sbroenne__mcp-server-excel/PRIVACY.md

Privacy Policy

Last Updated: August 28, 2026

Overview

MCP Server for Excel ("ExcelMcp") is an open-source tool that enables AI assistants to interact with Microsoft Excel. This privacy policy explains how the software handles your data.

Data Collection Summary

Telemetry applies to the MCP Server only. The CLI (excelcli) and its background daemon send no telemetry of any kind — the code paths described below exist solely in ExcelMcp.McpServer.

ExcelMcp's MCP Server collects limited, anonymous telemetry to improve the software. The statements below describe telemetry collection, not the workbook data that a tool returns to your chosen AI assistant.

What We DO Collect (Anonymous Telemetry)

• Tool usage statistics - Which tools and actions are used (e.g., "range/get-values")
• Performance metrics - How long operations take (duration in milliseconds)
• Invocation outcome - Whether an operation succeeded, returned an expected negative diagnostic result, or failed
• Failure class - A fixed privacy-safe label for input or state, an external dependency, timeout or cancellation, Excel runtime, an internal product fault, or an unclassified failure
• Session information - A random session ID generated each time the server starts
• Anonymous user ID - A hashed identifier based on machine identity (not personally identifiable)
• Application version - Which version of ExcelMcp is running
• Session alias compatibility - Whether a session-bound MCP request used the fixed sessionId compatibility alias. Its custom properties contain only the tool, declared action, alias name, and application version. Like other MCP telemetry, the standard context also includes the anonymous user ID and the random MCP server process telemetry session ID described above.
• Unhandled exceptions - Error type, approved source, and project-owned failure site only (never exception messages or stack traces)

What We DO NOT Collect

• ❌ File contents - Workbook content is never included in telemetry
• ❌ File names or paths - File paths are not included in telemetry
• ❌ Personal information - No names, emails, or account information
• ❌ Spreadsheet data - Cell values, formulas, and query results are never included in telemetry
• ❌ Prompts or messages - Requests and conversations are never transmitted
• ❌ Host or client details - Working directories and MCP client names are not collected
• ❌ User accounts - No registration or sign-in required
• ❌ Error details - Error messages, response content, exception names, and stack traces are not included in invocation outcome telemetry
• ❌ Workbook session IDs or tool arguments - Compatibility telemetry never includes the Excel workbook session_id/sessionId value, workbook path, request arguments, or user content. The random MCP server process telemetry session ID listed above is separate from workbook identity.

Purpose of Telemetry

We use anonymous telemetry to:

• Understand which features are most used
• Identify and fix performance issues
• Prioritize development of new features
• Detect and fix bugs

Invocation telemetry reads only the structured category produced by ExcelMcp; it does not inspect response or error text. Unknown or missing categories remain unclassified.

Telemetry Infrastructure

Telemetry is sent to Azure Application Insights, a Microsoft service. Data is:

• Transmitted over HTTPS
• Stored in accordance with Microsoft's data handling policies
• Retained for analytics purposes only
• Filtered at ingestion so exception records without ExcelMcp's explicit sanitization marker are discarded
• Filtered at ingestion so framework trace logs are discarded in full

How It Works

ExcelMcp drives the Excel application on your local machine:

1. Local Processing - All Excel operations are performed locally via Microsoft's COM API
2. Your Files Stay Local - Excel files are read from and written to your local filesystem only
3. Tool Results - Data requested by your AI assistant is returned through your MCP client or CLI process; your assistant's privacy policy governs how it handles that data
4. Optional Network Features - Remote M/DAX formatting and Python in Excel use external services only when you request those features
5. Telemetry - Release builds of the MCP Server can send the anonymous usage metrics listed above to Azure Application Insights; the CLI sends no telemetry

Data Flow

When you use ExcelMcp with an AI assistant (like Claude):

1. You send a request to the AI assistant
2. The AI assistant calls ExcelMcp tools on your local machine
3. ExcelMcp performs the requested Excel operations locally
4. Requested results are returned to the AI assistant through your MCP client
5. The MCP Server can send anonymous usage telemetry to Azure Application Insights

Some operations have additional data flows:

• Setting formatMCode=true sends the supplied M code to powerqueryformatter.com.
• Setting formatDax=true sends the supplied DAX formula to DAX Formatter.
• Python in Excel sends Python code and referenced worksheet data to Microsoft's cloud execution environment. This feature requires Microsoft 365, internet access, and Python in Excel to be enabled.

Remote formatting is disabled by default and requires explicit consent. ExcelMcp preserves M and DAX code locally unless you opt in.

Note: The AI assistant you use (for example, Claude or GitHub Copilot) has its own privacy policy governing conversations, tool calls, and returned workbook data.

Third-Party Services

• Azure Application Insights - Anonymous telemetry is sent to this Microsoft service. See Microsoft's Privacy Statement.
• Microsoft Excel - ExcelMcp requires Microsoft Excel installed on your machine. Excel is subject to Microsoft's privacy policy.
• Microsoft Python in Excel - Python code and referenced worksheet data run in Microsoft's cloud when you use the Python in Excel operations.
• Power Query Formatter - Receives M code only when you set formatMCode=true.
• DAX Formatter - Receives DAX formulas only when you set formatDax=true.
• AI Assistants - When used with AI assistants like Claude, those services have their own privacy policies.

Open Source

ExcelMcp is open source software. You can review the complete source code at: https://github.com/sbroenne/mcp-server-excel

Security

• ExcelMcp runs with the same permissions as your user account
• It can only access files and Excel instances that your user account can access
• No elevated privileges are required or requested

Children's Privacy

ExcelMcp does not knowingly collect any information from anyone, including children under 13 years of age.

Changes to This Policy

If we make changes to this privacy policy, we will update the "Last Updated" date above and publish the updated policy in our GitHub repository.

Contact

For questions about this privacy policy or the ExcelMcp project:

• GitHub Issues: https://github.com/sbroenne/mcp-server-excel/issues
• Repository: https://github.com/sbroenne/mcp-server-excel

---

Summary: ExcelMcp drives local Excel and reads or writes workbook files on your machine. Requested tool results can be returned to your chosen AI assistant. Remote code formatting and Python in Excel use external services only when requested. The MCP Server can send anonymous usage telemetry, but that telemetry excludes workbook contents, file names, paths, and personal information. The CLI sends no telemetry.