From a longtime Zotero user
I have used Zotero for years to manage my personal reference library, and I built Scholar Sidekick to solve a different problem. They are not the same kind of tool, and a fair comparison should say so up front.
Zotero is a reference manager - a stateful, library-based application for collecting, organising, annotating, and citing scholarly sources over the long term. Scholar Sidekick is a stateless API and MCP server for resolving identifiers and producing deterministic citation output, designed to be called from scripts, agents, and pipelines.
These tools are complementary, not competitors. The page below sets out what each does best, the small overlap between them, and how to use them together.
| Need | Reach for | Why |
|---|---|---|
| Building a long-term reference library for a thesis, book, or research programme | Zotero | Persistent storage, tags, notes, attachments, full-text search across PDFs. |
| Citing sources inside a Word, LibreOffice, or Google Docs manuscript | Zotero | Mature word-processor plugins; live cite-while-you-write integration. |
| Resolving a list of DOIs or PMIDs to formatted citations from a script or CI job | Scholar Sidekick | REST API with batch input, deterministic output, no account or local install. |
| Letting an AI agent (Claude, Cursor, Copilot) format or verify citations on demand | Scholar Sidekick | Native MCP server; no headless Zotero rig required. |
| Capturing references from a webpage as you browse | Zotero | The Zotero Connector is the gold standard for browser capture and metadata extraction. |
| Producing a one-off citation for a talk, poster, or grant - without installing anything | Scholar Sidekick | Free anonymous web access and a browser extension; nothing to set up. |
| Sharing a reference collection with a research group | Zotero | Group libraries with sync, permissions, and offline access. |
| Generating reproducible, version-pinned citation output for a publication or audit | Scholar Sidekick | x-scholar-transform-version header pins the formatter chain to a specific snapshot. |
| Feature | Zotero | Scholar Sidekick |
|---|---|---|
| Primary model | Stateful library | Stateless API |
| Account required | Recommended for sync | No (anonymous tier) |
| Cost | Free; paid storage tiers | Free anonymous tier; paid via RapidAPI |
| License | AGPL (open source) | Proprietary SaaS; OpenAPI spec public |
| Identifier resolution | Browser Connector + DOI/ISBN lookup | DOI, PMID, PMCID, ISBN, ISSN, arXiv, ADS bibcode, WHO IRIS URL |
| Citation styles | 10,000+ CSL styles | 10,000+ CSL styles + 5 hand-tuned builtins (Vancouver, AMA, APA, IEEE, CSE) |
| Export formats | BibTeX, RIS, CSL JSON, EndNote, MODS, Wikipedia, and more | BibTeX, RIS, CSL-JSON, EndNote XML, RefWorks, NBIB, RDF, CSV, plain text |
| Batch resolution | Yes, via UI or API | Yes, via REST or NDJSON streaming endpoint |
| MCP / agent integration | Community plugins (zotero-mcp, etc.) | First-party MCP server |
| Browser extension | Zotero Connector (rich capture) | Right-click formatter for any selected identifier |
| Word processor plugin | Word, LibreOffice, Google Docs | None (use Zotero for cite-while-you-write) |
| Library / collections | Yes (folders, tags, related items, notes) | No (stateless by design) |
| PDF storage and annotation | Yes (built-in reader in Zotero 7) | No |
| Group / shared libraries | Yes (free up to 25 MB; paid tiers from $20/yr) | Not applicable |
| Mobile app | iOS, Android (newer; basic) | Web (mobile-friendly); no native app |
| Offline use | Yes (desktop app) | No (web service) |
| Determinism / output stability | Depends on installed CSL pack version | x-scholar-transform-version header pins formatter chain |
| Provenance manifest | Documented in source code | Public /.well-known/sources.json with resolver chain and fallback order |
| Self-verification kit | Manual | /verification page with copy-paste curl commands |
| Plugin ecosystem | Mature (Better BibTeX, ZotFile, Zutilo, others) | Not applicable (server-side product) |
| Maturity | Released 2006; widely adopted in academia | Released 2025; micro-SaaS |
Zotero is a complete reference manager, and that is the bar Scholar Sidekick is not trying to clear. If your work involves any of the following, Zotero is the right tool and there is no honest case for replacing it:
Scholar Sidekick is a programmatic citation surface - built for scripts, agents, CI pipelines, and any workflow where citations need to be generated, exported, or verified without a human in the loop. The places it pulls ahead of Zotero are exactly the places Zotero is not designed for:
The most useful workflow for many academics is to use Zotero and Scholar Sidekick side by side, each for the job it is best at. A few concrete patterns:
No, and we do not recommend it. Scholar Sidekick is a stateless citation API and MCP server; it is not a reference manager. If you need a long-term library, PDF storage, browser capture, or word-processor cite-while-you-write, Zotero is the right tool. Scholar Sidekick is built for scripts, agents, and pipelines that need formatted citations on demand.
No. Scholar Sidekick is stateless. Each request resolves identifiers and returns formatted output; nothing about your library or read history is retained. See the privacy policy and the /.well-known/sources.json manifest for what data flows through the resolver.
There is no library to import into. If you have a Zotero collection and want a formatted bibliography or export file, export the collection as a list of DOIs or BibTeX from Zotero, then send those identifiers to the Scholar Sidekick /api/format or /api/export endpoint. Many users wire this up as a one-line script.
Zotero, by a wide margin. Cite-while-you-write inside Word, LibreOffice, or Google Docs is what reference managers exist for, and Zotero's plugins are mature and reliable. Scholar Sidekick does not try to replace that workflow.
Different jobs. Community Zotero MCP servers expose the contents of your local library to an AI agent - useful when the agent needs to read your saved references. Scholar Sidekick's MCP server resolves arbitrary identifiers from any source into formatted citations and export files - useful when the agent is generating new citations, not reading existing ones. The two can run side by side.
For citation formatting, both tools use the same Citation Style Language definitions, so the rendered output should match for any given style. Differences typically come from upstream metadata: Scholar Sidekick documents its resolver chain and fallback order in /.well-known/sources.json and pins behaviour with the x-scholar-transform-version header. The verification page contains curl commands that reproduce specific outputs against the live API.
Cite Zotero by referencing the project at zotero.org and noting the version used. Cite Scholar Sidekick by URL (https://scholar-sidekick.com) and including the x-scholar-transform-version value from the response, which pins the formatter chain to a specific snapshot for reproducibility.