google_mcp
Minimal, secure Google Workspace MCP server — 43 tools across 10 services (Gmail, Calendar, Drive, Docs, Sheets, Slides, Forms, Tasks, Chat, Meet). Only official Google libraries, zero third-party wrappers, fully auditable. Works with Claude Code, Kiro, Cursor, Windsurf, Codex.
Install / Use
claude mcp add rishapgandhi -- npx -y github:rishapgandhi/google_mcpIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
Development & EngineeringSupported Platforms
Skill content
View source on GitHubGoogle Workspace MCP Server
Minimal, secure, self-hosted MCP server for Google Workspace. Zero third-party wrappers — only official Google libraries. ~500 lines of fully auditable code.
Works with Claude Code, Kiro, Cursor, Windsurf, Codex, GitHub Copilot, and any MCP-compatible AI coding agent.
Why This Exists
Most Google Workspace MCP servers use third-party wrappers with thousands of lines of code and dozens of dependencies you can't audit. This server:
- ~300 lines total — you can read every line in 10 minutes
- Only official Google libraries —
google-api-python-client,google-auth-oauthlib - Zero external network calls — data flows only to
googleapis.com - Credentials stored locally with
chmod 600permissions - No telemetry, no analytics, no SaaS dependency
Services & Tools (43 tools)
| Service | Tools | Description |
|---------|-------|-------------|
| Gmail | gmail_search, gmail_read, gmail_send, gmail_reply, gmail_list_labels, gmail_modify_labels, gmail_delete | Full email management |
| Calendar | calendar_list_events, calendar_create_event, calendar_update_event, calendar_delete_event, calendar_get_event | Events with invite emails |
| Drive | drive_search, drive_read_file, drive_create_file, drive_delete_file, drive_list_folder, drive_share_file | Files, folders, sharing |
| Docs | docs_create, docs_read, docs_append | Create and edit documents |
| Sheets | sheets_read, sheets_write, sheets_create | Read, write spreadsheets |
| Slides | slides_create, slides_read | Create and read presentations |
| Forms | forms_create, forms_get, forms_add_question, forms_get_responses | Create forms, add questions, read responses |
| Tasks | tasks_list_tasklists, tasks_list, tasks_create, tasks_update, tasks_delete | Manage task lists and tasks |
| Chat | chat_list_spaces, chat_send_message, chat_list_messages, chat_create_space | Spaces, messages |
| Meet | meet_create_space, meet_get_space, meet_end_conference, meet_list_conferences | Meeting links, conferences |
Quick Start
1. Google Cloud Console Setup (Free)
- Go to console.cloud.google.com
- Create a new project
- Go to APIs & Services → Credentials → Create OAuth Client ID
- Application type: Desktop Application
- Note your
Client IDandClient Secret
- Go to APIs & Services → Library and enable:
- Gmail API
- Google Calendar API
- Google Drive API
- Google Docs API
- Google Sheets API
- Google Slides API
- Google Forms API
- Google Tasks API
- Google Chat API
- Google Meet REST API
2. Install
git clone https://github.com/rishapgandhi/google_mcp.git
cd google_mcp
cp .env.example .env
# Edit .env with your Client ID and Secret
pip install -e .
3. First-time Authentication
python3 -m src.server
# Opens browser for Google OAuth consent
# Token saved to .credentials/token.json (chmod 600)
4. Connect to Your AI Agent
Kiro CLI
{
"mcpServers": {
"google_workspace": {
"command": "python3",
"args": ["-m", "src.server"],
"cwd": "/path/to/google_mcp",
"env": {
"GOOGLE_OAUTH_CLIENT_ID": "your-client-id",
"GOOGLE_OAUTH_CLIENT_SECRET": "your-secret"
}
}
}
}
Claude Code
claude mcp add google_workspace -- python3 -m src.server --cwd /path/to/google_mcp
Cursor / Windsurf
Add to .cursor/mcp.json or .windsurf/mcp.json:
{
"mcpServers": {
"google_workspace": {
"command": "python3",
"args": ["-m", "src.server"],
"cwd": "/path/to/google_mcp",
"env": {
"GOOGLE_OAUTH_CLIENT_ID": "your-client-id",
"GOOGLE_OAUTH_CLIENT_SECRET": "your-secret"
}
}
}
}
Security
| Aspect | Implementation |
|--------|---------------|
| Dependencies | Only 4 official packages, all pinned |
| Network | Only connects to googleapis.com |
| Credentials | OAuth tokens stored with chmod 600 |
| Code | ~300 lines, fully auditable |
| Telemetry | None — zero external reporting |
| Data flow | Your machine → Google APIs (nothing else) |
Dependencies
| Package | Publisher | Purpose |
|---------|-----------|---------|
| mcp | Anthropic | MCP protocol SDK |
| google-api-python-client | Google | Official API client |
| google-auth-oauthlib | Google | OAuth2 authentication |
| python-dotenv | — | .env file loading |
Project Structure
google_mcp/
├── pyproject.toml
├── .env.example
├── .gitignore
├── LICENSE
├── README.md
└── src/
├── server.py # MCP server (stdio transport)
├── auth/
│ └── oauth.py # OAuth2 — only talks to accounts.google.com
└── tools/
├── gmail.py # Search, read, send
├── calendar.py # List events, create event
├── drive.py # Search, read, create files
├── docs.py # Create, read, append
├── sheets.py # Read, write, create
├── slides.py # Create, read
├── forms.py # Create forms, add questions, get responses
├── tasks.py # Task lists, create/update/delete tasks
├── chat.py # Spaces, send/list messages
└── meet.py # Create meetings, list conferences
Contributing
PRs welcome. Keep it minimal — the goal is a small, auditable codebase.
License
MIT — use commercially, fork, modify, redistribute freely.
Related Skills
Agent-Reach
76.5kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
ruflo
69.7k🌊 The original agent meta-harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
headroom
68.0kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
CowAgent
46.7kOpen-source super AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-model, multi-channel. Lightweight, extensible, one-line install. (formerly chatgpt-on-wechat)
