Zotero MCP (zotero-mcp) connects a Zotero library to Claude, ChatGPT and other MCP clients, enabling semantic search, metadata access, annotations, and write operations via multiple install paths and optional extras.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Zotero MCP: Chat with your Research Library—Local or Web—in Claude, ChatGPT, and more.
How it works
The project provides a server (zotero-mcp-server) that interfaces with Zotero via local or web APIs and exposes features including semantic search, metadata extraction, annotation handling, and write operations. It supports multiple embedding models for semantic search (Default local, OpenAI, Gemini, Ollama) and optional extras to enable heavier ML/PDF features. The server can be configured for local reads only or hybrid mode (local reads + web API writes). It includes a standalone CLI (zotero-cli) and a set of setup/config steps for Phaser-like MCP clients through Claude Desktop/Claude Code, Autohand Code, and Cherry Studio.
Getting started
Quick install (core tools)
uv tool install zotero-mcp-server
zotero-mcp setup # Auto-configure (Claude Desktop supported)
Discarding extra features keeps the base install lightweight.
Installing via pip (example)
pip install zotero-mcp-server
zotero-mcp setup # Auto-configure ( Claude Desktop supported )
Optional extras
| Extra | What it adds | Install command |
| semantic | Semantic search via ChromaDB, sentence-transformers, OpenAI/Gemini embeddings | pip install "zotero-mcp-server[semantic]" |
| pdf | PDF outline extraction and EPUB annotation support | pip install "zotero-mcp-server[pdf]" |
| scite | Scite citation intelligence (no account needed) | pip install "zotero-mcp-server[scite]" |
| all | Everything above | pip install "zotero-mcp-server[all]" |
Example for all features:
uv tool install "zotero-mcp-server[all]" # Full install with all features
Semantic search setup
zotero-mcp setup
Or separately:
zotero-mcp setup --semantic-config-only
Available embedding models:
- Default (all-MiniLM-L6-v2) (local)
- OpenAI (requires API key)
- Gemini (requires API key)
- Ollama (local via Ollama API)
Ollama usage example:
ollama serve
ollama pull nomic-embed-text
zotero-mcp update-db --force-rebuild
Configuration examples for semantic embedding paths and timeout settings are provided in the README, including example JSON for embedding_config and notes on rebuilding embeddings when models change.
Getting started (semantic and data extraction)
After initial setup, build the semantic search database:
zotero-mcp update-db
Additional commands exist for batch embedding via OpenAI, status checks, and full-text indexing:
zotero-mcp update-db --openai-batch
zotero-mcp openai-batch-status
zotero-mcp openai-batch-import
zotero-mcp update-db --fulltext
Traction
Stars: 4541 Forks: 370 Open issues: 50
Behind the repo
The README indicates a PyPI package zotero-mcp-server and a registry entry, with a server.json describing the package. External clients include Claude Desktop, Claude Code, Cherry Studio, and Chorus.
Caveats
Requirements include Python 3.10+, Zotero 7+ (for local API with full-text access). Local API enabling steps are described for Zotero and client configurations for Claude/Claude Code, including environment variable usage for local vs hybrid modes. Updating installation and semantic search configuration may require re-building indices and re-extraction when changing embedding models or attachment priorities.