mcp-google-sheets MCP — 사용 사례, 설치 & 라이브 데모

왜 쓰나요

핵심 기능

19 tools spanning create, read, update, search, share, charts
Batch updates to avoid rate limits on large writes
Service account or OAuth — fits personal and automation setups
Works across multiple spreadsheets in one call

라이브 데모

실제 사용 모습

google-sheets.replay ▶ 준비됨

0/0

설치

클라이언트 선택

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

{
  "mcpServers": {
    "google-sheets": {
      "command": "uvx",
      "args": [
        "mcp-google-sheets"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop → Settings → Developer → Edit Config 열기. 저장 후 앱 재시작.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "google-sheets": {
      "command": "uvx",
      "args": [
        "mcp-google-sheets"
      ],
      "_inferred": true
    }
  }
}

Cursor는 Claude Desktop과 동일한 mcpServers 스키마 사용. 프로젝트 설정이 전역보다 우선.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "google-sheets": {
      "command": "uvx",
      "args": [
        "mcp-google-sheets"
      ],
      "_inferred": true
    }
  }
}

Cline 사이드바의 MCP Servers 아이콘 클릭 후 "Edit Configuration" 선택.

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "google-sheets": {
      "command": "uvx",
      "args": [
        "mcp-google-sheets"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop과 같은 형식. Windsurf 재시작 후 적용.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "google-sheets",
      "command": "uvx",
      "args": [
        "mcp-google-sheets"
      ]
    }
  ]
}

Continue는 맵이 아닌 서버 오브젝트 배열 사용.

~/.config/zed/settings.json

{
  "context_servers": {
    "google-sheets": {
      "command": {
        "path": "uvx",
        "args": [
          "mcp-google-sheets"
        ]
      }
    }
  }
}

context_servers에 추가. 저장 시 Zed가 핫 리로드.

claude mcp add google-sheets -- uvx mcp-google-sheets

한 줄 명령. claude mcp list로 확인, claude mcp remove로 제거.

사용 사례

실전 활용법: mcp-google-sheets

Append daily metrics to a Google Sheet from any data source

👤 Ops / growth engineers ⏱ ~20 min intermediate

언제 쓸까: You want a lightweight dashboard in Sheets fed by whatever the agent can reach (Sentry, Stripe, a DB).

사전 조건

Google service account with Sheets + Drive API enabled — console.cloud.google.com → create SA → download JSON → set SERVICE_ACCOUNT_PATH
Sheet shared with the SA email — Sheets UI → Share → add [email protected] as Editor

흐름

Identify the target sheet

Use mcp-google-sheets to find the spreadsheet titled 'Daily KPIs'. Return its ID.✓ 복사됨

→ Spreadsheet ID
Append today's row

Append a row to sheet 'summary' with columns [date, signups, mrr, errors]. Use today's values from [your source].✓ 복사됨

→ add_rows confirms row appended at the bottom
Update a chart if needed

Verify the trailing-30d chart picked up the new row; if not, refresh it.✓ 복사됨

→ Chart shows today's data

결과: A live Sheets dashboard updated without opening the file.

함정

Rate limit (429) after many single-cell updates — Use batch_update_cells — one call can modify hundreds of cells
Service account doesn't see the sheet — You must explicitly share the sheet with its email — SAs have no implicit access

함께 쓰기: sentry · postgres

Deduplicate entries across multiple Google Sheets

👤 Data ops, CRM admins ⏱ ~25 min intermediate

언제 쓸까: Team members maintain separate Sheets and you need one clean consolidated list.

흐름

Enumerate source sheets

List all spreadsheets in my 'Leads 2026' Drive folder with list_spreadsheets.✓ 복사됨

→ List of 5-20 sheets
Pull all rows into memory

Use get_multiple_sheet_data to pull the 'Leads' tab from each. Combine by email.✓ 복사됨

→ Merged dataset with dup count
Write consolidated sheet

Create a new spreadsheet 'Leads 2026 - Consolidated' with deduped rows sorted by latest-touch.✓ 복사됨

→ Clean destination sheet

결과: One canonical sheet; originals untouched for audit.

Generate a Sheets report with formulas, not frozen values

👤 Finance analysts ⏱ ~30 min advanced

언제 쓸까: You want the sheet to recalculate when source tabs update.

흐름

Create the shape

Create a sheet 'Q1 Report' with tabs [summary, by_region, by_product].✓ 복사됨

→ Sheet + tabs created
Inject formulas referencing source

In summary!B2, write =SUMIFS('source'!C:C, 'source'!A:A, ">=2026-01-01"). Use batch_update with formula values.✓ 복사됨

→ Live-computing cells

결과: A self-updating report — change source data, numbers change.

Import a CSV and enrich each row with AI-generated columns

👤 Marketers, sales ops ⏱ ~30 min intermediate

언제 쓸까: You have a list of URLs/companies and want AI-generated descriptions or classifications per row.

흐름

Import CSV

Create a spreadsheet from this CSV: [paste]. Name it 'Leads Enriched'.✓ 복사됨

→ Sheet with N rows
Add and fill derived columns

Add columns 'industry' and 'one_line_pitch'. For each row, derive them from the url/company and write back via batch_update_cells.✓ 복사됨

→ Two new populated columns

결과: Enriched list ready for outreach.

함께 쓰기: firecrawl

조합

다른 MCP와 조합해 10배 효율

google-sheets + sentry

Daily crash-free % logged to a KPI sheet

Every day, pull crash-free session % from Sentry and append a row to my 'Daily KPIs' sheet.✓ 복사됨

google-sheets + firecrawl

Scrape competitor pages weekly, write pricing into a sheet

Run firecrawl_extract on these 5 competitor pricing pages and write each plan as a row in 'Competitor Pricing'.✓ 복사됨

google-sheets + postgres

Export a daily DB query to Sheets for non-tech teammates

SELECT * FROM signups WHERE day = CURRENT_DATE; write to tab 'Signups' in my ops sheet.✓ 복사됨

도구

이 MCP가 노출하는 것

도구	입력	언제 호출	비용
list_spreadsheets	folder_id?	Discover sheets in a folder	1 Drive API call
create_spreadsheet	title: str, sheets?: str[]	Fresh document	1 API call
get_sheet_data	spreadsheetId, sheet, range?	Read a range	1 API call
update_cells	spreadsheetId, range, values	Small writes	1 API call
batch_update_cells	spreadsheetId, updates	Any write >10 cells — avoids rate limits	1 API call
add_rows	spreadsheetId, sheet, rows	Append-only like logging	1 API call
add_chart	spreadsheetId, sheet, chart_spec	Visualize a range	1 API call
share_spreadsheet	spreadsheetId, email, role	Grant collab access	1 Drive call
search_spreadsheets	query: str	Find a value across many files	N API calls

비용 및 제한

운영 비용

API 쿼터: Sheets API: 300 read + 300 write req/min/project, 60/user. Drive API: 1k req/100s/user
호출당 토큰: 100–5000 depending on range size
금액: Free within Google's default quotas
팁: Batch writes aggressively — batch_update_cells is one API call no matter how many cells

보안

권한, 시크릿, 파급범위

최소 스코프: https://www.googleapis.com/auth/spreadsheets https://www.googleapis.com/auth/drive.file

자격 증명 저장: Service account JSON at SERVICE_ACCOUNT_PATH, or OAuth tokens at TOKEN_PATH

데이터 외부 송신: sheets.googleapis.com, drive.googleapis.com

절대 부여 금지: https://www.googleapis.com/auth/drive (full)

문제 해결

자주 발생하는 오류와 해결

403 The caller does not have permission

Share the sheet with the service account's email (client_email in the JSON)

확인: Open sheet → Share → confirm SA is listed

429 Quota exceeded

Use batch_update instead of many single updates; Sheets API limit is per-minute

Invalid range A1 notation

Range needs the sheet name: 'Sheet1!A1:C10', not 'A1:C10'

gcloud auth works but mcp doesn't

Set SERVICE_ACCOUNT_PATH explicitly to skip ambiguity

확인: echo $SERVICE_ACCOUNT_PATH

대안

mcp-google-sheets 다른 것과 비교

대안	언제 쓰나	단점/장점
Zapier MCP	You already pay for Zapier and want non-code routing	Per-task pricing; less direct control
Notion MCP (databases)	You prefer Notion's database UX and don't need formulas	No formula engine

mcp-google-sheets