haris-musa/excel-mcp-server

What it solves

It allows AI clients (such as Claude, Cursor, and VS Code Copilot) to programmatically create, read, and edit Excel files without requiring Microsoft Excel to be installed on the system. It bridges the gap between LLMs and complex spreadsheet operations, enabling AI to handle advanced tasks like generating charts, PivotTables, and complex formulas.

How it works

Built on the openpyxl library, the project implements the Model Context Protocol (MCP), allowing it to act as a server that AI clients connect to via stdio or HTTP. It includes a built-in calculation engine that can evaluate approximately 260 Excel functions, meaning it can return the results of formulas even if the file has never been opened in Excel. It supports a wide range of tools for manipulating workbooks, sheets, cells, formatting, and objects, and includes security features like folder confinement and formula safety checks to prevent malicious execution.

Who it’s for

Developers and AI users who want to give their AI assistants the ability to perform data analysis, report generation, and spreadsheet management directly within Excel files (.xlsx, .xlsm, .xltx, .xltm).

Highlights

• Advanced Excel Objects: Create and manage charts (including Excel 2016 types), real PivotTables with slicers and timelines, and sparklines.
• Formula Engine: Built-in engine calculates formulas and supports dynamic arrays (e.g., FILTER, SORT, UNIQUE).
• Comprehensive Toolset: Tools for everything from basic cell reading/writing to conditional formatting, data validation, and sheet protection.
• Large File Support: Uses streaming to keep memory usage flat even when processing sheets with hundreds of thousands of rows.
• VBA Support: Can read VBA code and optionally write it (opt-in), though it never executes macros for security reasons.