cursor-notebook-mcp

概述

cursor-notebook-mcp是一个基于Python的模型上下文协议(MCP)服务器，旨在帮助AI代理与Jupyter笔记本文件交互。它提供全面的工具用于创建、读取、编辑和导出笔记本文件，解决了Cursor在代理模式下无法直接编辑笔记本的限制。该服务器使用nbformat进行安全操作，使用nbconvert提供导出功能，并采取措施将操作限制在用户定义的目录中。

试试问 AI

装完之后，这里有 6 个你可以让 AI 做的事：

• notebook_create

  Creates a new, empty notebook file.

  试试:Create a new empty notebook试试:Start a new data analysis notebook
• notebook_read

  Reads an entire notebook and returns its structure as a dictionary.

  试试:Read the contents of my notebook.ipynb试试:Show me all cells in the analysis notebook
• notebook_add_cell

  Adds a new code or markdown cell after a specified index.

  试试:Add a new code cell after cell 3试试:Insert a markdown cell explaining the results
• notebook_edit_cell

  Replaces the source content of a specific cell.

  试试:Update cell 5 with new visualization code试试:Fix the error in cell 2
• notebook_delete_cell

  Deletes a specific cell.

  试试:Remove cell 4 which has no output试试:Delete the markdown introduction cell
• notebook_export

  Exports the notebook to another format (e.g., python, html).

  试试:Export the notebook to Python script试试:Convert my analysis to HTML format
• notebook_split_cell

  Splits a cell into two at a specified line number.

  试试:Split cell 3 after line 5试试:Divide the long cell into two parts
• notebook_merge_cells

  Merges a cell with the cell immediately following it.

  试试:Combine cell 4 and cell 5试试:Merge adjacent analysis cells
• notebook_search

  Searches cells for a keyword, showing contextual snippets.

  试试:Search for 'matplotlib' in the notebook试试:Find cells containing 'pandas'
• notebook_get_outline

  Produces an outline showing cell numbers with major headings/functions.

  试试:Show me the notebook structure试试:Generate an outline of the analysis notebook
• notebook_clear_all_outputs

  Clears outputs and execution counts for all code cells.

  试试:Reset all cell outputs试试:Clear the notebook before running fresh
• notebook_rename

  Renames/moves a notebook file from one path to another.

  试试:Rename my_notebook.ipynb to final_analysis.ipynb试试:Move the notebook to the reports folder

可对比工具

安装

pip install cursor-notebook-mcp

git clone https://github.com/jbeno/cursor-notebook-mcp.git
cd cursor-notebook-mcp
python -m venv .venv
source .venv/bin/activate  # Windows上使用 `.venv\Scripts\activate`
pip install -e .

{
  "mcpServers": {
    "notebook_mcp": {
      "url": "http://127.0.0.1:8080/mcp"
    }
  }
}

cursor-notebook-mcp --transport streamable-http --allow-root /path/to/your/notebooks --host 127.0.0.1 --port 8080

FAQ

这个服务器支持哪些传输模式？

该服务器支持stdio、流式HTTP和SSE传输模式。建议新设置使用流式HTTP。

除了Cursor，这个服务器可以与其他AI工具一起使用吗？

是的，虽然它是为Cursor设计的，但该服务器可以配置为与VS Code（Insiders）、Claude Code或任何支持MCP的模型/代理一起使用。

导出功能需要哪些外部依赖？

Pandoc用于大多数非HTML导出格式。对于PDF导出，nbconvert还需要LaTeX（推荐XeLaTeX）。

cursor-notebook-mcp 对比