zotero-mcp

Overview

Zotero MCP is a sophisticated integration that connects AI assistants directly to your local Zotero library through the Model Context Protocol. The plugin provides an integrated MCP server using Streamable HTTP protocol, eliminating the need for a separate server process. It offers advanced features including multi-dimensional library search, PDF content extraction, annotation analysis, collection management, semantic search via embedding vectors, and write operations for notes, tags, and metadata. The system supports both English and Chinese documentation and provides a streamlined experience for academic research and writing.

Try asking AI

After installing, here are 6 things you can ask your AI assistant:

• search_libraryq, title, titleOperator, yearRange, fulltext, fulltextMode, itemType, includeAttachments, mode, relevanceScoring, sort, limit, offset

  Advanced library search with multi-dimensional filtering and relevance scoring

  Try:Find papers about machine learning from 2020-2023Try:Search all documents containing 'neural networks' in full text
• search_annotationsq, itemKeys, types, colors, tags, mode, limit, offset

  Search annotations by query, colors, or tags

  Try:Find all yellow highlights in my PDFsTry:Show me annotations tagged with 'important'
• search_fulltextq, itemKeys, mode, contextLength, caseSensitive

  Full-text search across all document content with context snippets

  Try:Find all mentions of 'cognitive dissonance' in my libraryTry:Search case-sensitive for 'DIAGNOSIS' in PDFs
• get_item_detailsitemKey, mode

  Get complete metadata for a single item

  Try:Show me all details for item ABC123Try:Get full metadata for this citation
• get_contentitemKey, attachmentKey, mode, include, contentControl, format

  Extract PDF full-text, notes, abstracts, webpage snapshots from items

  Try:Get the full text of this PDFTry:Show me all notes for this article
• get_collectionsmode, limit, offset

  Get all collections in the Zotero library

  Try:List all my collectionsTry:Show me first 10 collections
• get_collection_detailscollectionKey

  Get details of a specific collection

  Try:Show details for my 'Machine Learning' collectionTry:What's in collection XYZ789?
• semantic_searchquery, topK, minScore, language

  AI-powered semantic search using embedding vectors

  Try:Find papers similar to about transformer architecturesTry:Search conceptually for studies on attention mechanisms
• write_noteaction, parentKey, noteKey, content, tags

  Create or modify Zotero notes

  Try:Add a note to this paper about methodologyTry:Update my existing note on this article
• write_tagaction, itemKey, tags

  Add, remove, or replace tags on items

  Try:Add tag 'important' to this paperTry:Remove tag 'toread' from this item
• write_metadataitemKey, fields, creators

  Update metadata fields on items

  Try:Change the title of this paperTry:Update the authors for this citation
• write_itemaction, itemType, fields, creators, tags, attachmentKeys, parentKey

  Create new items or reparent existing attachments

  Try:Create a new article entryTry:Move this PDF to be a child of this citation

Comparable tools

Installation

{
  "mcpServers": {
    "zotero": {
      "transport": "streamable_http",
      "url": "http://127.0.0.1:23120/mcp"
    }
  }
}

FAQ

What AI clients are supported?

The plugin supports Claude Desktop, Cherry Studio, Cursor IDE, and any client implementing the Streamable HTTP MCP protocol.

Does it work offline?

Yes, the plugin operates locally only, ensuring complete data privacy with no external server communication.

How do I enable semantic search?

Semantic search uses embedding vectors from OpenAI or Ollama APIs. Configure your API keys in Zotero preferences and rebuild the index via collection/item context menus.

Compare zotero-mcp with