MCP Jupyter Server

Instructions

Implementation Reference

• src/jupyter-handler.js:95-120 (handler)

  The handler function that reads a Jupyter notebook file, extracts information about each cell (index, type, and preview of source), and returns a formatted text list of all cells.

  async listCells(notebookPath) {
    const notebook = await this.readNotebook(notebookPath);
    const cellsInfo = notebook.cells.map((cell, index) => {
      const source = Array.isArray(cell.source) ? cell.source.join('') : cell.source;
      const preview = source.length > 100 ? source.substring(0, 100) + '...' : source;
      
      return {
        index,
        type: cell.cell_type,
        preview: preview.replace(/\\n/g, ' ')
      };
    });
  
    return {
      content: [
        {
          type: "text",
          text: `Notebook: ${notebookPath}\nTotal cells: ${cellsInfo.length}\n\n${
            cellsInfo.map(cell => 
              `[${cell.index}] ${cell.type}: ${cell.preview}`
            ).join('\n')
          }`
        }
      ]
    };
  }

• src/index.js:39-52 (schema)

  The input schema definition for the 'list_cells' tool, specifying the required 'notebook_path' parameter.

  {
    name: "list_cells",
    description: "List all cells in a Jupyter notebook with their indices and types",
    inputSchema: {
      type: "object",
      properties: {
        notebook_path: {
          type: "string",
          description: "Absolute path to the Jupyter notebook file"
        }
      },
      required: ["notebook_path"]
    }
  },

• src/index.js:331-333 (registration)

  The dispatch case in the CallToolRequestSchema handler that routes 'list_cells' calls to the JupyterHandler's listCells method.

  case "list_cells":
    return await this.jupyterHandler.listCells(args.notebook_path);