Zotero MCP MCP — Use Cases, Install & Live Demo

Why use it

Key features

Search by author, title, tag, collection, full-text
Pull BibTeX/CSL-JSON for any item
Extract highlights and annotations from PDF attachments
Works with Zotero 7 local API and Zotero Web API
Read-only by default — safe to attach to your research library

Live Demo

What it looks like in practice

zotero-mcp.replay ▶ ready

0/0

Install

Pick your client

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "zotero-mcp": {
      "command": "uvx",
      "args": [
        "zotero-mcp"
      ]
    }
  }
}

Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "zotero-mcp": {
      "command": "uvx",
      "args": [
        "zotero-mcp"
      ]
    }
  }
}

Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "zotero-mcp": {
      "command": "uvx",
      "args": [
        "zotero-mcp"
      ]
    }
  }
}

Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "zotero-mcp": {
      "command": "uvx",
      "args": [
        "zotero-mcp"
      ]
    }
  }
}

Same shape as Claude Desktop. Restart Windsurf to pick up changes.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "zotero-mcp",
      "command": "uvx",
      "args": [
        "zotero-mcp"
      ]
    }
  ]
}

Continue uses an array of server objects rather than a map.

~/.config/zed/settings.json

{
  "context_servers": {
    "zotero-mcp": {
      "command": {
        "path": "uvx",
        "args": [
          "zotero-mcp"
        ]
      }
    }
  }
}

Add to context_servers. Zed hot-reloads on save.

claude mcp add zotero-mcp -- uvx zotero-mcp

One-liner. Verify with claude mcp list. Remove with claude mcp remove.

Use Cases

Real-world ways to use Zotero MCP

Synthesize a literature review from tagged papers

👤 Grad students, researchers ⏱ ~60 min intermediate

When to use: You've tagged 30 papers with #thesis-ch3 and need to write section 3.2.

Prerequisites

Zotero running (local API enabled) — Zotero 7: Settings → Advanced → enable local API
API key if using web — zotero.org/settings/keys

Flow

Pull the set

List every item tagged thesis-ch3 in my Zotero. Just titles + years + first author.✓ Copied

→ Clean inventory, ~30 items
Cluster

Group those by methodological approach. Cite each item by zotero key.✓ Copied

→ 3–5 clusters with per-cluster citations
Extract + write

For cluster "quantitative RCTs", pull my highlights from each item and draft a synthesis paragraph with in-line citations.✓ Copied

→ Paragraph with (Author, Year) citations backed by actual highlights

Outcome: A drafted section 3.2 you can edit instead of starting from blank.

Pitfalls

PDF has no highlights, just scans — Run OCR in Zotero first (right-click → OCR PDF), then re-pull

Combine with: filesystem

Generate the bibliography for a paper you're writing

👤 Anyone writing with LaTeX ⏱ ~15 min beginner

When to use: You've cited 40 items across a draft and need a single references.bib.

Flow

Scan for cites

In /paper/main.tex, list every \cite{} key.✓ Copied

→ All cite keys enumerated
Resolve

For each, pull BibTeX from Zotero and write the result to /paper/references.bib. Flag any key with no match.✓ Copied

→ bib file written; unresolved cites listed

Outcome: A clean references.bib tied to your Zotero — no manual copy-paste.

Pitfalls

Cite key mismatch (Better BibTeX vs default) — Install the Better BibTeX extension in Zotero for stable pinned keys

Combine with: filesystem

Combinations

Pair with other MCPs for X10 leverage

zotero-mcp + filesystem

Write the extracted synthesis to your paper draft

Write the synthesis from Zotero into /paper/section-3-2.md.✓ Copied

zotero-mcp + memory

Remember per-paper key arguments across sessions

Store the 1-sentence thesis of each item from cluster A in memory.✓ Copied

Tools

What this MCP exposes

Tool	Inputs	When to call	Cost
search_items	query: str, tag?: str, collection?: str	Find items in library	free
get_item	key: str	Inspect one item in depth	free
get_annotations	key: str	Pull your highlights/notes on a PDF	free
get_bibtex	keys: str[]	Build bibliography for a draft	free
list_collections	none	Navigate your organization	free

Cost & Limits

What this costs to run

API quota: Local API: unlimited. Web API: rate-limited by Zotero.
Tokens per call: Item metadata ~200 tokens; annotations can run large
Monetary: Free
Tip: Prefer search_items with tag filters over listing all items

Security

Permissions, secrets, blast radius

Minimum scopes: library:read

Credential storage: API key in env var (web mode only)

Data egress: api.zotero.org (web mode) or localhost (local mode)

Never grant: library:write unless truly needed

Troubleshooting

Common errors and fixes

ECONNREFUSED localhost:23119

Zotero isn't running, or local API is disabled. Open Zotero; Settings → Advanced → enable local API.

Verify: `curl http://localhost:23119/api/users/0/items`

Web API 403

API key missing the library read scope. Regenerate at zotero.org/settings/keys.

BibTeX keys are auto-generated nonsense

Install Better BibTeX add-on and set a stable citekey format

Alternatives

Zotero MCP vs others

Alternative	When to use it instead	Tradeoff
Obsidian MCP	Your notes live in Obsidian, not Zotero	Different data model — not paper-centric
Direct Zotero CSL export	One-time export, no LLM involvement	No querying, no synthesis

More

Resources

📖 Read the official README on GitHub

🐙 Browse open issues

🔍 Browse all 400+ MCP servers and Skills