ai-scheduler-mcp
@torohash
google tasks apiとgoogle calendar apiを利用したスケジューリング用のsse対応のmcp server
Overview
What is ai-scheduler-mcp?
ai-scheduler-mcp is an MCP (Model Context Protocol) server that integrates Google Tasks and Calendar APIs. It allows MCP clients (such as Roo Code) to access and manage Google Tasks and Calendar functionality via Docker and SSE transport.
How to use ai-scheduler-mcp?
Set up a Docker network named mcp-network, obtain OAuth 2.0 credentials from Google Cloud Console, download credentials.json, build the Docker image using the provided script, run initial authentication to generate token.json, then start the server with ai_scheduler_mcp start. Connect your MCP client to the SSE endpoint at http://localhost:<PORT>/sse.
Key features of ai-scheduler-mcp
- Integrates Google Tasks and Calendar APIs
- Runs in a Docker container
- Uses SSE (Server-Sent Events) transport
- Configurable via environment variables and command options
- Provides a management script for build, start, stop, logs, and delete
Use cases of ai-scheduler-mcp
- Managing tasks and calendar events through AI assistants
- Automating scheduling workflows with MCP-compatible clients
- Integrating Google Tasks and Calendar into custom automation pipelines
FAQ from ai-scheduler-mcp
What Google APIs are required?
Google Tasks API and Google Calendar API must be enabled in your Google Cloud Console project.
How do I set up authentication?
Create an OAuth 2.0 client ID (desktop or web app type), download credentials.json to the project root, run ai_scheduler_mcp start-auth, authorize via the displayed URL, and paste the returned code.
What transport protocol does the server use?
The server uses SSE (Server-Sent Events). The connection URL is http://<host>:<port>/sse.
Can I change the default port?
Yes. Use the -P <port> option when starting the server (e.g., ai_scheduler_mcp start -P 8080) or set the PORT environment variable.
Where is the authentication token stored?
After the initial authentication, a token.json file is generated in the project root. This file is sensitive and must not be committed to Git.
