EndNote Library Reader
Enables programmatic access to EndNote .enl libraries, allowing users to list, search, and extract full text from bibliographic references and their attached PDFs through MCP tools.
README
EndNote MCP Service
This project provides an automated service for reading EndNote .enl libraries and exposing their contents via the MCP (Modular Command Protocol) interface. It enables structured, programmatic access to bibliographic data and PDF full texts for downstream LLMs or automation clients.
Features
- Automatic EndNote Library Parsing: Reads
.enl(SQLite) files and associated.Datafolders. - MCP Tool Interface: Exposes all functions as MCP tools for easy integration with LLMs or workflow engines.
- List All References: Enumerate all bibliographic entries with metadata.
- Fuzzy Search: Search references by (partial) title or keywords, supporting both English and Chinese.
- PDF Full Text Extraction: Retrieve and extract the full text of attached PDF files by paper title.
- Backup Mode & Manual Refresh: Optionally operate on a
.enl.backupfile for safer access, with both automatic and manual refresh support.
Typical Usage
- List all papers:
list_papers() - Fuzzy search:
search_papers('distillation') - Extract full text:
read_paper('Paper Title') - Manually refresh backup:
refresh_backup()
Application Scenarios
- LLM-based literature review and knowledge extraction
- Automated research assistants
- Academic data mining and analysis
Requirements
- Python 3.10+
- EndNote
.enlfile and corresponding.Datafolder - MCP/fastmcp runtime environment
Installation
-
Install uv if you don't have it:
pip install uv # or see https://github.com/astral-sh/uv for other installation methods -
Install dependencies using uv (recommended):
uv pip install -r requirements.txtThe main dependencies are:
- fastmcp>=2.8.1
- pypdf>=5.6.0
A requirements.txt file is provided for compatibility, but uv is the recommended tool for dependency management.
Configuration
You need to provide the path to your EndNote .enl file and the corresponding .Data folder. Optionally, you can enable detailed logging.
You may also enable backup mode with the --use-backup or -b option. When enabled, all database operations are performed on a .enl.backup file, which is automatically refreshed from the original .enl file at startup. You can manually refresh the backup at any time using the refresh_backup() MCP tool.
Why Backup Mode?
EndNote software locks the
.enlfile while it is running, preventing other programs from accessing it. This means that when you are viewing or editing references in EndNote, this tool cannot operate on the original.enlfile. To solve this, backup mode was introduced: all operations are performed on a.enl.backupcopy. When you need to refresh the backup, you must first close EndNote to release the file lock; otherwise, the refresh will fail.
Running the Server
Start the MCP server with the required arguments:
python server.py --enl-file <path-to-your.enl> --data-folder <path-to-your.Data> [--enable-log] [--use-backup]
# or using short options:
python server.py -e <path-to-your.enl> -d <path-to-your.Data> -l -b
--enl-file,-e: Path to the EndNote.enlfile (required)--data-folder,-d: Path to the EndNote.Datafolder (required)--enable-log,-l: Enable detailed log output (optional)--use-backup,-b: Use.enl.backupfor all DB operations (optional, enables backup mode; the backup is auto-refreshed at startup)
When backup mode is enabled, all operations are performed on the .enl.backup file. The backup is automatically refreshed from the original .enl file at server startup, and you can manually refresh it at any time using the refresh_backup() tool.
MCP Server Configuration
To integrate this service with MCP-compatible clients or orchestration tools, you can define an entry in your mcp.json configuration. Example:
{
"mcpServers": {
"enl_reader_mcp": {
"command": "uv",
"args": [
"run",
"--directory",
"<path-to-your-clone>",
"-m",
"server",
"-e",
"E:/EndNoteLib/your_library.enl",
"-d",
"E:/EndNoteLib/your_library.Data",
"-l",
"-b"
]
}
}
}
- Replace
<path-to-your-clone>with the actual path where you cloned this repository. - Adjust the paths to your
.enlfile and.Data - Add
-bto enable backup mode if desired. When enabled, the service will operate on a.enl.backupfile and support manual refresh via therefresh_backuptool.
MCP Tools
list_papers(offset=0, limit=10): List all references with pagination.search_papers(query): Fuzzy search references by title or keywords.read_paper(title): Get metadata and PDF full text by (fuzzy) title.refresh_backup(): Manually refresh the.enl.backupfile (only available when backup mode is enabled; has no effect if backup mode is off).
Note: The refresh_backup tool is only effective when backup mode is enabled. Backup mode is recommended for scenarios requiring read-only or safer access to the EndNote library.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。