zotero-mcp and mcp-zotero

These are competitors offering overlapping MCP server implementations for Zotero integration with Claude, with Tool A featuring more mature development (441 vs 5 stars) and broader functionality including metadata management and full-text analysis, while Tool B offers a lighter-weight alternative with focused operations like DOI-based item addition and open access PDF discovery.

zotero-mcp
60
Established
mcp-zotero
45
Emerging
Maintenance 13/25
Adoption 10/25
Maturity 24/25
Community 13/25
Maintenance 10/25
Adoption 4/25
Maturity 18/25
Community 13/25
Stars: 441
Forks: 29
Downloads:
Commits (30d): 0
Language: TypeScript
License: MIT
Stars: 5
Forks: 2
Downloads:
Commits (30d): 0
Language: TypeScript
License:
No risk flags
No risk flags

About zotero-mcp

cookjohn/zotero-mcp

Zotero MCP Plugin 是一个 Zotero 插件,通过 MCP协议实现 AI 助手与 Zotero深度集成。插件支持文献检索、元 数据管理、全文分析和智能问答等功能,让 Claude、ChatGPT 等 AI 工具能够直接访问和操作您的文献库。 Zotero MCP Plugin enables integration between AI assistants and Zotero through MCP.

Implements an integrated MCP server directly within the Zotero plugin using Streamable HTTP protocol, eliminating the need for a separate server process. Beyond search and retrieval, it provides semantic search via embedding vectors (OpenAI/Ollama), fine-grained PDF content extraction across four modes, annotation analysis by color/tags, and write operations for creating notes and items. Built on Node.js and TypeScript, it targets Claude Desktop, Cherry Studio, and other MCP-compatible AI clients through HTTP-based bidirectional communication.

About mcp-zotero

Xevos117/mcp-zotero

MCP server that exposes Zotero library operations as tools for Claude. Supports searching, browsing collections, adding items by DOI, importing PDFs with fulltext indexing, open access PDF discovery via Unpaywall, and injecting Zotero citation field codes into .docx documents.

Built on the Model Context Protocol, it uses a **self-documenting architecture** where workflow instructions and tool guidance are embedded in MCP's `instructions` field, enabling LLMs to discover full functionality without external docs. The server implements 15 tools across library operations, PDF management, and citation injection, with optional Claude skill support for sandboxed citation injection in Claude Desktop and Claude.ai Projects (which lack filesystem access). Configuration supports granular safety controls via `UNSAFE_OPERATIONS` to prevent accidental deletions, and integrates with Unpaywall for open-access PDF discovery.

Scores updated daily from GitHub, PyPI, and npm data. How scores work