1. Open Google Cloud Console;
2. Click current project name to access project selector (top left button);
3. Select "New Project" and provide descriptive name (e.g., "sheets mcp agent");
4. Pick your organisation. If you're using a personal Gmail, this field can have the value "No Organisation";
5. Click Create;
6. Wait for project creation and switch to new project.

1. From the left menu, go to APIs & Services → Library;
2. In the search field, type Google Sheets API, select the corresponding API, and click the Enable button;
3. Then, go to Library in left menu, search for Google Drive API, select the corresponding API and enable it.

1. Navigate to the "Credentials" tab;
2. If prompted, click "Configure consent screen";
3. Click "Get started";
4. Enter application name (e.g., "gsheet agent");
5. Input your email address;
6. Select "External Audience" for user type, if you use general Gmail account;
7. Re-enter your email address when prompted;
8. Accept the user data policy;
9. Click Continue and Create.

1. Click "Create OAuth Client";
2. Select "Desktop App" for local AI agents and scripts;
3. Accept suggested naming or customize as needed;
4. Download the JSON file and save it as credentials.json in your project directory;
5. Now you need to move the file with credentials to the folder with the AI agent (/google_sheet_mcp). You can do that manually by moving the file icon to the opened folder in the file explorer;
6. Use Claude with the Desktop MCP agent to move file:

Move the credentials.json file from Desktop into the new
Documents/projects/google_sheet_mcp folder

Confirm successful file transfer and remove from original location.

In Claude write:

Update config file with google_sheets_mcp server uvx mcp-google-sheet@latest.
Add environment variables CREDENTIALS_PATH and TOKEN_PATH
(same as CREDENTIALS_PATH, but pointing to token.json)

• Fully exit Claude Desktop. (Mac: Cmd + Q; Windows: Alt + F4);
• Reopen Claude to finish the MCP launch.

• Reopen Claude;
• Now you can test the new Google Sheets AI agent. Here is the CSV file used in the video, along with a prompt example:

Create a new Google Sheet, build a pivot table that groups the data
by the product column, summarise the total quantity and the total
revenue for each product. Use the most recent
/Downloads/sales_dataset_large.csv file.
Use Python libraries to analyse this file.

• Claude will analyze the file and generate a new Google Sheet in your Drive.