zotero_mcp

What is zotero_mcp?

This server connects Zotero via the Model Context Protocol (MCP) using a custom server and client, allowing interaction without tools like the Claude app or Cursor. It is for users who want to programmatically query Zotero library content through an LLM.

How to use zotero_mcp?

Set up a Python 3.12 conda environment and install requirements from requirements.txt. Create a .env file with keys: zotero_api_key, api_key (for the LLM), model, llm_api_base, llm_api_key, and library_id. Run python ./server.py to start the server, then in a new terminal run python ./client.py to interact.

Key features of zotero_mcp

• Custom MCP server and client for Zotero
• No dependency on Claude app or Cursor
• Query document content from Zotero via MCP
• Supports multiple LLM configurations via .env
• Simple Python setup with conda environment

Use cases of zotero_mcp

• Querying Zotero library documents programmatically
• Integrating Zotero references with LLM-based workflows
• Building custom MCP clients for academic research
• Retrieving and analyzing uploaded file content

FAQ from zotero_mcp

How do I obtain my Zotero API key?

Visit Zotero settings to create an API key. The key is set as zotero_api_key in the .env file.

Where can I find my library ID?

Your library ID is displayed on the Zotero keys settings page (shown in a screenshot in the README). Set it as library_id in .env.

What runtime dependencies are required?

Python 3.12, conda, and the packages listed in requirements.txt. A .env file with valid API keys is mandatory.

Can I use this server without the Claude app?

Yes, the custom server and client work independently; no Claude app or Cursor is needed.

What LLM configuration is needed?

The .env must include the LLM model name, its API base URL, and its API key. The api_key field is the LLM’s API key.