excel-mcp-rw
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@excel-mcp-rwReconcile PurchaseOrders.xlsx with SAP_Export.xlsx using PO_Number and SAP_Code keys"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Excel MCP Server (excel-mcp-rw)
High-Performance Read & Write Model Context Protocol (MCP) Server for Microsoft Excel on Windows.
Powered by Pandas, OpenPyXL, and Native Windows Microsoft Excel Automation (COM / PyWin32).
⚡ Why This Server?
Most Excel MCP servers are lightweight JavaScript wrappers that fail on real-world spreadsheets:
❌ They crash or exceed LLM context windows on large multi-megabyte files.
❌ They cannot calculate dynamic Excel formulas (
=XLOOKUP,=FILTER(), dynamic arrays, complex nesting).❌ They cannot refresh PivotTables or external database connections.
❌ They cannot run VBA macros (
.xlsm) or export pixel-perfect PDFs.
excel-mcp-rw solves this with a Dual-Engine Architecture:
Engine 1: Fast Python Data Science Stack (
pandas+openpyxl)Instant metadata streaming without loading multi-GB spreadsheets into RAM.
Vectorized filtering (
df.query()) and column statistics at compiled C-speed.Bounded pagination so the model never blows its context window.
Engine 2: Native Microsoft Excel Engine (COM Automation via
pywin32)Uses real Windows Microsoft Excel to execute
Application.CalculateFull().Refreshes PivotTables and data connections (
wb.RefreshAll()).Native print-engine PDF export.
Runs VBA macros and inspects active workbooks open on the desktop.
Related MCP server: BachStudio Excel MCP Server
🛠️ Available Tools
📊 Data Analysis & Manipulation (pandas / openpyxl)
Tool | Description |
| Read sheet names, dimensions, column header preview, and file size in MB in read-only streaming mode. |
| Preview first $N$ rows (default 10) with column data types and shape. |
| Filter rows with vectorized Pandas expressions (e.g. |
| Windowed slice read (e.g. rows 100 to 200) without loading the entire spreadsheet. |
| Comprehensive column statistics: min, max, mean, median, std, sum (numeric) or top 10 value frequencies (categorical). |
| Fast pattern and keyword search across all rows and columns with row number coordinates. |
| Create formatted |
| Add a brand new sheet tab to an existing workbook without touching other sheets (with optional data and headers). |
| Rename an existing sheet tab. |
| Delete a specific worksheet tab from an existing workbook. |
| In-place row append without rewriting the workbook. |
| Surgical updates to specific cells or formulas (e.g. |
| Export large sheets to clean CSV format. |
🔄 Multi-Key Reconciliation & Discrepancy Matching (pandas + RapidFuzz)
Tool | Description |
| Pre-flight Diagnostic: Evaluates key compatibility, nulls, duplicates, data type mismatches, exact match %, and fuzzy match potential across workbooks before any merge. |
| Reconciliation Engine: Matches on composite keys (2–3+ columns), calculates numerical/text variances on compared columns, applies fuzzy matching on unmatched keys, and outputs a clean 3-tab audit workbook ( |
🪟 Native Windows Microsoft Excel Tools (COM / PyWin32)
Tool | Description |
| Opens workbook in native Microsoft Excel, executes |
| Generates a pixel-perfect PDF using Microsoft Excel's internal print engine. |
| Refreshes all external data connections and PivotTables ( |
| Runs any VBA macro inside |
| Detects if Excel is open on your desktop and returns the active workbook, active sheet, and selected range. |
📋 Best-Practice Agent Workflow for Reconciliations
When asking your AI assistant to compare or reconcile two workbooks (e.g. Purchase Orders vs SAP/ERP records):
Step 1: Pre-Flight Analysis
The agent callsanalyze_reconciliation_keys(file_1, file_2, join_keys=['PO_Number', 'SAP_Code'])to diagnose key overlap, formatting quirks, and duplicate keys.Step 2: Diagnosis & Strategy Alignment
The agent reports match percentage and any recommendations (e.g. "5 rows have whitespace differences; 2 rows need fuzzy matching").Step 3: Execute & Generate 3-Tab Audit Workbook
The agent runsreconcile_and_mergeto produce:1_Reconciled_Matches: Exact matches + calculated price/qty differences.2_Probable_Fuzzy_Matches: Near matches flagged for human review with confidence %.3_Unmatched_Exceptions: Records missing in either file.
🚀 Quickstart & Installation
1. Clone & Install Dependencies
git clone https://github.com/indala/excel-mcp-rw.git
cd excel-mcp-rw
pip install -r requirements.txt⚙️ Client Configurations
1. Claude Desktop
Add this to your %APPDATA%\Claude\claude_desktop_config.json:
{
"mcpServers": {
"excel": {
"command": "python",
"args": [
"C:/path/to/excel-mcp-rw/src/excel_mcp/server.py"
]
}
}
}2. Claude Code CLI
Add globally to your user scope:
claude mcp add -s user excel -- python C:/path/to/excel-mcp-rw/src/excel_mcp/server.py3. Google Antigravity
Add to ~/.gemini/config/mcp_config.json:
{
"mcpServers": {
"excel": {
"command": "python",
"args": [
"C:/path/to/excel-mcp-rw/src/excel_mcp/server.py"
]
}
}
}4. Cursor / Windsurf
Add a new STDIO server in your MCP settings:
Name:
excelType:
command(orstdio)Command:
python C:/path/to/excel-mcp-rw/src/excel_mcp/server.py
📝 Example Prompts
Once connected to your assistant, you can ask:
"Give me an overview of all sheets in
C:/reports/financials_q3.xlsxwithout loading the whole file.""Filter the
Transactionssheet forAmount > 10000 and Status == 'Pending', and give me the top 20 rows.""What is the statistical distribution of the 'Salary' column in our employee database?"
"Search for customer ID 'CUST-8492' across all sheets and tell me which row it appears on."
"Recalculate all formulas in this sheet using native Excel and export it to PDF."
"What Excel spreadsheet do I currently have open on my desktop?"
📄 License
This project is licensed under the MIT License.
This server cannot be deployed
Maintenance
Related MCP Connectors
Excel analytics: inspect, query (JSON rows), charts, and JSON-to-xlsx workbook writing.
Generate PDF/DOCX/XLSX/PPTX from templates+JSON. Convert Office/HTML/MD to PDF. Universal templating
Formula-backed WorkPaper tools for workbook readback, input edits, and JSON persistence.
Composable APIs for document extraction, image transformation, and document & sheet generation.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceProvides Excel file manipulation capabilities without requiring Microsoft Excel installation, enabling workbook creation, data manipulation, formatting, and advanced Excel features.10MIT
- AlicenseNot gradedqualityNot gradedmaintenanceEnables creation, reading, and manipulation of Excel workbooks without Microsoft Excel installed, supporting formulas, formatting, charts, pivot tables, and data validation through multiple transport protocols.MIT
- AlicenseNot gradedqualityDmaintenanceAn MCP server for manipulating Excel files that features a headless engine for real-time formula calculation and data validation. It enables users to create, read, and manage workbooks, sheets, and charts without requiring Microsoft Excel installed.25 npmMIT
- AlicenseNot gradedqualityBmaintenanceEnables reading and writing of values, formulas, and formatting in Microsoft Excel files including .xlsx and .xlsm formats. It supports sheet management, table creation, and provides Windows-exclusive features like live editing and screen capture.8,110 npmMIT