Overview
What is Google Sheets Mcp?
Google Sheets Mcp is a Model Context Protocol (MCP) connector that enables AI agents to read, write, and manage Google Sheets spreadsheets directly through natural language or tool calls.
How to use Google Sheets Mcp?
Clone the repository, run npm install and npm run build, then place your Google OAuth credentials file (gcp-oauth.keys.json) in the dist directory. Start the server with npm run start and configure your MCP client with the sample JSON config pointing to the built index.js file. Specify a spreadsheetId or ask your agent to create a new spreadsheet.
Key features of Google Sheets Mcp
- Authenticate using Google OAuth 2.0 with desktop credentials
- List, create, and rename sheets and spreadsheets
- Read all data, specific rows, or columns from a sheet
- Edit individual cells, entire rows, or entire columns
- Insert new rows or columns at specified positions
- Refresh authentication when credentials expire
Use cases of Google Sheets Mcp
- An AI assistant automatically populates a weekly report spreadsheet
- A chatbot reads and updates a shared budget tracker
- A productivity agent creates new project sheets and manages task lists
- An automation pipeline inserts data rows into a live spreadsheet
FAQ from Google Sheets Mcp
What does Google Sheets Mcp do?
It acts as a bridge between AI agents and Google Sheets, allowing agents to perform spreadsheet operations via the Model Context Protocol.
How do I authenticate with Google Sheets?
You must create OAuth 2.0 credentials in Google Cloud Platform, enable the Sheets API, and save the downloaded gcp-oauth.keys.json file in the dist directory. The server will prompt you to authenticate on first run.
What operations are available?
It supports reading/writing cells, rows, and columns; creating and renaming sheets and spreadsheets; inserting rows/columns; and listing sheet names. A full table of 14 actions is provided in the README.
What runtime dependencies does it require?
Node.js and npm are needed to install dependencies and build the project. The server runs locally and communicates over stdio via the MCP protocol.
