# Google Tasks MCP Server
A Model Context Protocol (MCP) server for Google Tasks integration in Claude Desktop. This server enables AI assistants to manage Google Tasks through natural language interactions.
Forked from [Zach Caceres' (zcaceres) original implementation](https://github.com/zcaceres/gtasks-mcp) to enhance authentication handling, add support for multiple task lists, and provide a more robust npm package with Smithery integration for Claude Desktop. This fork is fully functional and properly deployed on Smithery, making it ready for immediate use with Claude.
[](https://smithery.ai/server/@alvinjchoi/gtasks-mcp)
[](https://www.npmjs.com/package/@alvincrave/gtasks-mcp)
[](https://opensource.org/licenses/MIT)
## Features
- List all tasks in your Google Tasks lists
- Search for specific tasks by query
- Create new tasks with title, notes, and due date
- Update existing tasks (title, notes, status, due date)
- Delete tasks
- Clear completed tasks from a task list
- Full integration with Google Tasks API
- Secure OAuth2 authentication
- Docker support for containerized deployment
## Task Creation Example
Here's a simple example of task creation using natural language with Claude:
With just a simple instruction like "create a task named 'eat a banana in the morning'", Claude can instantly create the task in your Google Tasks.
## Installation
### Installing via Smithery
To install Google Tasks Integration for Claude Desktop automatically:
```bash
npx -y @smithery/cli install @alvincrave/gtasks-mcp --client claude
```
### Manual Installation
```bash
npm install @alvincrave/gtasks-mcp
```
## Setup
1. Create a Google Cloud Project and obtain credentials:
a. Create a Google Cloud Project:
- Go to [Google Cloud Console](https://console.cloud.google.com/)
- Create a new project or select an existing one
- Enable the Google Tasks API for your project
b. Create OAuth 2.0 Credentials:
- Go to "APIs & Services" > "Credentials"
- Click "Create Credentials" > "OAuth client ID"
- Choose "Desktop app" as application type
- Give it a name and click "Create"
- You will get your `GOOGLE_CLIENT_ID` and `GOOGLE_CLIENT_SECRET`
c. Get Refresh Token:
- Go to [OAuth 2.0 Playground](https://developers.google.com/oauthplayground/)
- Click the gear icon (Settings) in the top right
- Check "Use your own OAuth credentials"
- Enter your OAuth Client ID and Client Secret
- In the left panel, find "Tasks API v1" and select "https://www.googleapis.com/auth/tasks"
- Click "Authorize APIs" and complete the OAuth flow
- Click "Exchange authorization code for tokens"
- Copy the "Refresh token" - this is your `GOOGLE_REFRESH_TOKEN`
2. Configure in Claude Desktop:
```json
{
"mcpServers": {
"gtasks": {
"command": "npx",
"args": ["-y", "@alvincrave/gtasks-mcp"],
"env": {
"GOOGLE_CLIENT_ID": "your_client_id_here",
"GOOGLE_CLIENT_SECRET": "your_client_secret_here",
"GOOGLE_REFRESH_TOKEN": "your_refresh_token_here"
},
"connectionTypes": ["stdio"],
"autoRestart": true
}
}
}
```
## Docker Support
You can run this server in a Docker container:
```bash
# Build the Docker image
docker build -t gtasks-mcp .
# Run the container
docker run -p 3000:3000 \
-e GOOGLE_CLIENT_ID=your_client_id \
-e GOOGLE_CLIENT_SECRET=your_client_secret \
-e GOOGLE_REFRESH_TOKEN=your_refresh_token \
gtasks-mcp
```
## Usage Examples
The server provides several tools that can be used through Claude Desktop:
### Search Tasks
```
search for tasks containing "meeting"
```
### List Tasks
```
show me all my tasks
```
### Create Task
```
create a task to "Prepare presentation" due on Friday
```
### Update Task
```
mark the "Send email" task as completed
```
### Delete Task
```
delete the task about "old project"
```
### Clear Completed Tasks
```
clear all completed tasks from my list
```
## Security Notes
- Keep your Google API credentials secure
- Regularly rotate your refresh tokens
- Store sensitive information in Claude Desktop configuration
- Never share or commit your credentials to version control
- The refresh token gives access to your Google Tasks, treat it like a password
## License
This project is licensed under the MIT License.
