sync history as Your Zotero storage is full. Sync keeps the two libraries matched. It isn’t a backup — for that, keep a copy of your Zotero data folder.

Connect

Create a key at zotero.org/settings/keys — read access is enough — and paste it in. Once connected, the card shows which library it’s tied to, the key’s last characters, whether it can write, and when the last import ran.

Settings on the card save immediately:

• Which library — your own library, or a group library by ID. A public group can be imported without any key. With a key that can write to the group, papers you find in CoChat are sent to the group too, with their PDFs (against the group owner’s Zotero storage); your notes stay personal and are never sent to a group.
• Sync direction — Both ways, Zotero → CoChat only or CoChat → Zotero only. A read-only key defaults to Zotero → CoChat only; switching to one direction parks the other side’s work without deleting anything.
• Zotero tags — import them as your personal tags. On by default.

How often it syncs. The first sync walks your whole library and records where it got to. From then on CoChat asks Zotero only for what changed and runs every hour; the card says syncing hourly. Run again on a history entry walks the whole library once more. Items you delete in Zotero are unlinked here (the paper stays in your Library).

Notes. Each paper’s note in CoChat mirrors one child note on the Zotero item: a note you write in Zotero appears here, a note you write here appears in Zotero tagged cochat. If both changed since the last sync, nothing is overwritten; the history lists 1 note needs a decision with Keep Zotero and Keep CoChat. Notes are not imported onto a paper whose note field belongs to a shared project question, because project members can read that field.

Sync runs on its own and whenever you click Sync now. Item imports run hourly once the first full sync of your items has completed, daily until then. A CoChat → Zotero only connection never imports items; it sends your note edits to Zotero hourly instead.

Zotero → CoChat

Click Sync now (or wait for the next automatic sync). Items with a DOI, an arXiv id or a resolvable URL become papers in your Library. Anything without an identifier is listed for review instead — it never becomes a phantom paper. Notes mirror both ways (see above); a paper’s PDF attachment comes along — mirrored from Zotero storage, or read from CoChat storage — and opens in the Library’s PDF tab; annotations aren’t imported yet.

Imported papers arrive as Checking… and are verified over time.

The import history below shows each run, live progress while one is running, what was imported, anything that needs an identifier, and Run again. A failed run changes nothing.

CoChat → Zotero

With the direction set to Both ways or CoChat → Zotero only, every paper you add to your Library travels to Zotero by itself: a Zotero item with its citation details lands in a CoChat collection, filed under a sub-collection named after the research question it belongs to (Inbox papers sit in CoChat directly), and when CoChat holds the PDF — your own upload, an open-access copy, or a cached one — it arrives as the item’s Full Text PDF — into CoChat storage, or into Zotero storage against your Zotero quota, whichever you chose below. Sync Zotero and it’s there.

Only papers with a real, verified DOI travel. Pasted files, uploaded documents and web pages without a DOI stay in CoChat. A paper you already have in Zotero is filed into the CoChat collection rather than duplicated, and its existing PDF is left alone. Turning storage on for the first time sends your whole Library over a few minutes; the storage card shows how many papers are synced, pending or skipped and why (no PDF found, storage quota, DOI not verified). This needs a Zotero API key with write access saved under Citations; a read-only key pauses the push until you save one that can write.

The manual route still exists: from Project → Knowledge → Sources, each research question has a Send to Zotero button (shown when your key can write). It exports the question’s included papers into that question’s collection, for a key you don’t want to leave storage on for.

Exporting again updates the same collection rather than making a new one. Papers already in Zotero aren’t duplicated, and if you’ve edited an item in Zotero since, CoChat leaves it alone rather than overwriting your changes. Progress appears under Citations on the Zotero tab.