Guide to using Zotero MCP to connect a Zotero reference library with AI assistants via the Model Context Protocol.
Tasks
Ask the AI assistant to search a specific Zotero collection by title keywords or author names and synthesize the returned bibliographic data into a literature review draft.
Request formatted citations in a specific style such as APA or MLA for selected items retrieved from your Zotero library through the AI assistant.
Instruct the AI assistant to summarize the PDF annotations and highlights you made for a specific paper in your Zotero library.
Inputs
Zotero reference library data including bibliographic metadata and full-text indexing
PDF annotations and standalone notes created within Zotero
Natural language queries sent from an MCP-compatible AI client
MCP server configuration parameters such as local API port 23119
Outputs
Natural language or structured search results from the Zotero library
Formatted citations in various styles such as APA, MLA, or Chicago
Collections lists and item metadata retrieved via the Model Context Protocol
Draft text and summaries generated from accessed annotations and notes
Limitations and checks
The MCP server provides read-only access and cannot modify the Zotero library
When used with a cloud-based AI assistant, content retrieved from the library is sent to the AI provider as part of the conversation context
Large libraries may result in slow response times
Full-text search requires Zotero full-text indexing to be enabled
Confirm Zotero is actively running before launching your AI client to ensure the MCP server can connect to the local API on port 23119.
Verify that Zotero's full-text indexing is enabled if your workflow depends on full-text content searches.
Review your AI provider's data retention and usage policies before accessing sensitive or embargoed research content through the integration.