Let an agent read your spreadsheets

@sezzlee/excel-mcp gives an agent read-only access to the .xlsx, .xlsm and .csv files in one folder, and to nothing outside it. It has no tool that writes, so a workbook is never modified.

You need Node.js 22 or later.

1. Pick a folder

Everything under the folder is readable, subfolders included, so pick the narrowest one that holds the files the agent needs. To follow along, create one and save sales.xlsx into it: twelve orders on a sheet called Orders.

mkdir -p ~/sezzlee-sheets

2. Add the server to your client

Pass the folder as an absolute path. A JSON configuration is not read by a shell, so ~ is not expanded there.

claude mcp add excel -- npx -y @sezzlee/excel-mcp /Users/you/sezzlee-sheets

Claude Desktop reads claude_desktop_config.json (Settings → Developer → Edit Config, then restart). Cursor reads .cursor/mcp.json in the project or ~/.cursor/mcp.json. VS Code reads .vscode/mcp.json. Any other client that starts a stdio server works with the command npx and the arguments -y, @sezzlee/excel-mcp and the folder.

3. Ask

Ask the agent: Which region sold the most in sales.xlsx?

Expect three calls: list_workbooks to find the file, describe_workbook to see its sheets, and aggregate_sheet to add up Total for each Region on the server. The agent gets four rows back instead of twelve, and the answer is West, at 1,865. On a sheet of fifty thousand rows the answer is the same size.

See what the agent receives

The recipes call the tools the way an agent does, through the MCP Inspector, so you can see each answer exactly. They need the server installed once and jq:

npm install -g @sezzlee/excel-mcp

Define this helper in your shell; it reads ~/sezzlee-sheets:

excel() {
  npx -y @modelcontextprotocol/inspector --cli sezzlee-excel ~/sezzlee-sheets \
    --method tools/call --tool-name "$@" | jq '.content[0].text | fromjson'
}

This is the call behind the answer above:

excel aggregate_sheet --tool-arg filePath=sales.xlsx \
  'groupBy=["Region"]' 'metrics=[{"fn":"sum","column":"Total"}]' \
  | jq -c '[.columns[].label], .rows[]'

["Region","sum(Total)"]
["East",1290]
["North",1660]
["South",1810]
["West",1865]

What it can read

• Cells as a compact grid: the header row once, then one array per row, as values, formulas or both.
• Answers instead of pages: totals, counts and groupings computed on the server, and cells located by value or formula.
• Structure: sheets, used ranges, defined names, merged cells, Excel Tables, data validation, conditional formatting and embedded pictures.
• CSV files, with the delimiter and encoding detected and every value kept as the exact text.

Charts, pivot tables and sparklines cannot be read. The server says so instead of answering with an empty list.