登录

zotero-mcp

54yyyuMIT活跃维护

把 Zotero 文献库接入 Claude 等 AI 助手的 MCP 服务，支持论文讨论、摘要生成、引用分析和语义检索。

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.

v0.17.0发布于 2026-11-10 16:30 UTC5,849star +58 / 7 天459fork同步于 2026-11-11 05:33 UTC
文献管理MCP服务语义检索引用分析AI助手
对比 GitHub 官网ai-agentchatgptclaudeclaude-codemcp

怎么上手

由 README 提炼，执行前请确认
原文 
uv tool install zotero-mcp-server
zotero-mcp setup
硬件要求按 README 说明

README 未说明硬件要求；仅要求 Python 3.10+，语义检索等可选功能依赖本地嵌入模型或 OpenAI、Gemini、Ollama 等外部服务。

下载：源码 zip·GitHub Releases（安装包）本站不镜像安装包

项目说明

由 README 自动翻译提炼
没人工点评？写下你的经验

把 Zotero 文献库接入 Claude、ChatGPT 等 AI 助手，让 AI 直接检索、阅读、标注和引用你的论文

把 Zotero 文献库接入 Claude 等 AI 助手的 MCP 服务，支持论文讨论、摘要生成、引用分析和语义检索。

核心能力

  • 按标题、作者、标签、全文或语义检索文献库
  • 读取 PDF 元数据、BibTeX、全文和指定页码范围
  • 在原文精确位置添加高亮、区域框和笔记
  • 按 DOI、URL、ISBN、BibTeX 或文件添加条目并管理合集标签
  • 本地模式直接读 zotero.sqlite，写入走 Zotero 10+ 或 Web API

容易踩的坑

需先在 Zotero 7+ 设置→高级中勾选允许其他应用与 Zotero 通信；语义检索、PDF、Scite 等功能需另装 extras，基础安装不含；Zotero 10+ 写入需先运行 zotero-mcp authorize-local，旧版则要配置 ZOTERO_API_KEY 和 ZOTERO_LIBRARY_ID。

内容由模型根据 README 生成，可能不完整或不准确，以官方文档为准。看原文 README

最近发版

5 个版本v0.17.0
全部 release
  1. v0.17.0最新正式版2026-11-10 · 2026-11-10 16:30 UTC

    What's Changed

    New Contributors

    Full Changelog: https://github.com/54yyyu/zotero-mcp/compare/v0.16.0...v0.17.0

  2. v0.16.02026-11-10 · 2026-11-10 00:02 UTC
    查看更新说明

    What's Changed

    Full Changelog: https://github.com/54yyyu/zotero-mcp/compare/v0.15.2...v0.16.0

  3. v0.15.22026-11-01 · 2026-11-01 00:32 UTC
    查看更新说明

    What's Changed

    New Contributors

    Full Changelog: https://github.com/54yyyu/zotero-mcp/compare/v0.15.1...v0.15.2

  4. v0.15.12026-10-16 · 2026-10-16 20:57 UTC
    查看更新说明

    What's Changed

    New Contributors

    Full Changelog: https://github.com/54yyyu/zotero-mcp/compare/v0.15.0...v0.15.1

  5. v0.15.02026-10-15 · 2026-10-15 20:45 UTC
    查看更新说明

    Added

    • The open-access PDF cascade asks OpenAlex for repository copies. When zotero_add_item or a batch import attaches a PDF for a DOI, it now tries every pdf_url OpenAlex lists after Unpaywall and before the publisher's citation_pdf_url, open-access locations first. OpenAlex knows institutional repository copies that Unpaywall does not count as open access. For Maraun 2013 in J. Climate, DOI 10.1175/jcli-d-12-00821.1, Unpaywall has nothing and the AMS page offers a 2-page preview, while OpenAlex lists the full 7-page PDF in the GEOMAR repository. A URL already tried under one source is not downloaded again under the next. Thanks @phil-c137.
    • Markdown and YAML files can be attached (#703). zotero-cli attach and add file accept .md, .yaml and .yml, stored as text/markdown and text/yaml so Zotero's own full-text indexer picks them up. .json stays a CSL-JSON import. Requested by @rizakardas.
    • zotero-cli --json get fulltext says when a read was cut short (#701). data.truncated is always present, and a capped read adds pages_read and page_count, so a script no longer has to parse the TRUNCATED heading out of data.text.

    Changed

    • Semantic search is its own toolset, and hidden on installs without the [semantic] extra (#572, #689). zotero_semantic_search moves from core into a new semantic toolset that is on by default, so default installs see no change. ZOTERO_MCP_TOOLSETS=none now leaves it out too; use none,semantic to keep it. A spec that uses none without naming semantic logs one warning at startup saying so; none,-semantic leaves it out without the warning. semantic and search-admin are left out automatically when ChromaDB isn't installed, so a base install no longer advertises tools that can only reply with install instructions; naming them in ZOTERO_MCP_TOOLSETS keeps them. The connector search keeps its keyword fallback. Reported by @zhuziqi97; thanks @falcon1-1-code.
    • The literature-review and contradicting-evidence prompts fall back to zotero_search_items when semantic search is not usable (#693). zotero_literature_review and zotero_find_contradicting_evidence always told the model to call zotero_semantic_search, which fails without the [semantic] extra and is absent when a toolset spec leaves it out. They now check at render time (ChromaDB found with find_spec, not imported, and the tool's toolset enabled) and otherwise name zotero_search_items with qmode='everything', one key term per query. With semantic search usable the text is unchanged. Thanks @falcon1-1-code.
    • zotero_advanced_search rejects a field name it does not know (#694). A typo such as titel was a valid-looking condition that matched nothing, so the tool answered "No items found matching the search criteria" and the mistake went unnoticed. It now returns an error naming the field, up to three close matches (Did you mean: title?) and the supported operations. Thanks @falcon1-1-code.
    • zotero_advanced_search accepts a single condition sent as a bare object (#695). {"field": "title", "operation": "contains", "value": "x"} instead of a list of one was rejected with a raw pydantic "Input should be a valid list" error, and the same object as a JSON string answered "No search conditions provided". It is now read as a one-condition list. An empty object is still an error. Thanks @falcon1-1-code.

    Fixed

    • A startup semantic update no longer holds up the MCP handshake (#706). With auto_update: true and update_frequency: "startup" (or a due daily run), the server ran the update in a worker thread of its own process, where native calls that hold the GIL (ChromaDB, loading or downloading the embedding model) and, on Windows, DLL loads stall the event loop as well: Claude Desktop's initialize went unanswered until the update finished, timed out, and shut the server down. The update now runs as a child zotero-mcp update-db proces

用户点评

登录用户的真实使用记录，可按版本和硬件筛选
共 0 条

    还没有人点评。第一个说出真实体验的人最有价值。

    README 原文

    未翻译
    在 GitHub 打开

    Zotero MCP connects your Zotero research library with ChatGPT, Claude, and other AI assistants (e.g., Cherry Studio, Chorus, Cursor) via the Model Context Protocol. Search your library, read and annotate papers, add and organize items, and find research by meaning.

    AI agents: read docs/for-agents.md first. It covers which route to use, setup, and the commands in one place.

    Three ways in

    One package, zotero-mcp-server, ships all three. Pick the one that fits where you work.

    MCP serverzotero-cli + agent skillZotero Agent
    ForChat apps that speak MCP but have no shell: Claude Desktop, ChatGPT, Cherry Studio, ChorusAgents with a shell: Claude Code, Codex, Cursor, Windsurf, Gemini CLI, Amp, OpenCodeChatting inside Zotero, next to the PDF you are reading
    Howzotero-mcp setupzotero-mcp install-skillzotero-cli plugin, then install the .xpi in Zotero
    WhyWorks in any MCP client98 tokens in context until it is needed, instead of ~17kKnows the item, page and selection you have open; answers link to pages
    GuideGetting startedCLI and agent skillZotero Agent plugin

    What it does

    • Search by title, author, tag, collection, full text, or meaning (semantic search with local, OpenAI, Gemini, or Ollama embeddings)
    • Read metadata, BibTeX, full text, and page ranges of PDFs, with page images where text extraction garbles math, figures, and tables
    • Annotate: highlights and area boxes placed on the exact words, figure, table, or equation; notes; PDF annotation extraction
    • Write: add papers by DOI, URL, ISBN, BibTeX, or file (with open-access PDFs), manage collections and tags, merge duplicates
    • Local or web: in local mode reads come straight from zotero.sqlite; writes go to the running Zotero 10+ or through the web API
    • Scite citation tallies and retraction alerts (optional)

    Quick start

    1. Install (Python 3.10+):

    uv tool install zotero-mcp-server     # or: pip install zotero-mcp-server

    New to the command line? Try the community-built Zotero MCP Setup: a macOS GUI installer, one-click scripts for Mac and Windows, and a step-by-step guide.

    2. Enable Zotero's local API: in Zotero 7+, open Settings → Advanced and tick Allow other applications on this computer to communicate with Zotero.

    3. Connect your assistant:

    zotero-mcp setup      # auto-configures Claude Desktop

    or add the server by hand to claude_desktop_config.json:

    {
  "mcpServers": {
    "zotero": {
      "command": "zotero-mcp",
      "env": { "ZOTERO_LOCAL": "true" }
    }
  }
}

    Teach the agent to drive zotero-cli (the cheaper route, see below):

    zotero-mcp install-skill

    Or use the MCP server: for Claude Code, add the same mcpServers entry as above to ~/.claude.json.

    See Getting started.

    zotero-cli plugin     # where the .xpi is, and how to install it

    Then in Zotero: Tools > Plugins, the gear, Install Plugin From File. Details: Zotero Agent plugin.

    4. Writes (optional): on Zotero 10+, run zotero-mcp authorize-local once and choose Always Allow. On older Zotero, add ZOTERO_API_KEY and ZOTERO_LIBRARY_ID to write through the web API.

    Then ask things like "Find papers in my library on attention mechanisms", "Summarize the key findings of this paper", or "Highlight the main claims of this PDF".

    Optional extras

    The base install covers search, reading, annotations, and writes. Heavier features are extras:

    ExtraWhat it addsInstall command
    semanticSemantic search via ChromaDB, sentence-transformers, OpenAI/Gemini embeddingspip install "zotero-mcp-server[semantic]"
    pdfPDF outlines, page layout and page images (PyMuPDF), EPUB annotationspip install "zotero-mcp-server[pdf]"
    sciteScite citation tallies and retraction alerts (no account needed)pip install "zotero-mcp-server[scite]"
    allEverything abovepip install "zotero-mcp-server[all]"

    Update any time with zotero-mcp update.

    MCP server or agent skill?

    If your agent has a shell (Claude Code, Cursor, Codex, Windsurf, Gemini CLI, Amp, OpenCode …), one command teaches it to drive zotero-cli:

    zotero-mcp install-skill

    An MCP server sends every tool's schema on every request, before you type anything. The skill costs 98 tokens until the agent decides it is relevant:

    RouteIn contextPaid
    MCP server, default profile (41 tools)16,815every request
    MCP server, all tools (53)20,942every request
    Agent skill, frontmatter only98always
    Agent skill, body loaded2,400when it fires
    Agent skill + full command reference6,158worst case

    Use the MCP server when your client speaks MCP but has no shell (Claude Desktop, ChatGPT); use the skill when it has a shell. Both share one config. Details: CLI and agent skill.

    Zotero Agent: chat inside Zotero

    A plugin that adds a chat panel to Zotero itself, driven by your own Claude Code, Codex or pi. It knows which item and page you have open and works your library through zotero-cli. Answers cite papers with real zotero:// links, so a click opens the PDF at that page. It ships in the wheel: run zotero-cli plugin for the file and the install steps, or paste the one-paragraph prompt in the docs to your agent and let it do the setup. Details: Zotero Agent plugin.

    Documentation

    GuideWhat's in it
    Getting startedConnecting Claude Desktop and Claude Code, ChatGPT, Cherry Studio, Chorus, Autohand, and other MCP clients
    ConfigurationEnvironment variables, local writes, web and hybrid modes, the SQLite read backend, global search, text extraction, command-line options
    Semantic searchEmbedding models, building and updating the index
    ToolsEvery MCP tool, tool groups (ZOTERO_MCP_TOOLSETS), related items, PDF annotation extraction
    CLI and agent skillzotero-cli command reference, --json output, install-skill
    Zotero Agent pluginThe chat panel inside Zotero: install, first run, agents and sign-in, privacy
    DockerContainer images and runtime modes
    TroubleshootingCommon problems and fixes
    For AI agentsOne guide for an agent setting up or using Zotero MCP

    Website: stevenyuyy.com/zotero-mcp · Changelog

    Contributing

    Issues and pull requests are welcome. Run the tests with uv run pytest tests/. A live integration test plan, meant to be run by Claude against a real library, is in docs/integration-test-plan.md. Changes to the plugin in plugin/ (AGPL-3.0) are accepted on the condition that the maintainer may also release them under other terms, including commercial ones.

    Thanks to everyone who has contributed code, fixes, and ideas to Zotero MCP.

    Support

    Zotero MCP is free and open source.

    If it saves you or your lab time, sponsoring helps cover the unglamorous parts: Windows and WSL2 edge cases, Zotero schema changes, group-library support, and the embedding/search infrastructure.

    Labs and institutions: the $50 and $200 tiers are meant to be expensable, and include priority triage on the issues affecting your workflow.

    License

    The MCP server and zotero-cli are MIT (LICENSE). The Zotero Agent plugin in plugin/ is AGPL-3.0-or-later (plugin/LICENSE), the license Zotero itself uses.

    相关项目

    同分类里 star 最高的项目
    对比全部 →
    n8n
    n8n-io·TypeScript

    可视化搭建工作流并支持自定义代码的自动化平台，内置 AI 能力，提供 400+ 集成，可自托管或上云，用来把各类 API 和数据流串起来。

    工作流自动化+4MCP 生态
    210058+32661k[email protected]2026-11-11 04:05 UTC
    Scrapling
    D4Vinci·Python

    Python 网页抓取框架，从单次请求到大规模爬取都能处理，内置 Playwright 支持，还提供 MCP 服务方便接入 AI 工作流。

    网页抓取+4MCP 生态进阶
    101258+228211kv0.5.12026-11-09 10:36 UTC
    awesome-mcp-servers
    punkpeye

    一份精选的 MCP 服务器清单，帮你在接入 Model Context Protocol 生态时快速找到可用的服务端实现。

    MCP 服务器+3MCP 生态入门
    97585+18220k2026-11-06 17:59 UTC
    servers
    modelcontextprotocol·TypeScript

    Model Context Protocol 官方参考服务器集合，提供文件系统、Git、数据库等现成 MCP 服务实现，让 AI Agent 快速接入外部工具和数据源。

    MCP 服务器+4MCP 生态进阶
    91917+7412kv2.3.02026-11-11 05:16 UTC