Frequently Asked Questions about MCP Servers on UBOS

Q: What is an MCP Server?

A: MCP (Model Context Protocol) Server acts as a bridge, allowing AI models to access and interact with external data sources and tools. It standardizes communication between AI models and external resources.

Q: Why do I need an MCP Server?

A: You need an MCP Server to enable AI models (like LLMs) to interact with the real world. It allows them to access data, use tools (like Gmail or Calendar), and perform actions, extending their capabilities beyond pure language processing.

Q: What are some common use cases for MCP Servers?

A: Common use cases include intelligent email management, automated calendar scheduling, enhanced customer support (chatbots accessing CRM data), and streamlined data analysis.

Q: How does the MCP Server on UBOS integrate with Gmail and Google Calendar?

A: The MCP Server on UBOS provides pre-built integrations with Gmail and Google Calendar APIs, allowing AI agents to read emails, send emails, create calendar events, and manage schedules.

Q: Is the MCP Server secure?

A: Yes, the MCP Server uses OAuth2 authentication to ensure secure access to user data. It also leverages Cloudflare Workers’ security features.

Q: What is Cloudflare Workers and why is the MCP Server built on it?

A: Cloudflare Workers is a serverless computing platform. It provides scalability, reliability, and cost-effectiveness, making it ideal for hosting the MCP Server. The serverless architecture ensures that the server can handle fluctuating workloads without requiring manual scaling.

Q: Can I customize the MCP Server?

A: Yes, the MCP Server is open-source and customizable. You can adapt it to your specific needs by modifying the code or adding new integrations.

Q: What are Durable Objects and how are they used in the MCP Server?

A: Durable Objects are a feature of Cloudflare Workers that allows you to persist data across multiple requests. The MCP Server uses Durable Objects to store OAuth tokens and other state information.

Q: How do I deploy the MCP Server?

A: You can deploy the MCP Server to Cloudflare Workers using the npm run deploy command. You’ll need to configure OAuth credentials and a Cloudflare KV namespace first.

Q: What are the API endpoints exposed by the MCP Server?

A: The MCP Server exposes the following API endpoints: /tools (returns a list of available tools), /api/mcp (the main MCP API endpoint), /oauth/gmail (initiates Gmail OAuth authentication), and /oauth/gmail/callback (Gmail OAuth callback).

Q: How does the MCP Server relate to the UBOS platform?

A: The MCP Server is a valuable asset available on the UBOS platform that complements UBOS capabilities by enabling seamless integration with other services. The UBOS platform also helps user orchestrate AI Agents, connect them with enterprise data, build custom AI Agents with their LLM model and Multi-Agent Systems.

Q: What if I encounter issues during deployment?

A: Consult the “Troubleshooting” section in the documentation for common issues like KV namespace errors and OAuth errors. Also, ensure your TypeScript configuration allows for JavaScript file generation.