Zotero MCP Server

Access research libraries to summarize papers, search citations, and manage collections.

Data last scanned last week · Reviewed last month

23 2 dangerous tools40 tools★ 4,488PythonMIT

Overview

The Zotero MCP Server integrates your Zotero research database directly into AI assistants like Claude and ChatGPT. It supports advanced features such as vector-based semantic search across your library, enabling conceptual discovery through various embedding models like OpenAI, Gemini, or local alternatives. Users can retrieve detailed metadata in Markdown or BibTeX formats and extract PDF annotations alongside table of contents data for streamlined analysis. Beyond reading, the tool facilitates write operations including adding new papers via DOI, URL, or local file upload, as well as managing collections and updating item tags. It is designed to work with various clients like Cursor, Cherry Studio, and Chorus, providing a centralized interface for research workflows.

Our verdict

This project holds a security score of 53/100, indicating a high-risk profile that requires user caution. The score reflects that the server is community-maintained rather than vendor-backed, and critical security aspects—specifically the implementation of OAuth and read-only mode—have not yet been editorially reviewed. Given these unverified security mechanisms, it is best suited for power users who are comfortable managing local research data in a non-enterprise environment.

Zero direct dependencies minimizes supply-chain surface area
Active maintenance with recent commits
Strong community engagement with 67 contributors
Broad compatibility with multiple AI assistant platforms

OAuth and read-only modes lack security verification
Community-maintained status lacks formal vendor oversight

Tools

Tool	Description	Risk
zotero_semantic_search	AI-powered similarity search with embedding models	low
zotero_update_search_database	Manually update the semantic search database	medium
zotero_get_search_database_status	Check database status and configuration	low
zotero_search_items	Search your library by keywords	low
zotero_advanced_search	Perform complex searches with multiple criteria	low
zotero_get_collections	List collections	low
zotero_get_collection_items	Get items in a collection	low
zotero_get_tags	List all tags	low
zotero_get_recent	Get recently added items	low
zotero_search_by_tag	Search your library using custom tag filters	low

Compatibility

Client	Local	Docker	Remote	Read-only
ChatGPT
Claude Desktop
Cursor
VS Code
Windsurf

Frequently asked questions

›Can I use the Zotero MCP server without installing heavy ML dependencies?

Yes, the base installation is lightweight and includes core search, metadata retrieval, and annotation tools without pulling in optional ML or PDF-heavy dependencies.

›What is the difference between Local, Web API, and Hybrid access modes?

Local mode allows for offline library access without needing an API key, while Web API mode supports cloud-based library access; hybrid mode combines local reading with web API writes.

›Do I need a paid account to use Scite citation intelligence features?

No, the scite integration functions without a paid account by utilizing public API endpoints to provide citation tallies and retraction alerts.

›How can I perform semantic search over my research library?

You must install the server with the semantic extra using pip install "zotero-mcp-server[semantic]", which enables ChromaDB and various embedding models.

›Does the server provide a command-line interface independent of an AI assistant?

Yes, it includes a standalone zotero-cli that allows you to search, browse, and edit your library directly from the terminal using short aliases.