excel-mcp-server MCP — Use Cases, Install & Live Demo

Why use it

Key features

Create/read/update workbooks and worksheets
Formulas, tables, charts (line/bar/pie/scatter), pivot tables
Formatting: fonts, colors, borders, conditional formatting
Transport modes: stdio (local) and streamable-http (remote)
No Excel/Office install required — pure Python

Live Demo

What it looks like in practice

excel.replay ▶ ready

0/0

Install

Pick your client

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "excel": {
      "command": "uvx",
      "args": [
        "excel-mcp-server"
      ],
      "_inferred": true
    }
  }
}

Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "excel": {
      "command": "uvx",
      "args": [
        "excel-mcp-server"
      ],
      "_inferred": true
    }
  }
}

Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "excel": {
      "command": "uvx",
      "args": [
        "excel-mcp-server"
      ],
      "_inferred": true
    }
  }
}

Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "excel": {
      "command": "uvx",
      "args": [
        "excel-mcp-server"
      ],
      "_inferred": true
    }
  }
}

Same shape as Claude Desktop. Restart Windsurf to pick up changes.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "excel",
      "command": "uvx",
      "args": [
        "excel-mcp-server"
      ]
    }
  ]
}

Continue uses an array of server objects rather than a map.

~/.config/zed/settings.json

{
  "context_servers": {
    "excel": {
      "command": {
        "path": "uvx",
        "args": [
          "excel-mcp-server"
        ]
      }
    }
  }
}

Add to context_servers. Zed hot-reloads on save.

claude mcp add excel -- uvx excel-mcp-server

One-liner. Verify with claude mcp list. Remove with claude mcp remove.

Use Cases

Real-world ways to use excel-mcp-server

How to generate a monthly Excel report from a prompt

👤 Ops, finance, anyone sending Excel to non-technical stakeholders ⏱ ~20 min intermediate

When to use: Every month-end when you refresh 'the spreadsheet everyone asks for'.

Prerequisites

uv installed — curl -LsSf https://astral.sh/uv/install.sh | sh
A source of truth (CSV, DB query, or prior month's file) — Export to CSV, or pair with postgres MCP

Flow

Create the workbook

Create /reports/2026-04.xlsx with tabs 'Summary', 'Revenue', 'Costs', 'Charts'.✓ Copied

→ File created, sheets exist
Populate data

Fill 'Revenue' from /data/apr-revenue.csv; add a total row with SUM formulas.✓ Copied

→ Rows + totals present
Add charts

On 'Charts' tab add a line chart of daily revenue from 'Revenue' A2:B31 titled 'April Revenue'.✓ Copied

→ Chart visible on open in Excel
Format

On 'Summary', bold headers, apply thousands separators to currency, and conditional-format any red (negative) values.✓ Copied

→ Professionally formatted file

Outcome: A reviewer-ready .xlsx with live formulas and charts.

Pitfalls

Pasting 100k rows via prompt blows the context window — Have the MCP read the CSV directly from disk rather than pasting values
Pivot tables written by openpyxl don't auto-refresh on open in some Excel versions — Instruct users to right-click Refresh; or use a formula-based summary instead

Combine with: postgres · filesystem

How to extract tables from an Excel deck into markdown

👤 Engineers turning finance spreadsheets into docs ⏱ ~10 min beginner

When to use: A stakeholder sends you a workbook and you need it in a wiki.

Flow

Enumerate sheets

Open /inbox/Q1-plan.xlsx and list sheets + dimensions.✓ Copied

→ Sheet index
Extract each as markdown

For each sheet, read the used range and render as a markdown table.✓ Copied

→ Markdown blob ready to paste

Outcome: Wiki-ready tables.

Pitfalls

Merged cells mangle markdown rendering — Unmerge before export, or handle merged regions explicitly

Combine with: markdownify · notion

How to clean a messy Excel file in one pass

👤 Data wranglers ⏱ ~15 min intermediate

When to use: A partner sends a sheet with inconsistent dates, trailing whitespace, and duplicated headers.

Flow

Profile the problems

Audit /inbox/vendors.xlsx — columns with mixed types, whitespace, duplicate rows.✓ Copied

→ Issue report
Apply fixes

Trim whitespace, parse column E as dates in ISO, remove full-duplicate rows. Save to /cleaned/vendors.xlsx.✓ Copied

→ Clean file saved

Outcome: A clean workbook ready for ingestion.

Pitfalls

'Fixing' dates that were intentionally text loses meaning — Confirm with the owner before aggressive coercions

Combine with: postgres

Combinations

Pair with other MCPs for X10 leverage

excel + postgres

Query the DB and write the result to Excel for stakeholders

Run SELECT * FROM monthly_revenue WHERE month = '2026-04' and write the result to /reports/apr.xlsx with a chart.✓ Copied

excel + filesystem

Batch-process every xlsx in a folder

For every file in /inbox/vendors/, unify the headers to {vendor,sku,qty,price} and write to /cleaned/.✓ Copied

excel + markdownify

Convert xlsx tables to markdown docs

Extract each sheet of /inbox/plan.xlsx to markdown sections and write /docs/plan.md.✓ Copied

Tools

What this MCP exposes

Tool	Inputs	When to call	Cost
create_workbook	path: str	Start a new .xlsx	free
create_worksheet	path, name	Add a tab	free
write_data_to_excel	path, sheet, start_cell, data: 2d array	Populate a range	free
read_data_from_excel	path, sheet, range	Extract values	free
apply_formula	path, sheet, cell, formula	Add a formula	free
create_chart	path, sheet, chart_type, range, title	Insert a chart	free
create_pivot_table	path, source, rows, cols, values	Summarize data	free
format_range	path, sheet, range, styles	Styling	free
apply_conditional_formatting	path, sheet, range, rule	Color-code cells	free

Cost & Limits

What this costs to run

API quota: None — all local
Tokens per call: Depends on how much data flows through the prompt; prefer file-path operations
Monetary: Free (MIT)
Tip: Read ranges, not whole sheets, when a file has 100k+ rows.

Security

Permissions, secrets, blast radius

Minimum scopes: Filesystem read/write to the folder containing the workbooks

Credential storage: None

Data egress: Local file I/O only; data only leaves your machine via your LLM provider

Never grant: Write access outside a sandboxed folder when running untrusted prompts

openpyxl doesn't run macros or VBA — .xlsm macro logic is ignored.
Very large files (>50 MB) can OOM on small machines.

Troubleshooting

Common errors and fixes

Permission denied writing file

Another program has the file open (often Excel itself). Close it and retry.

Verify: lsof path/to/file.xlsx

Formula stored as text

Prefix strings starting with '=' via apply_formula, not write_data_to_excel.

Chart doesn't render in Excel

Check the data range covers the source columns with headers; Excel needs 2+ rows.

File saved but Excel says 'corrupted'

Usually mixing write_data and formulas in the same cell. Recreate the file, build bottom-up.

Alternatives

excel-mcp-server vs others

Alternative	When to use it instead	Tradeoff
Google Sheets via google-workspace MCP	You want collaborative cloud sheets, not local .xlsx	Needs OAuth; not a replacement for offline xlsx
LibreOffice headless MCP	You need macro execution or conversion to PDF	Heavy install; slower

More

Resources

📖 Read the official README on GitHub

🐙 Browse open issues

🔍 Browse all 400+ MCP servers and Skills