home / mcp / zotero mcp server
Zotero MCP: Connects your Zotero research library with Claude and other AI assistants via the Model Context Protocol to discuss papers, get summaries, analyze citations, and more.
Configuration
View docs{
"mcpServers": {
"54yyyu-zotero-mcp": {
"command": "zotero-mcp",
"args": [
"serve"
],
"env": {
"ZOTERO_LOCAL": "true",
"GEMINI_API_KEY": "YOUR_GEMINI_KEY",
"OPENAI_API_KEY": "YOUR_OPENAI_KEY",
"ZOTERO_API_KEY": "YOUR_API_KEY",
"GEMINI_BASE_URL": "https://gemini.googleapis.com",
"OPENAI_BASE_URL": "https://api.openai.com/v1",
"ZOTERO_LIBRARY_ID": "YOUR_LIBRARY_ID",
"ZOTERO_LIBRARY_TYPE": "user",
"GEMINI_EMBEDDING_MODEL": "gemini-embedding-001",
"OPENAI_EMBEDDING_MODEL": "text-embedding-3-small",
"ZOTERO_EMBEDDING_MODEL": "default"
}
}
}
}Zotero MCP connects your Zotero research library with AI assistants such as Claude, ChatGPT, and others, enabling you to search, summarize, and extract insights from your library locally or via a web API.
You use Zotero MCP by running the local server on your machine and connecting an MCP client (for example Claude Desktop, ChatGPT, Cherry Studio, or Chorus) to it. You can perform semantic searches over your Zotero library, retrieve item metadata and full text when available, access notes and attachments, and extract PDF annotations for easy review. The server supports both offline local access and web-based access for remote collaboration.
Prerequisites you need before starting:
Choose a installation method and run the commands shown to install and set up Zotero MCP.
Default installation (recommended) using uv
uv tool install zotero-mcp-server
zotero-mcp setup # Auto-configure (Claude Desktop supported)Alternative installation using Python package manager
pip install zotero-mcp-server
zotero-mcp setup # Auto-configure (Claude Desktop supported)Alternative installation using Python via a wrapper (pipx) that keeps it isolated
pipx install zotero-mcp-server
zotero-mcp setup # Auto-configure (Claude Desktop supported)Updating your installation while preserving configuration
# Check for updates
zotero-mcp update --check-only
# Update to latest version (preserves all configurations)
zotero-mcp updateZotero MCP can be run locally for offline access or exposed via a web API for remote access. You can configure semantic search with different embedding models and update frequencies. The setup flow includes optional semantic search configuration and choices about how often the library index updates.
For local CLI usage and remote clients, you typically start the MCP server and then connect your client to the served endpoint or socket. Use the client to issue search queries, request metadata, export citations, or retrieve full text where Zotero provides access.
Configure your server to use the local Zotero API or the web API, and set environment variables to control embedding models, API keys, and database paths.
Key environment variables you may encounter include the following. Use placeholders where values are not provided in your setup.
Zotero MCP includes advanced PDF annotation extraction, allowing you to process annotations directly from PDFs, search through PDF annotations, and work with image annotations. It integrates with Zoteroβs native annotation system.
For best results, install the Better BibTeX plugin for Zotero, as annotation-related features are enhanced when the plugin is present.
When you use PDF annotation features for the first time, the necessary tools will download automatically.
General issues you may encounter and how to resolve them include verifying that Zotero is running and the local API is enabled, checking API keys and library IDs for web API access, and ensuring you are on a compatible Zotero version for local full-text access.
Semantic search related tips cover common issues such as missing environment variables, deprecation warnings, and indexing status checks.
If update commands fail, verify your internet connection and consider forcing an update. Preserve configurations by default, but review the MCP client config directory if you notice discrepancies after updates.
If you find Zotero MCP useful, consider supporting its development.
License: MIT
AI-powered similarity search with embedding models
Manually update the semantic search database
Check database status and configuration
Search your library by keywords
Perform complex searches with multiple criteria
List collections
Get items in a collection
List all tags
Get recently added items
Search your library using custom tag filters
Get detailed metadata (BibTeX export available)
Get full text content for items where available
Get attachments and notes for an item
Get annotations including direct PDF extraction
Retrieve notes from your Zotero library
Search notes and annotations including PDF-extracted content
Create a new note for an item (beta feature)