Opyt vs Zotero

Zotero is the closest existing thing to Opyt's store. It fills one click at a time.

Of everything compared here, Zotero is the closest structural match to Opyt, and that is worth saying before anything else. It calls itself "a free, easy-to-use tool to help you collect, organize, annotate, cite, and share research"; it is open source, run by a nonprofit that says it has "no financial interest in your private information"; and "All library data is stored in an SQLite database, zotero.sqlite, within the Zotero data directory". It indexes the embedded text of attached PDFs, EPUBs and HTML in the background while idle. A local, private, full-text store of work other people published is the same shape as Opyt's store.

It is also already reachable by an agent. Recent desktop versions serve a local copy of the Zotero Web API on localhost:23119 under /api/, full-text endpoint included, once you enable "Allow other applications on this computer to communicate with Zotero" in Settings → Advanced; requests return 403 until you do. Zotero's own developer docs say direct access to the database "should be done only in a read-only manner". Several community MCP servers are built on that API, so a Zotero library already in place can be queried without Opyt.

The difference is the step before the store, and the unit inside it. Zotero fills when you click the Connector on a page you are already looking at, so the library is exactly the set of pages you visited and chose to save. And its unit is a citeable reference: an item type, authors, a container, a year. A thread, a repo and a Substack post are not references, and a person's whole archive is not a list of pages you clicked through one at a time.

At a glance

Where Zotero is better

• Citation styles, bibliographies, and Word, LibreOffice and Google Docs integration. This is why Zotero exists, and Opyt has none of it.
• Reading and annotating PDFs in the app, with notes attached to the item. Opyt keeps extracted text and has no reader.
• Group libraries, where storage is drawn from the owner's account at no extra charge. Opyt's sharing is a read-only query API, not a shared library.
• Mature, nonprofit and open source, on Windows as well as Mac and Linux, with 300 MB of free sync and a plain upgrade path at $20, $60 or $120 a year (checked 2026-08-31).
• If your sources are papers you already collect by hand, Zotero plus a community MCP server is a complete answer and Opyt adds nothing.

Where Opyt is better

• Acquisition. A Zotero library is the set of pages you personally visited and clicked to save. Opyt starts from 691 bookmarks, 99 liked authors, 492 follows and 19 subscriptions already sitting in your account, none of which a Connector click produced.
• Person-expansion. One confirmation pulls 3,412 posts, 38 repos and 96 essays. Zotero has no notion of an author's archive, only of items you added one at a time.
• Sources that are not references. 2,417 posts, 267 repos, 28 blog archives and 24 papers rank in one index; a thread has no natural item type in a bibliographic schema.
• Semantic ranking. Zotero's full-text search is keyword matching over indexed attachment content. Opyt ranks with BM25 plus cosine over 8,095 stored 4,096-dimension vectors, measured at 229 ms and 84 MB resident.
• The growth loop. 67 documents in the live store arrived from standing queries nobody typed. Zotero never goes and gets anything.

Which one

Zotero is the closest existing thing to Opyt's store, and it is better at everything downstream of one: annotation, citation, sharing. Opyt is better at the step before it. Zotero waits for a click; Opyt starts from what you already follow and goes and gets the rest.

install Opyt free, MIT, five minutes