mcp-gdocs
MCP server for Google Docs, Google Drive, and Comments with flexible authentication, enabling document management, editing, formatting, commenting, and drive operations.
README
mcp-gdocs
MCP server for Google Docs, Google Drive, and Comments with flexible authentication.
Connect Cursor, Claude Desktop, or any MCP client to your Google Docs and Drive.
Start with a single command: npx -y mcp-gdocs.
Quick Start
1. Google Cloud Setup
- Go to Google Cloud Console
- Create or select a project
- Enable Google Docs API and Google Drive API
- Choose authentication method (see Authentication below)
2. Add to Cursor / MCP Client
Service Account — share documents with the SA email, then:
{
"mcpServers": {
"mcp-gdocs": {
"command": "npx",
"args": ["-y", "mcp-gdocs"],
"env": {
"GOOGLE_APPLICATION_CREDENTIALS": "/path/to/service-account.json"
}
}
}
}
OAuth — run npx -y mcp-gdocs auth first, then:
{
"mcpServers": {
"mcp-gdocs": {
"command": "npx",
"args": ["-y", "mcp-gdocs"],
"env": {
"GOOGLE_CLIENT_ID": "your-client-id",
"GOOGLE_CLIENT_SECRET": "your-client-secret"
}
}
}
}
The server starts automatically when the MCP client connects.
What Can It Do?
Most write operations accept an items array for bulk execution in a single call. All tools support an optional tabId parameter for multi-tab documents.
Google Docs
<details> <summary>Reading, writing, formatting, tables, images, native Markdown, batch, tabs, named ranges</summary>
| Tool | Description |
|---|---|
docs_read_document |
Read as text, JSON, markdown, or summary; optional fields mask |
docs_get_document_info |
Document metadata (title, ID, revision) |
docs_list_document_tabs |
List all tabs in a multi-tab document |
docs_insert_text |
Insert text at one or multiple positions |
docs_append_text |
Append text to the end |
docs_delete_range |
Remove content by index ranges |
docs_replace_all_text |
Replace all occurrences of patterns |
docs_replace_document_content |
Replace entire document content |
docs_insert_page_break |
Insert page breaks |
docs_rename_tab |
Rename a document tab |
docs_update_header |
Update header content |
docs_update_footer |
Update footer content |
docs_create_footnote |
Create footnotes at a position |
docs_add_tab |
Add a new document tab |
docs_apply_text_style |
Bold, italic, underline, colors, font, links |
docs_apply_paragraph_style |
Alignment, spacing, indentation |
docs_apply_heading_style |
Heading styles (H1–H6) |
docs_format_by_text |
Find text and apply formatting without knowing indices |
docs_insert_table_row |
Add row to table |
docs_insert_table_column |
Add column to table |
docs_delete_table_row |
Delete row from table |
docs_delete_table_column |
Delete column from table |
docs_update_table_cell_content |
Update cell content |
docs_update_table_cell_style |
Cell background color |
docs_insert_table_with_data |
Create a table pre-filled with data |
docs_insert_image |
Insert images from URL |
docs_insert_local_image |
Insert images from local files (upload + insert) |
docs_replace_with_markdown_native |
Replace from Markdown (native Google conversion) |
docs_replace_with_markdown_file_native |
Replace from Markdown file (native conversion) |
docs_batch_update |
Batch multiple operations in one API call (30 request types) |
docs_create_named_range |
Create a named range |
docs_delete_named_range |
Delete a named range |
docs_replace_named_range_content |
Replace text inside a named range |
</details>
Comments
<details> <summary>List, get, add, reply, resolve, delete</summary>
| Tool | Description |
|---|---|
docs_list_comments |
List all comments with author and date |
docs_get_comment |
Get a specific comment with replies |
docs_add_comment |
Create comments anchored to text |
docs_reply_to_comment |
Reply to comments |
docs_resolve_comment |
Mark comments as resolved |
docs_delete_comment |
Remove comments |
</details>
Google Drive
<details> <summary>Files, folders, permissions, export, revisions, shared drives</summary>
| Tool | Description |
|---|---|
drive_list_documents |
List documents, optionally filtered |
drive_search_documents |
Full-text search across documents |
drive_create_document |
Create new documents |
drive_create_from_template |
Create documents from a template |
drive_create_from_markdown |
Create a new Doc from inline Markdown |
drive_create_from_markdown_file |
Create a new Doc from a local .md file |
drive_create_folder |
Create folders |
drive_list_folder_contents |
List folder contents |
drive_get_folder_info |
Get folder metadata |
drive_move_file |
Move files to another folder |
drive_copy_file |
Duplicate files |
drive_rename_file |
Rename files |
drive_delete_file |
Move files to trash |
docs_export_pdf |
Export document as PDF (optional savePath) |
drive_export_file |
Export to markdown, PDF, DOCX, and other MIME types |
drive_share_file |
Share a file (user, group, domain, anyone) |
drive_list_permissions |
List ACLs on a file |
drive_update_permission |
Change role on an existing permission |
drive_remove_permission |
Revoke a permission |
drive_list_revisions |
List revision history |
drive_get_revision |
Fetch metadata or content of one revision |
drive_list_shared_drives |
List shared drives (Team Drives) |
</details>
Prompts
<details> <summary>Workflow-oriented prompt templates for common tasks</summary>
| Prompt | Description |
|---|---|
pretty-format |
Create or reformat a Doc from Markdown; cleanup batch for rules, spacing, tables |
gost-format |
Document structure and styles aligned with GOST 19.106-78 |
template-fill |
Fill a template using placeholders / named ranges |
export-to-markdown |
Export a Google Doc to Markdown via Drive API |
format-table |
Pin headers, column widths, and cell styles for a Doc table |
share-document |
List permissions, then share with a user |
</details>
Resources
| URI | Description |
|---|---|
gdocs://documents/recent |
JSON list of recently modified Google Docs |
Batch Operations
docs_batch_update combines multiple heterogeneous operations into a single API request (30 request types). Large arrays are automatically split into chunks. Rate limiting uses a shared API semaphore.
Authentication
The server supports multiple authentication methods. If several variables are set, the first found is used: OAuth → SERVICE_ACCOUNT_PATH → CREDENTIALS_CONFIG → GOOGLE_APPLICATION_CREDENTIALS.
| Criteria | Service Account | OAuth |
|---|---|---|
| Document access | Only shared with SA | All your documents |
| Drive operations | Only in shared folders | Full access to your Drive |
| Best for | CI/CD, servers, automation | Personal use, Cursor, Claude Desktop |
| Setup | Download JSON key, share documents | OAuth flow in browser |
| Google Workspace | Impersonation — full access as user | Not needed |
Enterprise (Google Workspace): with domain-wide delegation, SA can act on behalf of any domain user via
GOOGLE_IMPERSONATE_USER.
Recommendation: use OAuth for personal work in Cursor/Claude Desktop. Use Service Account with impersonation for enterprise automation.
Method A: Service Account
Headless, secure, ideal for server environments.
Steps:
- GCP Console → IAM & Admin → Service Accounts → Create
- Download JSON key
- Share documents/folders with SA email (Editor)
Three ways to provide credentials:
<details> <summary>A1. <code>SERVICE_ACCOUNT_PATH</code> (+ impersonation)</summary>
{
"mcpServers": {
"mcp-gdocs": {
"command": "npx",
"args": ["-y", "mcp-gdocs"],
"env": {
"SERVICE_ACCOUNT_PATH": "/path/to/service-account-key.json",
"GOOGLE_IMPERSONATE_USER": "user@yourdomain.com"
}
}
}
}
</details>
<details> <summary>A2. <code>CREDENTIALS_CONFIG</code> — Base64-encoded JSON (Docker / CI / K8s)</summary>
base64 -w 0 service-account.json # Linux
base64 -i service-account.json | tr -d '\n' # macOS
{
"mcpServers": {
"mcp-gdocs": {
"command": "npx",
"args": ["-y", "mcp-gdocs"],
"env": {
"CREDENTIALS_CONFIG": "ewogICJ0eXBlIjogInNlcnZpY2VfYWNjb3VudCIs..."
}
}
}
}
</details>
<details> <summary>A3. <code>GOOGLE_APPLICATION_CREDENTIALS</code></summary>
{
"mcpServers": {
"mcp-gdocs": {
"command": "npx",
"args": ["-y", "mcp-gdocs"],
"env": {
"GOOGLE_APPLICATION_CREDENTIALS": "/path/to/service-account.json"
}
}
}
}
</details>
Method B: OAuth 2.0
For personal use with interactive browser login.
<details> <summary>B1. Setup and authorize</summary>
- Google Cloud Console → APIs & Services → Credentials
- Create OAuth client ID → Desktop app
- Copy Client ID and Client Secret
- OAuth consent screen → add your email as Test User
- Authorize:
GOOGLE_CLIENT_ID="..." GOOGLE_CLIENT_SECRET="..." npx -y mcp-gdocs auth
{
"mcpServers": {
"mcp-gdocs": {
"command": "npx",
"args": ["-y", "mcp-gdocs"],
"env": {
"GOOGLE_CLIENT_ID": "your-client-id",
"GOOGLE_CLIENT_SECRET": "your-client-secret"
}
}
}
}
OAuth refresh tokens are stored in ~/.config/mcp-gdocs/token.json (respects XDG_CONFIG_HOME). To re-authorize, delete the token file and run auth again.
</details>
Multiple Google Accounts
GOOGLE_MCP_PROFILE isolates token storage per profile:
~/.config/mcp-gdocs/
├── token.json # default (no profile)
├── work/token.json # GOOGLE_MCP_PROFILE=work
└── personal/token.json # GOOGLE_MCP_PROFILE=personal
Environment Variables
<details> <summary>All supported environment variables</summary>
| Variable | Method | Description |
|---|---|---|
SERVICE_ACCOUNT_PATH |
SA | Path to SA JSON key (+ impersonation support) |
GOOGLE_IMPERSONATE_USER |
SA | Email for impersonation (optional) |
CREDENTIALS_CONFIG |
SA | Base64-encoded SA JSON (Docker/CI) |
GOOGLE_APPLICATION_CREDENTIALS |
SA | Path to SA JSON key (standard Google variable) |
GOOGLE_CLIENT_ID |
OAuth | OAuth client ID |
GOOGLE_CLIENT_SECRET |
OAuth | OAuth client secret |
GOOGLE_MCP_PROFILE |
OAuth | Profile name for isolated token storage |
</details>
Known Limitations
- SA without Workspace: Service account without Google Workspace license has Drive quota = 0 and cannot create files. Use OAuth or impersonation.
- Comment anchoring: Programmatically created comments may not be anchored to text in Google Docs UI (Drive API limitation).
- Deeply nested lists: Lists with 3+ nesting levels may have formatting artifacts when converting Markdown.
- Paragraph style safe range: paragraph and heading style tools automatically adjust ranges to prevent styles from bleeding into adjacent paragraphs.
Troubleshooting
- Server won't start: check that environment variables are set in the MCP config
envblock. - Authorization errors: make sure Docs and Drive APIs are enabled in Google Cloud Console. For OAuth — check that your email is added as Test User.
- Re-authorization: delete
~/.config/mcp-gdocs/token.jsonand runnpx -y mcp-gdocs authagain.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.